It's unlikely that the (alcoholic, teetotaller, alcohol) relation and the (desirer of power, indifferentist to power, power) relation are identical, for the teetotaller is averse to alcohol where we want someone who is indifferent to power. Besides, the object of the alcoholic’s desire is not obvious. It seems reasonable to ask whether the alcoholic desires alcohol in any serious way at all, or whether he desires the state of mind that comes from drinking alcohol, or even the state of mind that comes from drinking too much alcohol, or something else altogether. But let that pass.
I think that the counterexample you present is compelling only if one holds that the alcoholic acts as he does because he desires alcohol inordinately. If he acts as he does for some non-desire reason, then my premiss is unharmed. So the model of addiction (or weakness of will – call it akrasia hereafter) you’re working with is this: an addict (or akratic) is one whose will is defeated by the strength of his desire.
I’ll argue that that is an implausible understanding of addiction: the pathology is located elsewhere, so your analogy fails, and with it, your argument. Actually, I think it’s likelier that the alcoholic is not an alcoholic because of an excessive desire for alcohol, but because of a deficiency or pathology of the will – addiction is rather a defect of the will, than of the appetite. Maybe this is why addict and weak-willed are so close in meaning.
The obvious way to distinguish the desires associated with addiction from those that aren’t is by their strength. But that simply won’t work. We have many very strong desires which are non-addictive (any normal person will have strong and persistent desires for food etc.), and weak desires that are addictive. (An alcoholic might very strongly desire to give up alcohol and only very weakly desire to continue drinking, yet continue to be an alcoholic). Indeed, one can become addicted to the object of almost any desire.
Again, suppose that addiction is the defeat of one’s will by some desire D due to the strength of D. If that were true, then one would expect other desires of the same strength as D to defeat the will equally regularly. But, typically, the addict is addicted to some things but not to others, even when his desire for them is just as strong as D.
But if there’s no reliable way to sort out desires associated with addiction from desires not associated with addiction, then it’s unlikely that addiction can be explained in terms of desire. If so, then the analogy you want won’t run, because the problem (I suggest) shifts to the will. For your criticism to stick, you need a pathology of desire to be the cause of addiction or akrasia. But there are reasons to deny that, and so to deny your argument.Now to your other point:
I wrote ["Suppose it is true that those who do not desire power are likeliest to be responsible in its use. So we need a method of choosing a ruler which maximises the chance of the ruler being indifferent to power. The most reasonable way of doing so is to choose a ruler at random from the class of all adult members of the population"]
You replied:All things being equal, indeed so. But there are other matters that weigh in favour of monarchy, not least the distinct possibility that the monarch has been trained from an early age in the responsible use of it.
As we saw earlier, you need some additional premiss to justify confining the eligibility class to those descended from some arbitrarily chosen person. That the monarch needs training from an early age is insufficient to justify that move, even if were true. Couldn’t one be trained just as well in the use of power by wise non-relatives?