I have recently placed a post on this topic on the blog of the sculptor Dr. Gindi. The post is here:
Posts by others on the topic of the infinite are here:
I have recently placed a post on this topic on the blog of the sculptor Dr. Gindi. The post is here:
Posts by others on the topic of the infinite are here:
"If no one comes from the future to stop you from doing it, then how bad can it be?"
This slogan has been doing the rounds on social media. It is in a photograph of the title page to a spiral bound document. The document is attributed to US Robots and Mechanical Men Inc., a fictitious company in Isaac Asimov's Robot stories.
In this post we shall assume that people from the future could come back to our time and make adjustments on the basis of what they could see would be the consequences of actions taken now. They might do so in order to make things better for some people (perhaps us or themselves) at some time later than now. And they might do so indirectly, by leaving information that was relevant to predictions somewhere where we would find it. Even if they intervened directly, for example by changing people's brains just before they made certain decisions or by introducing fatigue into machines that people were using so that the machines did not carry out the actions that the people had chosen, the affected people would not be aware of the fact of the interventions. They might in retrospect be puzzled that they had taken decisions which were out of character, or that previously reliable machines had failed at critical moments, but that would be all.
What consequences might there be for some of our decision-making, and what would be the scope of and limits on the actions of the time-travellers?
In order to keep track, we shall give numbers to years.
Year 1 is now.
Year 600 is the year in which the time-travellers who visit us live. We shall assume that after visiting us they would go back there. It would be where all their friends lived and where the lifestyle was one to which they were accustomed.
Year 900 is a time when everyone alive in year 600 and in the next few generations after them will have died.
We shall speak of people living in, for example, year 600, to mean people who would be confined to that year and decades either side of it if there were no time travel. That is, the people who live in year 600 are the people for whom that year and decades around it are their home period of time.
One of the problems with selecting actions in year 1 on the basis of their consequences is that it is impractical to work out more than a few of the consequences, or to work out the long-term consequences. It is possible that an apparently harmless action would in fact have disastrous consequences years into the future, although it is also arguable that little blame should attach to a specific action in year 1 because many other actions not yet taken would also influence what happened years into the future. Consequentialists would love to be able to foresee all significant consequences of actions, but they cannot do so.
Once one reached, say, year 50, it would at least theoretically be possible to see what the long-term consequences up to that point of an action in year 1 had been. It would not be easy, because many other things would have happened in the intervening years. And to do the job in a way that would be of practical help to consequentialists in year 1 who had magically gained access to information that was in fact only available in year 50, one would have to work out how the world would have turned out following all of the alternative actions or inaction that might plausibly have produced a better outcome. That might remain impossible, or it might require the running of many simulations on the lines envisaged by the simulation argument that Nick Bostrom puts forward.
This is the message of the slogan. If actions in year 1 would lead to disastrous consequences, people living in year 600 would come back to year 1 and either prompt us to different actions or make other changes so as to break the causal chains that would have led to disaster. Then consequentialists could do the best they could in year 1, and be confident that their truly massive mistakes would be nullified. But would people living in year 600 bother? Could they coherently intervene? And what might people living in year 900 do?
We might expect people living in year 600 to intervene if it would make life better for themselves. And it might be that year 1 would be the easiest point at which to intervene, before the consequences of actions in year 1 had spread too widely and deeply. On the other hand, it might be hard to foresee the effects in year 600 of adjustments to year 1. Perhaps the best option would be a broad-brush adjustment to year 1 followed by finer adjustments to years 100, 200, 300, 400, 500, 550 and 575, making adjustments to each year as it would be modified by adjustments to earlier years. (Later we shall note a reservation about making adjustments to years close to year 600.)
Would people living in year 600 care about people earlier than themselves? They might foresee bad consequences of year-1 actions for people living in year 2, or year 50, or year 300. Those people would include the ancestors of people living in year 600. There is a large philosophical debate about what we owe to our near and distant descendants. Here we ask whether, given the opportunity, people should benefit their ancestors.
There is a natural feeling that people probably would do so, assuming no significant cost to themselves. We care about the living elderly. We may not give much thought to the states of mind of the dead, given that there are no such states of mind and there is no current way to affect the states of mind of the dead when they were alive. But if we could do something that would have benefited them back in their lifetimes, doing so would be a natural extension of our current habit of caring for the living.
Having said that, if there were a significant cost to people living in year 600 of benefiting their ancestors, one can envisage them saying that if they did not act the ancestors would never be aware of the lost opportunity to be helped. But then, if intervention in the past became routine, people living further forward (say in year 800) might complain that they seemed to have get things wrong, say in year 799, making year 800 a bad year, and that they had not been helped by people living in year 1200. There might come to be an expectation that future generations, able to estimate the consequences of actions, should help. That expectation might be seen by those future generations as creating an obligation on them to help.
A common issue in discussions of backward time travel is that of coherence. If someone living in year 600 comes back and changes something in year 1, it seems likely that things in year 600 would have to change too because the intervening course of history would have been different. Moreover, all the history between years 1 and 600 would need modification.
There are extreme concerns over coherence, for example that someone coming back from year 600 might not have existed under the re-written history, which would at the least seem to make it difficult for them to return to year 600. Then there are apparently more modest concerns that things would have had to be a bit different in various years.
Concerns of both sorts would however be of comparable severity in one sense. It is a fundamental principle that the world has, from any particular point of view, one history. At least, this is so subject to the isolation of some regions of spacetime from others on account of the finite speed of light and the expansion of space.
Call the history up to year 600 absent any intervention History-F (for first). An alternative, History-S (for second), would have to displace History-F from the point of intervention onward. Then what would have happened to History-F?
One answer involves the idea of multiple histories. History-F continues to exist and play out, but at the point at which a visitor from year 600 to year 1 intervenes, History-S starts to run in parallel. Anybody after that point of intervention is either in History-F and can only look back over that history, or in History-S and can only look back over that history. The histories over which they can look back are the same at all times before the point of intervention.
Whatever merits this approach may have in the interpretation of quantum mechanics, it does seem a bit much at the level of human life. It could be correct, but it would raise further questions about which was the real history, where there would be space for all the alternatives, and so on. The counter-intuitive nature of an approach does not show that it is mistaken. We must put up with a great deal that is counter-intuitive in physics. But at the human level, it at least intuitively (and therefore perhaps circularly) seems that intuition should carry more weight, if not as a guide to what is correct then at least as an indicator of the manifest incorrectness of some ideas.
Such concerns would arise even if a split into History-F and History-S was followed by a merger well before year 600, so that the picture was like that of a main railway track and a track that branched off to one side, ran in parallel to the main track, and then rejoined it. There would be some points at which there would be two different histories running in parallel. Indeed, concerns over coherence would be more acute. Without a subsequent merger one might at least say that there was no instant of time at which two histories were running, because each history would have its own time. But with a merger there would be much more pressure to accommodate a common timescale with two histories running at the same time, simply because one could count backward in time from a post-merger point. Moreover, there would be a question of which history should be considered to have preceded the merger. It would be a condition of merger that only one set of records and memories existed. So only one history would be written, and there would be no awareness of the branching and the merger. Nonetheless one could consider the unverifiable possibility of those events and the period of parallel histories having happened. This would challenge the intuitive idea that however obscure the causal pathways, the present was caused by a single past.
(We may note effects on motivation if there were such a merger, so that year 600 and years around it would be the same with or without the intervention. On the one hand the merger would make people living in year 600 more willing to intervene in the past, because it would make no difference to their lives. On the other hand they might say that there would be no benefit to them from intervening, so they might not bother. If the point of merger were before year 400, there would not even be any benefit to anyone whom anyone alive in year 600 had ever met.)
Now let us put all such concerns, with or without a merger, to one side. Just suppose that multiple histories would be created and that someone comes back from year 600 in History-F. They intervene and make a change to the benefit of people in some year later than year 1 (it need not be people living in year 600). That would make some people in History-S better off than they would have been had their history been the same as History-F. But it would leave unchanged the lives of everyone in History-F, including all of the intervener's friends back in the year 600 as it was when they set out on their journey (and as it would be on their return if there was a merger).
We could then ask what the benefit of the intervention would be. If it was to save people from some disaster, there would be a new set of people who did not experience the disaster but the old people, in History-F, would still experience it. So there would be extra people who would have a better life. If it would overall be a good life, we could see the point of the intervention. But then, would people living in year 600 bother to create better histories for people who would otherwise not have existed at all? There are connections both with the question we posed above as to whether one should care about people who no longer exist, and with the view that every positive life is worth creating which leads to Parfit's repugnant conclusion. (Intervention in the case we consider might however not be as repugnant as the breeding of as many people with even marginally positive lives as possible, because in History-S, and even in History-F, lives might be far better than marginally positive.)
