The Elimination Fallacy

Posted in:

I haven't found this fallacy discussed anywhere. It might be considered as in instance of “false dilemma” or “Black-or-White”, but it seems to be worth discussing in its own right. It can come up in various contexts - in politics, ethics, or discussing the merits of technical solutions to problems.

The fallacy looks simply like this:

  1. X is bad

  2. Therefore we should eliminate X.

There are also hypothetical versions — “X would be bad, we must absolutely ensure it does not happen”.

The logical appeal of this is obvious. The flaw is also not hard to see - it is simply the law of unintended consequences. But I've seen lots of instances of this fallacy in use.

I'll start with a hopefully fairly uncontentious example from computer programming (skip three paragraphs if you are not a programmer).

We know that duplication in code is bad (Don't Repeat Yourself, Once and Only Once etc.). So we conclude that we must eliminate duplication.

The problem is that 100% elimination of duplication can require massive effort, and introduce all kinds of other problems. So, when I'm faced with the fact that my client side validation is duplicating server side validation, I don't necessarily have to conclude that I need to rewrite my server side logic in NodeJS so that the same Javascript code that runs in the browser runs on the server. I would then have all the problems associated with Javascript [1], not to mention the time it will take me. In another case, I might eliminate some duplication with an abstraction, but it is the wrong abstraction which turns out to be more expensive than the cost of the duplication.

In reality I just need to reduce duplication within the bounds of what is reasonable. The remaining duplication that I cannot eliminate can often be handled using simpler techniques that don't come at such great cost (e.g. a comment left in each place where a piece of information is duplicated about the existence of the other places — this is the technique I use if I have the same constant needed by code written in multiple languages).

However, the problem of the elimination fallacy doesn't lie so much is its initial logical appeal, which we can reason ourselves out of, but in its rhetorical power. Use of the elimination fallacy makes for a massively better sound bite than a more balanced expression. For example:

“Make poverty history!”


“Let's tackle unnecessary poverty and try to massively reduce it, while of course making sure that we don't significantly damage other things that also really matter, such as personal freedom and the ability of people to make decisions for themselves in ways that actually affect their destiny.”

Sometimes the rhetoric is just that, and should be interpreted as such. “Make poverty history” is not in fact a call to eliminate poverty whatever the unintended consequences (for example — the elimination of human poverty could be achieved by exterminating the human race, or possibly by a globally enforced extreme form of socialism, but neither are what we mean). It is a rallying cry, a call to massively reduce our tolerance of hugely unnecessary poverty and injustice — and one I agree with. Speaking in absolutes to communicate a relative truth has a long and honoured tradition [2].

Similarly, at the time of writing, there is the “Never Again” campaign that wants to see some action on gun control laws. The supporters are not actually suggesting any and every means to stop this from happening ever again. Nor are they suggesting a significant loss to anything really important [3].

However, the danger comes when we start with a great rallying cry, and then actually pursue it in the absolute form we stated it, rather than with the nuance it actually needs. You see the same slogans used by every extremist group – for example, “Never again” in response to an immigrant crime, meaning we need to keep “them” out of “our” country for good.

A proportional response can be so much harder to defend, as well. A number of years ago in the UK there was a mountain climbing accident involving children. Such things are rare. Obviously we want to avoid them where possible, but the resulting legislation made it harder and more expensive to run summer camps that do outdoor activities in the UK (I was involved with one of these camps), and kids from poorer backgrounds especially could no longer afford those kinds of holidays. The unintended consequence is that children spend more of their summers in cities, where they are actually much more likely to get hit by a car or end up bored and doing drugs.

However, when faced with a campaign to tighten laws on outdoor activities, led by well-meaning and mourning parents, it is very difficult to say “Actually, no, we don't want to eliminate the possibility of children falling off mountains” (because to do so we'd have to either eliminate mountains, or introduce draconian laws that stop people from ever climbing them). There are in fact many really bad things that we shouldn't attempt to completely eliminate — slander, poverty, theft, murder, adultery, disease, the list is endless. But not wanting to eliminate these things seems pretty indefensible.

All of this leaves us with some big questions:

  1. If we express ourselves in absolutes, how do we avoid being called extremist and being misrepresented, or being lumped with people who use the same slogans for genuine extremism? Or do we just say ”haters gonna hate” and risk misrepresentation?

  2. If we express ourselves with nuance, how do we retain rhetorical punch?

  3. How do we avoid being taken in by our own rhetoric and taking things too far?

  4. Finally, if we are conscious of the elimination fallacy, how do we avoid using it as an excuse for not going far enough or defeatism?


Comments §

Comments should load when you scroll to here...