One alternative which would remove many of the concerns we have raised would be to have the intervention wipe out History-F, so that it no longer occurred. This would however not be a panacea. It would go against our intuition that once a course of events has unrolled in time, the events cannot be undone. And interveners would have to leave a year 600 from History-F before returning to a year 600 from History-S. A question of the identity of the leavers with the returners would arise, although one might borrow from work on possible worlds and in particular David Lewis's counterpart theory. Alternatively one might claim that History-S was the only history there ever was, it being fixed from the start that the interveners would do their work. One might compare that idea with the conjecture of superdeterminism in quantum mechanics.
Another answer is offered by the possibility of changes made in year 1 by visitors from year 600 making no difference that would bother humanity before year 601. It is not that no change would be made anywhere. Information and its transmission have energy costs. There is no free information, and no free causation. But if the difference made, which would technically create a split in the universe, were parked somewhere out of sight, for example on a planet that orbited a distant star, the idea of a split history would not be so disturbing. Nobody would have to contemplate the existence of alternative human beings who had split off from them and who would have noticeably different lives, even if there were such splits in theory. The effects of the distant change would then work their way back to Earth at the appropriate time.
The challenge of possible incoherence would not go away, but at least it would not be so directly disturbing to our idea of how human lives play out in time. The effects of the change made by interveners who came back from year 600 would not enter into the human realm until year 601 or later, so there would be no split in human history.
The motive of interveners who came back from year 600 would also be clear. They would be making life better for their own society, including themselves once they returned to year 600. Indeed, the approach suggested here would not allow the interveners to arrange any benefit to people between years 1 and 600, on pain of reintroducing incoherence that would be visible in the human realm.
On the other hand, if that were all that would be done, it would not be clear that there was much point in intervening in year 1. Why not just use, in year 600, the technology then available to change how things would be in year 601 or later years? It is plausible that if people living in year 600 had the technology for time travel and to effect changes far away from Earth, they would have the technology to effect local changes at reasonably short notice. And that would probably be safer, both because time travel might not work as intended and because the causal chain from some distant planet over a period of 600 years would be highly uncertain.
People living in year 900 would have more information than people living in year 600. They would know how the world turned out in years 601 to 899, and they might attribute some of the things that happened in that timespan to interventions made in years before 600 by people who lived in year 600. They might also have more information relevant to the choice of interventions that would influence years before 600, because they might have developed better methods of simulation of alternative realities.
It would therefore not be surprising if people living in year 900 intervened to over-write the work of people who lived in year 600, in addition to making changes in years from 601 to 899.
The layering of interventions on interventions in the period from year 1 to year 599 would not generate new problems of coherence, although the problems already identified would still arise, and in the slightly stronger form that there would be a need to handle three or more alternative histories instead of just two.
There would be a new concern for people who lived in year 600. They could have no confidence that their interventions would be final. But there would also be new reassurance for them. If their interventions were not in fact ideal, and if people who lived in year 900 cared enough, those people would intervene to make corrections.
The process could of course go on, with subsequent generations making further interventions. There would however be two possible restrictions, both of which might also apply to people who lived in year 600 who made the first set of interventions.
The first possible restriction is this. People at any time T might be reluctant to make changes directly within the lifetimes of people who were still alive at time T or who had been personally known to people still alive at time T. The thought here is that there might be a reluctance to disrupt the lives of people one knew or had known, in whatever sense of disruption was appropriate given how the problems of coherence and multiple histories were handled. And direct disruption at points within the lifetimes of those people might be considered to be significantly more offensive than disruptions at earlier points in time which would likewise change the lives of the people in question. Any such gradation of offence would probably be based on a feeling that people at the point of intervention would feel a sudden jolt. That would be likely to be a mistaken feeling, but unless the effects of intervention in the past became much better understood by year 600 than they are now, it would be hard to displace the feeling.
The second possible restriction is this. People might eventually lose interest in the welfare of people in the distant past, and would also think it safer to adjust their own welfare by making interventions in the more recent past so as to keep the causal chains shorter and their simulations less prone to serious error. So people living in year 600 might want to intervene as far back as year 1, but people living in year 900 might not go back further than year 300, people living in year 1200 might not go back further than year 600, and so on. This would limit the number of layers of intervention in any particular year that would build up.
In the science fiction stories of Star Trek, there is the Prime Directive. There does not seem to be a canonical text, but the gist is that Starfleet personnel shall not interfere in the normal development of any alien society, for example by introducing technology the society has not yet discovered for itself. Even intervention in order to save a starship or its crew is not allowed.
It is possible that people living in year 600, while they had overcome the technical challenges of time travel, would decide that the philosophical challenges were too great. Then they might adopt their version of the Prime Directive and leave us alone, even though we could hardly be called an alien society.
This post is about the relationship between our discourse about the world and the world itself.
We have already explored the topic in Accounts and Reality:
There we put forward a way to look at the relationship which would allow us to put to one side the question of scientific realism and parallel questions outside the natural sciences.
The aim here is different. It is to explore ideas from two authors, Immanuel Kant and Ayn Rand. Our main focus will be on Rand.
Our aim is not exegesis. We shall freely borrow material and take ideas out of context to suit our purpose. We shall also develop Rand's ideas beyond what she published. She might or might not have agreed with all of the developments we propose. But her central ideas do provide the framework for our developments, and to that extent we remain true to her thought.
There is indubitably a world that exists and that has its nature independently of our thoughts about it. That nature is stable over time. It has a constancy that we capture by formulating laws of nature. We would never have evolved if there were not such an independent and stable world.
We give sophisticated descriptions of the world, sometimes in the terms of physics and chemistry, sometimes in the terms of other natural sciences, and sometimes in the terms of the social sciences and the humanities. We shall concentrate on descriptions that are expected to be very widely applicable across different places (on and off our planet) and times. Such descriptions are given in the natural sciences.
Our problem is to find a satisfactory way to describe the relationship between the independent world and our descriptions of it. In what sense, if any, do our descriptions set out the actual nature of the world?
The task is immediately made difficult by the fact that when we want to discuss the relationship between two things, we normally start by understanding the natures of the things separately. But here, we cannot say anything about the independent world without describing it. We can draw aside the curtain of a description, for example when we reduce some high-level phenomenon to lower-level phenomena, but then we have another description.
Kant, in his Critique of Pure Reason, tells us that our inability to escape from a sequence of descriptions points directly to the nub of the problem. We can never get beyond our own perceptions and descriptions of the world, which are perceptions and descriptions of the phenomenal world, the world as it appears to us.
This is a serious obstacle because in formulating our descriptions we impose mental structures, such as those of space, time and causation, which we have to impose in order for us to be subjects of coherent experience and to acquire knowledge.
Moreover, we should not conclude that the applicability of these mental structures discloses the nature of things as they might be independently of experience, things which he calls the noumena. In order for us to conclude that, things as they might be independently of experience would have to be presented to us in experience. Then the mental structures would inevitably apply, simply as a condition of experience, and we could not conclude that the structures were required by virtue of the nature of the noumena. All we can do is take it that there is some nature of reality as it is independently of experience, and then say nothing about that nature.
We should however be content with the phenomenal world, the world as it appears to us. It is the world as described by our sciences. It has stability and we can all know about a single phenomenal world. It is not our fleeting and subjective perceptions.
In this section we shall set out Rand's scheme as elaborated by us. (This qualification is important: we have added elements that are not in Rand's writings. And as we are not engaged in exposition of her views, we have mostly done so silently. Our aim is to show that there is a useful system on Randian lines.)
Rand regards as a huge mistake the Kantian generation of mystery as to the nature of the world as it is in itself. Her specific criticisms of Kant are contested. But she also offers an alternative approach.
For Rand we have the intrinsic, which is the world that is independent of our thoughts. We also have our percepts, where a percept is "a group of sensations automatically retained and integrated by the brain of a living organism" (Rand, Introduction to Objectivist Epistemology, chapter 1). Percepts may well differ from one person to another, and their nature will depend on the perceptual apparatus of the organism. Finally, we have the objective. This is the set of facts about the world. Facts are discovered when we apply the appropriate concepts to make sense of our percepts, where those percepts are in turn generated by the intrinsic world (which includes our own perceptual apparatus). The objective is therefore a relation between the intrinsic world and the appropriate concepts.
We need to interpret the idea of a relation in the right way. We should see relations not as processes of interaction but as connections, for example the connection of siblinghood which can exist between two people and which can be seen when we put two siblings side by side. For a given state of the intrinsic world and a given set of concepts, a conceptual scheme, there will be a given set of purported facts that can be seen. If the concepts are appropriate, they will be facts and therefore parts of the objective.
The objective must comprise a single set of facts. All rational creatures, and all other objects, exist in a single world, and things may be said about that world. If the notion of a fact is to make sense at all, each fact has to be a fact for everyone. The alternatives would be on the one hand disjoint sets of facts for different groups of rational creatures, as if they were speaking languages between which no translation was possible, and on the other hand a shared language or translatable languages within which different creatures expressed their personal truths without any requirement to avoid contradicting one another. Neither of these options would make any sense.
The need for a single set of facts gives rise to a problem for Rand's approach. Different people may have different percepts. And if we consider non-human rational creatures, we can expect them to have different perceptual apparatus. So for a given state of the intrinsic world we would need a function from ordered pairs, each of which comprised a set of percepts and a conceptual scheme, that always had the same value. That value would be the single set of facts. How could this be so?
The solution lies in the role of concepts. Different conceptual schemes would be appropriate to rational creatures of different natures, specifically those with different perceptual apparatus and the consequently different sets of percepts. (We restrict ourselves to differences in percepts by virtue of differences in apparatus, and exclude differences by virtue of happening to observe different things.) So rational creatures with different perceptual apparatus could be taken to the same facts. The differences in conceptual schemes would balance out the differences in percepts.
There is however one more stage. When appropriate concepts are used the result will be statements of facts, and the use of different conceptual schemes will result in different statements. For Rand's approach to work, it must be possible to claim that different statements could be seen as stating the same facts. But this claim could be made. We may envisage scope to hold different statements together by applying rules of translation between them. Such identities of facts stated in different ways might however be limited to statements of large and coherent sets of facts. Individual facts might not be identifiable across different conceptual schemes, because schemes would carve up the world in different ways.
In such a way we can see the objective as common property among all rational creatures, and a particular statement of it as common property within a given community of rational creatures, although different members of a community may grasp parts of the objective with varying degrees of sophistication and some parts may not yet be grasped by anyone. Any conflicts as to contents of the objective, that is, as to whether some given purported facts are facts, will indicate that mistakes have been made, so that at least one party to a disagreement has not in fact captured the relevant parts of the objective. A particularly likely reason is that they are not using the appropriate concepts. We have indeed spoken of the (single) conceptual scheme that is appropriate to rational creatures of a given nature, rather than an appropriate scheme. We shall say more about the appropriateness of conceptual schemes below.
In speaking of different concepts and different statements of the same facts, we have added to Rand's thought. But this development is within the spirit of Rand. She is very keen to emphasise that facts are as they are, independently of what we or any other rational creatures might think.
Having said that, our development does require careful handling in order to remain consistent with what Rand says. For her, the objective is a relation between the intrinsic world and the appropriate concepts. If we recognise the role of different conceptual schemes and then consider the relation between the intrinsic world and a given conceptual scheme, it may seem that the relation would have to comprise not the objective but a given statement of the objective. This would be a significant deviation from Rand's thought and not a development of it. Fortunately, we can save the day and allow the relation to be the objective itself.
One cumbersome way to save the day would be to say that the objective was a relation between the intrinsic world and the entire set of appropriate conceptual schemes. But that would raise questions about the definition of the boundary of the set. For example, which types of perceptual apparatus should be considered?
A more straightforward way is the following. While a given appropriate conceptual scheme would only take us to one statement of the objective, it would be a statement of the one objective. And it would be the same objective whichever appropriate scheme was used. Thus the relation is still the objective rather than a statement of it. It might however be necessary to extend the chosen conceptual scheme in order to allow it to cover all the ground covered by all conceptual schemes, so as to avoid having parts of the objective accessible to some rational creatures but not to others.
It is sometimes appropriate to think of the objective as a single entity that comprises all the facts. This can help us to bring out the structure of Rand's approach and contrast it with Kant's. But when we consider questions of human progress and error, it is more helpful to consider specific purported facts and ask whether they are indeed facts, meaning that they are elements in a relation between the intrinsic world and either the whole appropriate conceptual scheme or some parts of it. (Henceforth, "the appropriate conceptual scheme" will be used to mean the scheme appropriate to rational creatures of the relevant nature.) And when we refer to facts without qualification, rather than purported facts, this is what we shall mean.
Concentration on individual facts will have the advantage of taking away any need to think it practical ever to set out the entire appropriate conceptual scheme. What will matter will be concepts that are needed to relate to the intrinsic world in order to yield the facts in question. These will need to be appropriate concepts, in the sense that they would have their places within the appropriate scheme.
Defining the objective as a relation between the intrinsic and the appropriate conceptual scheme does not allow for the distinction between the world as it is in itself and the world as it appears to us on which Kant relied. The objective considered as a single entity which can be identified given a pair that comprises the intrinsic world and the appropriate conceptual scheme is not to be seen as parallel to the world in itself in the way that for Kant, the world as it appears to us is parallel to the world as it is in itself. ("Parallel" is not meant to imply separability. We mean to permit a one-world interpretation of Kant under which one would regard the world as it was in itself and the world as it appeared to us as two aspects of a single world rather than as two worlds.) Rather, while the intrinsic world remains the world to which we relate (and in which we exist), we get away from thinking of an objective world. Instead we should speak only of the objective and statements of it. The intrinsic world is the only world. The objective is how things are, a notion that is more a concern of epistemology than of metaphysics. To echo Wittgenstein (Tractatus 1.1), it is the totality of facts and not of things. But to contrast with Wittgenstein, while this totality of facts is the objective, it is not the world. We know facts that are parts of the objective, and we state those facts in ways that reflect our concepts, but we know the world, the intrinsic world rather than the objective.
We have spoken of the appropriate conceptual scheme for rational creatures of a given nature as being paired with the intrinsic world to yield the objective as a relation between them. We need to give some substance to this notion of appropriateness.
A conceptual scheme should take us from our percepts to the objective which is set out in the statements of facts that we make (and which would equally be set out in different statements given by other rational creatures). The objective should not depend on our perceptual apparatus, but our percepts clearly do depend on that apparatus. It is the conceptual scheme that provides the necessary flexibility. The appropriate scheme for rational creatures of some given nature is the one that is appropriate to their perceptual apparatus, where that apparatus encompasses both their external organs and whatever processing in their brains may be considered involuntary and inevitable. The conceptual scheme must be the one that will take the rational creatures from whatever percepts they have to the objective rather than to some merely apparent objective.
This point extends to routes to the objective that run through scientific instruments. Rational creatures construct them to pick up the values of certain variables. Then the creatures read off the data and draw conclusions. The instruments become parts of the creatures' perceptual apparatus. The creatures still need to find the appropriate conceptual scheme that will lead them to the objective, given how the instruments have been constructed.
The objective includes not only facts that once established may be expressed without reference to percepts, but also facts that are to be expressed in perceptual terms. The sizes, shapes, speeds and colours of objects, the pitches and volumes of sounds, and so on, are all facts that fall within the objective.
Moreover, all the facts included in the objective should in principle be open to being stated by creatures with any perceptual apparatus (although some facts would be hard to verify with the wrong perceptual apparatus). Colours and pitches as perceived by human beings might not be detectable directly by creatures with the wrong apparatus. They would have to say "to human beings, this object is blue", or something similar. But they could say it like that, picking out the fact of blueness to human beings. Human beings could identify the same fact. It is just that they would not normally bother to say "to human beings".
What we have said so far defines the appropriateness of a conceptual scheme in terms of leading to the facts. This fits very well with Rand's approach. The facts are there, whether we grasp them or not. And we have to make use of our perceptual apparatus and develop our conceptual scheme in order to improve our grasp of the facts. Now we shall look at how the development of conceptual schemes may proceed. In so doing we shall get a better idea of how we can tell whether our concepts are indeed appropriate, so that we can have confidence that the purported facts we identify are indeed facts.
For Rand, our concepts arise neither in the intrinsic world alone (a realist position) nor in our consciousness alone (a nominalist position). Rather, they arise out of the need of consciousness to get a grip on the intrinsic world, with the grip being tested by reference to the perceptual appearance of the intrinsic world. They may be regarded as appropriate concepts when there is no conflict between the purported facts that are stated using the concepts and our observations of the world.
An important stage in the development of concepts which will give us a grip on the world is the process of omitting inappropriate details, a process that Rand calls measurement-omission. For example, to arrive at a sensible concept of a table we need to focus on the presence of a flat surface that can support objects and of legs that keep the surface off the floor. We should omit specifics of size, colour, or the wood or metal used.
Inappropriate details are those which should be ignored because they would destroy the concept by excessive fragmentation or would be irrelevant given the nature of the concept.
On fragmentation, the concept of an animal remains valid even after cats are distinguished from dogs. To incorporate details of the specifically feline or canine into the concept of an animal would be to destroy it by fragmentation.
On irrelevance, it is for example in the nature of the concept of a cat not to mention physical location, so location would be an irrelevant detail. We are of course free to mention the locations of specific cats, and we might regard it as significant that a given subspecies only bred in a specific place. But we might not incorporate location even into the concept of that subspecies, because we might want to allow accidental genetic twins in other parts of the world to be covered by the same concept.
Error is for Rand an issue in relation to all concepts, from the most fundamental to the most superficial. This is because for her we gain, or through error fail to gain, a grip on the intrinsic world.
Specifically, Rand could not use a defence against error that Kant could use in relation to the most fundamental concepts. Kant could say that we simply had to use those concepts. Their use would not yield a false picture of the world as it was in itself, because it would not give us any picture of that world. It would only give us a grip on the world as it appeared to us. And since the concepts moulded that appearance, the result would necessarily be a grip on the world as it appeared to us. There would therefore be no such thing as error through choosing the wrong fundamental concepts.
For Rand, there is no such recourse to a mere appearance in relation to which any concepts would be bound to be appropriate because they gave us only a grip on the world as it appeared under the influence of precisely those concepts. And we cannot ignore the possibility of error, given that different people could use conflicting sets of concepts.
The response of Rand is that we must choose the right concepts to fit with the intrinsic world and yield a statement of the objective. If we choose the wrong concepts, we shall eventually find this out because the result will be a set of beliefs that contradict one another or that give us no grip or a poor grip on the world. (This is a point at which percepts do real work, alongside the work they do in guiding the development of our concepts. If what our concepts incline us to think is misaligned with our percepts, we can tell that something is wrong.) The objective, the set of actual facts rather than erroneous fictions, is to be discovered by getting things right, under the harsh discipline of the intrinsic world.
Conceptual schemes can become more refined, capturing more details of the world. The objective can be seen as captured in more or less detail by more or less refined conceptual schemes. Moreover, progress in that way would not require the rejection of facts established using less advanced conceptual schemes.
Conceptual schemes can also change more radically, in ways that require rejection of what had earlier been thought of as facts. The development of relativity and of quantum mechanics are examples.
This is not in itself an objection to Rand's approach. Indeed it reinforces her message that we must find the concepts to fit the intrinsic world, and not expect the intrinsic world to mould itself to our concepts. But given the risk of conceptual revision, how can we have confidence that what emerges as the apparent objective is anywhere close to the relevant parts of the objective. It might be primarily a product of our conceptual scheme, with our appreciation of the intrinsic world having been distorted by that scheme.
The answer is that we cannot be wholly confident. But we can have reasonable confidence, in the sense that our apparent objective is as good as we can get for the time being and we have no specific reason to think that we are mistaken. (Occasionally we do have specific reason to think that while we may be on the right lines, there are still defects in our understanding.) We may also be able to celebrate the fact that what we currently take to be a statement of relevant parts of the objective is empirically adequate to a very high degree: there is no direct clash between the purported facts that we state and the observations we are so far able to make. And we may argue that our taking of sets of purported facts to be parts of the objective is regulated by the intrinsic world because it is sometimes precisely because of observations that we revise our view of what the facts are.
We should not however be wholly confident that we are making steady progress toward a full statement of the objective along a path that will avoid the need for any significant backtracking. It would be possible for us to pursue a line of conceptual development, and to gather data in ways that were shaped by our concepts, such that we tended toward a merely locally optimal set of concepts and apparent grasp of the intrinsic world which was still seriously defective. A large-scale rethink would then be required, although it might be many years before we discovered that this was so.
Different conceptual schemes would give rise to different relations with the intrinsic world. Sometimes the outcome would be the same objective, because the differences between schemes merely counteracted differences between percepts that were produced by different perceptual apparatus. But sometimes the outcome would be different apparent objectives. This would not however mean that there was really more than one objective or that relativism could gain a foothold.
We should start by discarding mistaken conceptual schemes, on the basis of their leading us into contradiction or on the basis of conflicts between their predictions and our actual percepts.
After discarding mistaken schemes, we should be left with access to a single objective that could be set out in ways which were more or less refined, depending on the specific concepts from a single scheme that we chose to deploy. (The ways in which the objective was set out might or might not all be reducible to some fundamental way. The idea of a single conceptual scheme does not require us to believe in universal reducibility.)
At least that could be expected in the natural sciences, although even there it would not be safe to assert that the current conceptual scheme would last for ever, only that it should be regarded as the appropriate one given the evidence currently available. We would in practice only be able to say that we had high confidence that we had reached the objective, because we could not be wholly confident that all mistaken schemes had been discarded.
In the social sciences and the humanities we might expect more scope for there to be schemes that yielded purported facts which were inconsistent with one another, while the various schemes were all considered to be acceptable for the time being so that no scheme could justifiably be selected as the single appropriate one.
The need to allow for that possibility is however a challenge to the notion that relevant assertions made in the social sciences and the humanities are facts. Where there is scope to put different conceptual schemes to work and conflicting purported facts emerge from the interactions between the different schemes and the intrinsic world as presented through percepts, we should hesitate to regard any of the purported facts as parts of the objective at all. The most we could confidently say would be that all but one of the schemes which yielded conflicting purported facts might at some time be discarded, so that some of the purported facts might justifiably come to be regarded as facts.
It might seem that this way of keeping relativism at bay was circular. It seems to amount to saying that if relativism appears to get a foothold, then it is not the objective that is discovered, or at least that not more than one of any set of conflicting facts can be regarded as part of the objective. That is, relativism about the objective is excluded by fiat.
There is truth in this charge of circularity. It is assumed that there is a single objective to be found, so contradictory purported facts cannot be admitted. And a consequence is that not every study of the world can be counted as consistently uncovering the objective. But the success of some studies of the world, particularly in the natural sciences, gives us every reason to think that there is an objective to be uncovered. And there is plenty of scope for the objective in the social sciences and the humanities too. Many conclusions survive shifts between conceptual schemes.
The ideas of Immanuel Kant and of Ayn Rand are far apart, and in some respects directly opposed. Nonetheless we can explore scope to modify their ideas in ways that might allow some reconciliation, and scope to find common ground even without modification.
One way to seek a reconciliation between Kant and Rand would be to make Kant's world as presented by phenomena as good to Rand as the reality of which she speaks. But Rand would require reality as presented by phenomena to be the full extent of reality.
Kant would agree that reality as presented by phenomena was all the reality we could talk about in any detail. And the fact that given a certain intrinsic world and a certain set of concepts, Rand's objective would have to be as it was, would be entirely acceptable to Kant. His position is one of empirical realism. In his view we make empirical discoveries, with the world and not our preferences being dominant. That thought fits well with Rand's approach.
But for Kant, there are still the noumena. The noumena feature in a metaphysical theory. The supposition of the noumena is motivated by a view that certain metaphysical questions make sense, even though all attempts to tackle them as if they were questions about the world as we experience it would end in paradox and confusion, and by the need to find a place for the free will of human beings. But nothing about the noumena features in any natural science, and any attempt to learn anything specific about noumenal reality would be fruitless. Alas for any hope of reconciliation between Kant and Rand through matching the phenomenal world with the intrinsic world, being all the reality worth talking about in any detail would not be the same as being the full extent of reality, something that Rand would demand.
This might seem an odd demand to infer from Rand's work, since she spoke both of intrinsic reality and of the objective. But for her intrinsic reality, and nothing else, could be paired with a conceptual scheme so that the objective could be identified as a relation between those two elements. There is no space for two sorts of reality, or even for one reality considered in two different ways. The objective is not reality viewed differently. It is the view of reality that we get by putting the appropriate concepts to work. For Kant on the other hand there is a reality as it appears to us, and we then perceive it. This would be a real point of disagreement between Rand and Kant. Rand's assertion is not a kind of logical positivist one that there would be no sense in talking of anything apart from intrinsic reality. Rather, it is a metaphysical assertion that there really is nothing else.
Even if it were a logical positivist assertion, that would not be acceptable to Kant. He would agree that nothing specific could be said about the noumena, but he would also require that reference to the noumena not be regarded as vacuous.
Both Kant and Rand are concerned with what we know. For Kant, a study of our powers to know leads on to the conclusion that what is known is not the qualities of the world as it is in itself. For Rand, anything that counts as knowledge is knowledge of the nature of the intrinsic world. At that level, the conflict between Kant and Rand persists. But the centrality of knowledge to the approaches of both of them leaves open the hope of some agreement at the level of the nature of the act of knowing, and in particular in relation to the process of the application of concepts to get to grips with the world.
For Kant, there are certain forms of intuition (space and time) and concepts that we just have to apply. The need to use them reflects our nature, and the necessity of applying them should not mislead us into thinking that they reflect the nature of the world as it is in itself.
For Rand, there are three axioms that we must at least implicitly adopt if we are to think about the world at all. These are the axioms of existence, identity, and consciousness. The axiom of existence is that existence exists, or to spell this out a bit, that there really is a world out there, the intrinsic world, which is independent of our consciousness of it. The axiom of identity is the somewhat gnomic "A is A". To spell this out, each thing has the specific nature that it has, and its characteristics constitute its identity (Rand, The Ayn Rand Lexicon, entry for Identity). As a practical corollary, one must face up to the natures of things as they are. One can take action to change things in the world, but merely wishing that things were different or trying to view them using inappropriate concepts will leave them as they are. The axiom of consciousness is that one exists possessing consciousness, which is the faculty of perceiving that which exists. As Rand puts it to pull the picture together, "Existence is Identity, Consciousness is Identification" (Rand, The Ayn Rand Lexicon, entry for Identity).
The axioms underpin a way of thinking that Rand picks out as fundamental. She argues that a crucial step on the road to making sense of the world is to recognise units, such as a person or a table. These units are then grouped together under concepts. A modest degree of conceptualisation is needed to pick out units at all, as when someone identifies "that thing". But it is only once units have been picked out that a person can group them on the basis of some appreciation of similarities and differences, and then enquire into their characteristics so as to define concepts which allow sensible groupings. Moreover, even to get as far as picking out units in a way that will allow the application of concepts, things must be out there in the world (existence), they must have properties by which they can be recognised and grouped (identity), and we must be conscious of them (consciousness). We might get as far as picking out units and grouping them merely because existence, identity and consciousness prevailed, without our being aware of their prevalence. But in order to achieve knowledge at anything like the level we have in fact achieved we need to think about what we are doing, reflect on what we know or seem to know, and consciously work out what do do next in the way of making further empirical investigations or defining new concepts. So we need to be fully aware of existence, identity and consciousness.
The axioms are not concepts. But they correspond to the three axiomatic concepts, the concepts we must put to work, the concepts of existence, identity, and consciousness. And acceptance of the axioms and the concepts does parallel Kant's requirement to use the forms of intuition and the concepts he lists in the sense that acceptance is required in order to make any progress. There is even a noteworthy parallel in the role accorded to time, the form of inner intuition for Kant and something fundamental to our grasp of the world for Rand. In chapter 6 of Introduction to Objectivist Epistemology Rand notes that it is only once we consciously take on board the concepts of existence, identity and consciousness that we can make real progress by understanding continuity and change in the world. And continuity and change are essentially temporal concepts.
There is however a difference, particularly from the use of Kant's forms of intuition. For Rand, our thought must accord with her axioms in order for us to have any hope of getting things right. For Kant, we have to perceive in the ways that use of his forms of intuition ensures. So for Rand it is about thought, while for Kant it is about perception.
Having said that, the relevant sense of perception in Kant's thought is not the basic level of incoming sensations but the more elevated level of organising them, a level which arguably involves a degree of unconscious thought. This does however correspond to Rand's stage of percepts, which follows the stage of sensations but precedes the stage of the application of concepts.
For Rand, the one and only world is what we come to know. And while our knowledge is never likely to become complete, we can in principle explore everything there is, with nothing beyond our reach for reasons that would be supplied by metaphysics or epistemology. We may be delayed by reasons supplied by the sciences. It may for example be very difficult to make instruments of a desired refinement. And we may, particularly in fundamental physics, find ourselves not knowing whether there is information about what particles are doing that is inevitably hidden from us or whether there is no fact of the matter about what particles are doing. But our being faced with such a disconcerting dilemma would not reflect some inevitable metaphysical or epistemic principle. Rather, it would merely reflect the inappropriateness of our expectation that the microscopic world should conform to the style of the macroscopic world.
Another important point is that for Rand, the world has its own stability. Things do endure, and they participate in causal relationships in accordance with their natures. This makes knowledge possible. Kant would likewise be happy to allow stability in the sense of endurance, natures and causal relationships in the world as it appeared to us. But for Rand, the stability we find when we learn about the things in the world and connect their natures to their causal powers is a stability of the world as it is in itself. Kant is by his own theory debarred from even speculating about the stability of things in themselves.
An important part of Rand's picture is that we are elements in the world, whose perceptual and intellectual capacities are given by the nature of the world itself. We can explore how our eyes and brains work, and what we have discovered or may eventually discover is what there is to be known about our intellectual functioning. Our abilities to identify and re-identify objects, to conceptualise, and to trace chains of causes and effects give us immense power, but they do not make us beings outside the world. We are in the world and what we will discover is what is going on around us in that world. Moreover, there is no mystery about how we know. Our processes of perception and conceptualisation are natural, not magical.
There is a contrast here with Kant, who sees us as rational creatures with a true nature that is not to be defined in the terms of the world as it appears to us (although Kant would of course allow research into the eye and the brain). This view goes hand in hand with the idea that there is a true nature of the world which is outwith the grasp of empirical exploration.
Kant's view may also make it somewhat mysterious how we perceive the world. The one sure way to remove the mystery would be to adopt a firmly one-world interpretation of Kant that took an empirical description of our processes of perception to be a good guide to how those processes really worked. That would however risk making claims about us as we were in ourselves, claims of a sort that Kant would forbid.
Rand and Kant have markedly different theories of how the world is, and those differences spill over into differences as to the nature of our knowledge. But there are some points at which their two lines of thought can be related to each other. And their basic questions about the world and about knowledge are the same, even if their answers are different.
Kant, Immanuel. Critique of Pure Reason.
Rand, Ayn. Introduction to Objectivist Epistemology, expanded second edition edited by Harry Binswanger and Leonard Peikoff, with an additional Essay by Leonard Peikoff. New York, NY, Meridian, 1990.
Rand Ayn. The Ayn Rand Lexicon: Objectivism from A to Z, edited by Harry Binswanger and with an Introduction by Leonard Peikoff. New York, NY, Meridian, 1988.
Wittgenstein, Ludwig. Tractatus Logico-Philosophicus.
This post is about different ways to try to reduce the tax one pays, and about relationships between those ways, the law, and ethical considerations.
This post is not advice on tax law. Anyone with questions or concerns about the position of an actual taxpayer should seek appropriate professional advice. We will not provide advice or recommend an adviser.
The focus is on the UK. We shall therefore refer to the tax authority as HMRC, which is short for HM Revenue and Customs. Other countries handle some of the issues differently, but the same general style of enquiry may be useful in relation to those other countries.
We have written two earlier blog posts on issues related to legal provisions we shall discuss:
There is a traditional distinction between three things one can do in order to reduce the tax one pays: tax planning, tax avoidance, and tax evasion.
Broadly, planning involves things like putting money into tax-privileged investments such as pension funds, or ensuring that a home bought to rent out becomes the owner's main private residence for certain periods so as to minimise the proportion of the gain on sale that is taxable, or structuring a group of companies so that if one company makes a loss, it can be netted off against the profits made by another company to reduce the amount that is taxable. Avoidance is planning that looks too much like the exploitation of loopholes. It often involves complex structures and transactions that have no purpose other than to save tax. It is common to speak of avoidance schemes, where a scheme is a complete structure and set of transactions. Evasion is the provision of false information to HMRC, or the withholding of information, in order to pay less tax.
There is also a well-established distinction between different consequences of actions that in some way harm others. This is the distinction between restitution, making good the loss suffered, and penalties. In the context of tax underpaid, restitution arises in the form of payment to HMRC of the tax that would otherwise have gone unpaid.
Traditionally, tax avoidance which did not succeed because HMRC argued successfully for an interpretation of the law that left no loophole only led to restitution. The tax which the taxpayer had hoped to avoid had to be paid, perhaps with interest, but that was all. There was nothing illegal in setting up complex structures and carrying out complex transactions. They might simply fail to achieve their aim. Evasion, on the other hand, traditionally led to penalties.
We shall explore categories that are not the traditional ones, although they are related to the traditional ones. We shall do so because our alternatives seem to be more useful in understanding certain issues, particularly some relationships between the legal position and the ethical position.
Our first category, which we shall call failed avoidance, covers tax planning or avoidance (we can run them together) that does not achieve its aim, for legal reasons rather than because some mistake was made in implementing the planning or avoidance.
Our second category, which we shall call unethical avoidance, covers tax planning or avoidance that on ethical grounds should not be undertaken.
Our third category, which we shall call penalised conduct, covers conduct that is liable to lead to penalties as distinct from mere restitution. We shall however not be concerned with the deliberate falsification or concealment of information.
Our fourth category, which we shall call culpable conduct, covers conduct that on ethical grounds should lead to penalties.
One of our questions will be that of the sharpness of the boundaries of these categories. Another will be that of the match or mismatch between the boundaries of the first and second categories, and between the boundaries of the third and fourth categories. We shall also be concerned with the extent to which, contrary to tradition, failed avoidance can amount to penalised conduct.
We shall start with failed avoidance, then move on to unethical avoidance and relationships between failed and unethical avoidance. After that, we shall we shall move on to penalised and culpable conduct. We shall then consider the ethics of legislating in certain ways. We shall conclude with a discussion of categories and general ethical principles.
Tax planning can be very straightforward, for example putting money into a pension fund to get a tax deduction now and tax-free investment returns, subject to taxation (perhaps at a lower rate) when the pension is paid out. At the other extreme it can involve very complex structures and transactions, often using multiple companies, trusts and the like and mixtures of share capital and loans, all spread across several jurisdictions, in an attempt to ensure that the overall tax rate on a group's worldwide profits is far less than one would naturally expect it to be.
HMRC will seek to prevent tax avoidance from succeeding when the tax at stake is substantial in an individual case, when it would be substantial if a scheme were widely used, or when the avoidance is considered to be brazenly contrary to the policy of the tax system. Complexity is not essential for any of these conditions to be met, but it is quite common for schemes to be complex and to rely on loopholes in the law that may not have been apparent when it was drafted.
Avoidance in which the necessary steps were executed correctly can only be prevented from succeeding if the legal tools are available. Some avoidance is blocked by specific legal provisions, often ones which were inserted as soon as the form of avoidance in question came to light. But the enactment of such provisions can easily lead to the creation of new schemes that stay just outside the ambit of the new provisions. More significant for our purposes are general provisions that are intended to lead to the failure of classes of schemes defined by some aspect of their overall nature or by their results, rather than by their details. We shall look at two such provisions, the General Anti-Abuse Rule (the GAAR) and the Diverted Profits Tax (the DPT).
In 2013, the UK introduced the General Anti-Abuse Rule (the GAAR). Links to the legislation and the guidance can be found here:
The basic idea is that a tax avoidance scheme does not succeed in delivering the expected tax saving if it fails the double reasonableness test. In the words of Finance Act 2013, section 207(2):
(2) Tax arrangements are "abusive" if they are arrangements the entering into or carrying out of which cannot reasonably be regarded as a reasonable course of action in relation to the relevant tax provisions, having regard to all the circumstances including -
(a) whether the substantive results of the arrangements are consistent with any principles on which those provisions are based (whether express or implied) and the policy objectives of those provisions,
(b) whether the means of achieving those results involves one or more contrived or abnormal steps, and
(c) whether the arrangements are intended to exploit any shortcomings in those provisions.
One argument for the double reasonableness test is that it makes it quite difficult for tax avoidance to be caught by the GAAR. This is legislation with a much less well-defined scope than most of the legislation that targets specific tax avoidance schemes. It has to have a fuzzy boundary of application because if it had a precise boundary, avoidance schemes that stayed just the right side of the boundary would soon be devised. So to be fair to taxpayers, the grey area created by the fuzziness of the boundary should be such that it is easy to keep a safe distance away from it. Routine tax planning is definitely safe because it would be perfectly reasonable to regard it as reasonable. This is so even though there are also reasonable positions from which it would be regarded as unreasonable, for example a position that people should act without regard to tax considerations and then just accept whatever tax consequences happened to follow.
We should note the connections between the general notion of reasonableness and the circumstances to consider that are listed in section 207(2)(a), (b) and (c).
The effect of the circumstances mentioned in (a) and (c) is to link the notion of reasonableness to the intention of Parliament as it may be inferred from the relevant legislation and to the policy intention (an intention of the Government rather than of Parliament). Unreasonableness is indicated, although not demonstrated, by action which would frustrate those intentions. And (b) draws attention to contrived or abnormal steps as an independent indicator of unreasonableness.
Thus the notion of reasonableness at stake is closely tied to what Parliament and the Government at the time wanted the law to achieve. This is not surprising. The whole point of the GAAR was to give a new way to stop taxpayers dodging the intended effect of the law. But it will raise the question of whether what one might regard as unreasonable on general ethical grounds would align with what would be caught by the double reasonableness test of the GAAR. Has the GAAR defined a boundary of ethical significance, or merely one of governmental convenience?
The Diverted Profits Tax (DPT) was introduced in 2015. The legislation is in Finance Act 2015, part 3, and may be viewed here:
This is a tax in addition to corporation tax, not an element within corporation tax. It is imposed on profits which would have been subject to UK corporation tax if avoidance steps had not been taken.
For the tax to apply, the taxpayer must have taken steps which lacked sufficient economic substance (sections 80(1)(f) and 86(2)(e)). This condition is defined in section 110(4), (5), (6) and (7). There are two parts. It must be reasonable to assume that the steps were designed to secure a tax reduction. And it must not be reasonable to assume that the non-tax benefits would exceed the tax benefits (section 110(7)(b) introduces a variation on this relative benefits rule which we shall not discuss).
We here have a different kind of reasonableness test from that used in the GAAR. The important difference is not between single and double reasonableness, with reasonable-design replacing reasonable-reasonable. Rather, it is between the DPT's "it is reasonable to assume" design to secure a tax reduction and the GAAR's "cannot reasonably be regarded as a reasonable course of action". The use of "cannot" in the GAAR leaves scope to survey all reasonable views in search of one favourable to the taxpayer. The use of "is" in the DPT pushes us to suppose some authoritative standard of reasonableness on the basis of which various possible sets of steps taken by a taxpayer could be divided into those it was reasonable to assume were designed to secure a tax reduction and those about which it would not be reasonable to make that assumption. The GAAR avoids the need to suppose an authoritative standard by allowing the full range of reasonable positions to be surveyed.
While inability to survey a full range of reasonable positions would make it harder for a company to be confident that it was safe from the DPT than it would be to be confident that it would be safe from the GAAR, because the inevitable grey area would be closer to regular commercial practice, there is another difference that might, but need not, pull in the other direction. The GAAR looks directly at what the taxpayer did. The DPT legislation looks at whether what the taxpayer did appeared to be designed to reduce tax. That is, application of the DPT requires consideration of design.
Having said that, it is only design as may be inferred from action that matters, not actual design as might be disclosed by internal company documents. To avoid its being reasonable to assume that transactions were designed to achieve a tax reduction, one would at a minimum have to argue that the transactions could easily have been chosen without tax reduction in mind. This much would be needed because section 110(9)(b) allows there to be design to secure a tax reduction despite there also being design "to secure any commercial or other objective".
There is however one final safeguard for the taxpayer. This is the rule that the DPT does not apply if it is reasonable to assume that the non-tax benefits would exceed the tax benefits.
There is a penalty for failure to notify that the DPT might apply. It can be up to 100% of the tax at stake. Notification does not however imply an admission that the tax applies.
Tax avoidance may be seen as unethical. It at least tends to be against the spirit of the law, which is to collect tax at prescribed rates on income, gains, sales and so on as evaluated in a common-sense way. If for example someone receives income of £90,000, and there is no express provision to treat income of that kind as being less than is received, they should pay income tax on that amount minus the personal allowance at the applicable rates of 20% and 40%. It would be contrary to the spirit of the law to find a loophole which meant that tax was only levied on £20,000.
The ethical case against avoidance could be strengthened by pointing out that the avoider relied on goods and services provided out of taxation, such as education, healthcare and roads. Such goods and services are not only used directly in daily life. They also make it possible to run businesses so that people can have jobs or make investment returns. That is, they help to generate the income on which tax is supposed to be paid. If a democratically elected Parliament has decided how the cost of the tax-funded goods and services is to be shared out, it would seem to be unethical to dodge paying one's allocated share.
A contrary view would be that tax was an artificial imposition, with ethical obligations being defined entirely by the words of the law so that the only unethical conduct would be to act contrary to those words. If an avoidance scheme navigated round the words of the law without actually breaking any of the rules, nobody would have any ground to condemn the avoider as unethical. It would even be possible to take a positive view of avoidance, on the ground that the state had no business violating individual property rights without individual consent (as opposed to collective consent).
There is very unlikely to be universal agreement on the boundary between acceptable and unethical tax planning or avoidance. Some people think that if the law allows a piece of tax avoidance to succeed it is ethically acceptable, however contrived it may be. Others think that anything in the least bit contrived is ethically unacceptable. And there are plenty of options in between, allowing that certain degrees of avoidance are acceptable but condemning more extreme avoidance.
It is not surprising that there should be a wide range of views. There are many factors to consider in political questions, and several angles from which those questions may be viewed. We do not find the single dominant factors or angles which may allow there to be general agreement on conclusions such as the one that physical violence is wrong.
Not only is there no general agreement on where to place an ethical boundary. Almost any boundary would be hazy. The only positions that would be guaranteed to yield sharp boundaries would be extreme ones: that all tax planning was wrong, so that people should make their commercial decisions as if tax did not exist and then accept whatever tax consequences arose, or that all tax avoidance, however contrived, was ethically acceptable.
Even an ethic that was intrinsically disposed to precise conclusions as to what to do in given cases would be unlikely to help. A Kantian ethic might tell taxpayers to act from a good will, and to consider the feasibility of universalising the maxims on which they acted, but any one of a range of principles on which to approach tax planning, or indeed any one of a range of possible decisions on particular occasions, could easily comply with such requirements. And a Benthamite approach of maximising happiness could not be used to work out either principles on which to act or which choices to make on particular occasions, simply because of the impossibility of working out what would in fact do the most to promote human happiness. On the one hand it could be argued that the democratically controlled state was the best guardian of resources to be applied for the common good, so that tax planning should be minimal. On the other hand it could be argued that the flourishing of the economy would be vital, that this would demand not letting too much wealth leach into the public sector, and that the state was incompetent at allocating resources, so that tax payments should be minimised.
We shall now consider the relationship between the boundary that divides effective from failed avoidance and possible boundaries that would divide acceptable from unethical avoidance.
Perfect correspondence in fact, so that tax planning was ethically acceptable if and only if it was legally effective in the sense that it would lead the planner to keep their envisaged tax saving, seems unlikely. One would expect it if there were a general ethical principle that whatever planning the law did not defeat was acceptable, and whatever it defeated was unacceptable. One might argue for the first leg of this general principle on the basis that there was no natural morality about the levying of taxes, although on that basis ethical acceptability would amount not to endorsement but only to the absence of ethical objection. And one might separately argue for the second leg of the general principle on the basis that there was an ethical requirement to show respect for the law which went beyond abstention from illegal actions and extended to abstention from actions, the hoped-for consequences of which would be frustrated by operation of law. (Remember that tax planning, even aggressive avoidance, is in general not disobedience. There is in general nothing illegal about setting up complex commercial structures, even if the purpose is tax avoidance. The law may merely render the use of such structures ineffective for tax purposes.) But it would be very hard indeed to argue for both legs at the same time, unless one were to assert that the law was in the areas of life that it covered the definitive guide to the ethical. That would be to allow legal positivism to steamroll its way across ethical discourse.
The GAAR relies on the double reasonableness test: could the tax arrangements reasonably be regarded as a reasonable course of action?
The resultant haziness as to which tax avoidance would be defeated by the GAAR is inevitable, given that precise rules directed against specific avoidance schemes routinely leave scope to devise new schemes.
There is a special contribution that ethical thought might make to deciding specific cases. Ethical thought might contribute to deciding what would be reasonable, or at least (taking account of the double reasonableness test) to deciding what could reasonably be regarded as reasonable. What follows here on this point is however speculative. The potential scope to contribute is not recognized in current legislation, and attempts to make use of ethical thought might get nowhere in court. We include the topic here in order to suggest that ethical thought might have a role to play in the writing of future tax law or in the development of approaches to deciding cases.
A Kantian approach would best be suited to saying what could reasonably be regarded as reasonable. An understanding of the notion of a good will and of the categorical imperative would not suffice to say what would be reasonable in tax planning. The factors to consider would be too complex for that. But a claim that some particular piece of tax planning could reasonably be regarded as reasonable could be tested by asking whether it was plausible that the action might be inspired by a good will, or whether it was plausible that the maxims behind the tax planning could be universalised. (We here mean maxims specific to planning of the relevant sort, and do not envisage a single maxim or set of maxims that would cover tax planning of all sorts.)
A utilitarian approach would likewise best be suited to saying what could reasonably be regarded as reasonable. The only actions that would in fact be reasonable, according to such an ethic, would be actions that maximised total happiness. But it would be far too hard to work out consequences of particular pieces of tax planning or of general approaches to tax planning. On the other hand, it might be feasible to say that it was reasonable to regard some particular piece of tax planning, or some particular general approach, as having a plausible prospect of maximising total happiness.
Virtue ethics may have a better prospect of leading to judgements as to which specific pieces of tax planning, or general approaches, were in fact reasonable. A piece of tax planning or a general approach could be judged on the basis of whether it accorded with specific virtues. Honesty would be an obvious virtue, but so long as we were not considering evasion dishonesty might not be in question in any case. Community spirit might be a relevant virtue (although there are philosophies, such as that of Ayn Rand, which would not see it as a virtue of any significance). Paying for resources on which one relied even though they were technically provided free of charge at the point of use, such as the education of employees, might be a virtue. And so on.
We have noted the distinction between two two types of demand for payment that HMRC may make.
Restitution is making good a loss, for example when tax avoidance fails and the tax which it had been intended to save is payable after all. The restitution is of the loss to the state. It is not exactly like damages payable for loss in a civil action between private individuals or companies, but the basic idea is the same. Adversely affected parties are to be restored to the financial positions in which they would have been had something unfortunate not happened. Most importantly, for any payment to amount to restitution, its amount must be determined by reference to the loss that arose.
A penalty is a punishment that is inflicted for reasons other than restitution, for example in order to deter. The severity of a penalty may happen to be determined by the size of the loss in question, as when a penalty for making an inaccurate tax return is a percentage of the tax that would have been lost. But there is no need for there to be a link between the size of the loss and the size of the penalty.
Penalties do of course apply when a taxpayer deliberately falsifies tax returns, or conceals information that is required. Such actions fall under the traditional heading of evasion. We shall not be concerned with such conduct here.
One would naturally expect that failed avoidance would lead only to restitution. Indeed, that has been the traditional position. There is in principle nothing illegal about setting up complex structures and carrying out complex transactions, even if the aim is to reduce tax liabilities.
Things have however changed. Failed avoidance can incur penalties, and can do so even when all the structures and transactions are in themselves perfectly legal and full disclosure has been made to HMRC. We shall consider two examples here. The first one relates to mistaken claims to successful avoidance. The second one relates to the GAAR.
If a taxpayer completes a tax return on the basis that an avoidance scheme succeeds, when in fact it fails, the return will be inaccurate and there may be penalties for that reason, a reason which is distinct from the creation of a complicated commercial structure and its use to engage in complicated transactions.
This can happen even if the structure and the transactions involved are disclosed in full, and any applicable requirement to notify the use of an avoidance scheme or of transactions that may suggest avoidance is also met. (The use of some specific schemes must be notified. And transactions with certain characteristics that are typically associated with avoidance must also be notified, even if no avoidance is in fact involved.)
The return can be inaccurate because under the self-assessment regime that has prevailed in the UK since the 1990s, the taxpayer is responsible for working out what tax they owe. So if a taxpayer has engaged in tax avoidance that on close inspection would be found not to save tax, they should state their tax position without taking account of the purported saving. The normal penalty would be the one for carelessness in making an inaccurate tax return, although other penalties would also be possible (Finance Act 2007, schedule 24, paragraph 3A). The penalty for carelessness typically ranges from nil to 30% of the tax at stake (more if transactions involving entities in certain countries are involved), but if HMRC find out about the error before the taxpayer discloses it unprompted the minimum penalty is 15%.
In relation to the GAAR, the UK has adopted an approach which allows penalties to be imposed merely because avoidance fails. A taxpayer may be put on notice that HMRC believe the GAAR to frustrate a piece of tax avoidance. If the taxpayer does not promptly agree and re-assess their tax as if the avoidance did not succeed, a legal challenge may follow. If the taxpayer loses, they must pay not only the tax not saved but also a penalty of 60% of that tax. The only way to guarantee not to suffer this penalty is to concede that the scheme is defeated by the GAAR as soon as HMRC claim that it is so defeated. If the taxpayer fights the case and loses, the penalty will be due.
The absence of general agreement on an ethical boundary, together with the fact that the steps taken in tax avoidance are when considered individually often steps that could just as well be taken for other reasons, steps such as setting up trusts in order to safeguard assets from the spendthrift or making loans in order to finance business ventures, make it arguable that when tax avoidance is frustrated by law the consequences should be limited to restitution and should not include a penalty. The law has to draw its boundary between effective and failed avoidance in one place, and there is not much prospect of general agreement that it is ethically the right place. So there is not likely to be general agreement that a given taxpayer has acted in a way that would make penalties appropriate in addition to restitution.
This puts the spotlight on cases in which the approach is to impose penalties despite the fact that all this is involved is failed avoidance. Can it be right to do so, given the legality of the underlying structures and transactions?
One argument in favour of such penalties would be that the law could draw a secondary boundary within failed avoidance, between respectable and outrageous avoidance. Another argument would be that penalties would deter speculative avoidance which was unlikely to succeed unless HMRC failed to look at it. A third argument would be that while the structures and transactions might be perfectly legal, the motive would very often be one of tax reduction. Then the taxpayer should be regarded as unethical and the state should not be expected to pull its punches.
There are counter-arguments to all three of these. To take the first argument, while a legally defined boundary between respectable and outrageous avoidance might be said to exist already in the double reasonableness test of the GAAR, there would be little prospect of general agreement on the location of a corresponding ethical boundary. The deterrent effect in the second argument might go too far. It would extend to avoidance that had a reasonably good prospect of succeeding, and the state should not deter claims to what might well turn out to be the taxpayer's legal entitlement were the matter to go to court. (It is a feature of such cases that it often cannot be known in advance of a court hearing exactly what the law would require.) Finally, the argument that unethical conduct on the part of the taxpayer should allow the state a freer hand than it would normally have would introduce notions of fair dealing into the tax system in a very strong form. Not being bullied by the state would become dependent on being kind to the state.
The GAAR brings such issues sharply into focus. It may easily be unclear whether some tax avoidance would be defeated by the GAAR. There is guidance on the GAAR, but it is not going to answer every question, and in any case guidance is not the law under which the question of whether tax avoidance succeeds is determined. And yet Parliament has legislated to impose a penalty of 60% of the tax at stake on avoidance that fails by virtue of the GAAR, the only sure protection against which is to concede as soon as HMRC challenge the avoidance.
For the penalty actually to apply, the avoidance must be established to fail. Then it would be a penalty like the general penalty for making a return that was inaccurate by virtue of allowing for avoidance that was found to fail. But the scale of the penalty would be doubled, from a usual maximum of 30% to a fixed 60%, if the taxpayer, at the earlier stage of not knowing whether the avoidance would succeed, had the temerity to challenge HMRC's view that the avoidance would fail. Such an encouragement to taxpayers not to stand up for themselves but instead meekly to accept HMRC's view of the law is at the very least ethically dubious.
To raise ethical objections to this kind of penalty is not to endorse tax avoidance. The kind of avoidance involved will nearly always be highly contrived, and would not have been instigated unless the taxpayer was determined to pay significantly less tax than the general principles of the tax system would lead one to expect. But such considerations do not suffice to give the state a free pass.
We should add in favour of the state that a decision to challenge avoidance under the GAAR would not be made lightly or on the say-so of a single tax inspector. There is an Advisory Panel, and HMRC must seek the Panel's opinion on each case before invoking the GAAR. The Panel can decide that the GAAR should not be used, for example in the case discussed here:
One may however have reservations about the strictness of the control imposed by the existence of the Panel. While its members are not HMRC staff, they are appointed by HMRC and are supported by an HMRC-staffed secretariat. While HMRC must seek the Panel's opinion on each case before invoking the GAAR, they are not bound by that opinion but are only required to consider it (Finance Act 2013, schedule 43, paragraph 12). And the Panel does not apply the double reasonableness test, but only the test less favourable to the taxpayer of asking whether the taxpayer's actions were in fact a reasonable course of action (Finance Act 2013, schedule 43, paragraph 11(3)(b)).
Thus one can still be concerned about the existence of a penalty for avoidance, the only sure way of avoiding which is to concede to HMRC without taking the matter to a judicial hearing, even if one disapproves of tax avoidance and despite the existence of the Advisory Panel.
We do not here discuss penalties under the Serial Tax Avoidance Regime. We do however note that there would be things to say in connection with that regime about the imposition of penalties, rather than merely requiring restitution, on the basis of the fact that the taxpayer had previously attempted avoidance.
It is a general principle of law that individuals and companies should know in advance what the consequences of various possible actions would be. This argues in favour of clear boundaries, and against hazy ones. Legislation that does not provide such boundaries where it would be possible to provide them is arguably unethical, even if hazy boundaries would make it easier for the state to achieve its objectives.
In the present context this means that it should ideally be clear in advance which kinds of tax avoidance would fail, and which conduct would be penalised (as opposed to merely requiring payment of tax saved by way of restitution).
With avoidance that is defeated by the GAAR, the boundary between successful and failed avoidance is deliberately hazy. A precise boundary would lead to too narrow a range of avoidance being defeated by the GAAR, either because the boundary had been made precise by specifying commercial features of schemes (such as loan transactions that went through intermediate group entities which merely received money and passed it on, or the use of branches which were commercially significant but fell outside the definition of permanent establishments), or because the boundary had been made precise by reference to figures like effective tax rates. In either case, it would not be long before significant avoidance schemes which fell just outside the boundary were devised.
The same case for a hazy boundary could be made in relation to the DPT, although the potential scope of that tax is made a bit more precise by the need for a specific effect, the diversion of profits, which narrows the range of structures and transactions that could give rise to the charge.
In defence of haziness in both these cases, it could be said that those who engage in normal commercial transactions with little focus on saving tax will find themselves to be at a safe distance from the hazy boundary, although this is less true of the DPT than of the GAAR because of the absence with the DPT of an option to survey a full range of reasonable positions.
The argument for clear boundaries is particularly strong when penalties, rather than merely restitution, are involved. We naturally associate penalties with criminal conduct, where conduct must clearly fall within the scope of an offence defined in legislation. But penalties are involved both when the GAAR applies and when the failure of avoidance leads to a penalty for making an inaccurate return, another area of hazy boundaries because it can easily be unclear in advance whether avoidance will succeed even when the avoidance is clearly not within the scope of the GAAR. (The penalty for failure to notify that the DPT might apply is however rather different. It is not in itself enough to make haziness of the boundary objectionable, because notification does not imply liability.)
It might be thought that penalties, and even the GAAR penalty for having the temerity to challenge HMRC's view and then losing, would be acceptable so long as a taxpayer had unacceptable motives.
It might also be thought that dubious motives were indeed present, both when the GAAR applied and when the penalty for making an inaccurate return applied. After all, the pursuit of loopholes and the minimisation of one's contribution to the cost of providing infrastructure, a health service, and so on do not look like good objectives.
We should however recognise counter-arguments.
One is that the duty of taxpayers to the state is laid down in statute, and statute is nearly all drafted in terms of precise rules. The rules must be obeyed, but the limits of the rules are the limits of the taxpayer's obligations. There is nothing in statute about the spirit of the rules, and the relationship of taxpayer to state is not like the relationship of two cricket teams to each other in which the spirit of the game should be observed. (As it happens there is a law of cricket, law 41, which requires observation of the spirit.) The closest one gets in tax law to reference to the spirit of the rules is in references to the principles and policy objectives of legislation and to the exploitation of shortcomings in legislation which we find in the GAAR legislation quoted above (Finance Act 2013, section 207(2)(a) and (c)).
A second counter-argument is that the state gets enough money, and wastes a lot of it because politicians and officials who spend it are not personally at risk of financial loss, so it is good to keep the state on a tight rein by not paying more tax than one must.
A response to concerns about penalties might be that it would be impractical to handle all the tax avoidance there would be if attempting it were a game without risk, in that the taxpayer would in cases of failure merely have to pay the tax they would have had to pay without the attempt. Penalties change the balance of risk, and deter the taxpayer population as a whole. But that would be to say that the need to keep the population cooperative outweighed considerations of justice to individual taxpayers. One could make a case that this was so, but the case could in turn be challenged.
We have proposed four categories to analyse conflicts between on the one hand actions short of the outright criminal to reduce tax payable, and on the other hand the law and ethics: failed and unethical tax avoidance, and penalised and culpable conduct. These categories strike us as more useful than the traditional categories of planning (being sensible), avoidance (being too clever by half), and evasion (lying), when we want to think about relationships between legal rules and ethical principles.
Having said that, the categories we use are not particularly novel. They are clearly related to and inspired by the traditional categories. And the traditional categories still have important roles. They serve to explain how the law has developed. They also help to relate general ethical principles from outside the tax world both to the general scheme of the law and to ethical assessments of what taxpayers do: being sensible is generally accepted both in law and in ethics, being too clever by half may easily be seen as something to be blocked by the law and is also ethically debatable, and lying is both reasonable to punish and ethically unacceptable.
It is not that the traditional categories are essential to underpin judgements as to what is legally appropriate or ethically unacceptable. But whenever we tackle questions of what the rules should be or what is unacceptable conduct in a specific area, we do need some way to relate the question to more general principles. And it so happens that in the area of taxation the traditional categories can perform that role. Since they also need to be kept in play in order to understand the history of the current law and the context of much ethical comment on what taxpayers may do, and they do not interfere with other ways of thinking about the issues, we see no reason to discard them. It is merely that they are not the only useful categories.