Hacker News Reader

Comment by paulsutter | Hacker News Reader

When faced with a difficult decision, it probably doesn't matter which choice you take. Emotionally it's difficult to accept this. But logically it's easy:

- If one of the two choices was clearly better, it would be an easy decision

- But it is difficult. Therefore, neither of the two choices is clearly better.

- Therefore, it doesn't matter much which you choose.

This doesn't work for people who feel irrational regret (ie, regretting a decision that turned out badly because of information that was not known at the time of the decision). Yes, irrational regret is very common. But it's just another logical failing that can be overcome.

If you still feel anxiety when forced to choose between two good choices, go visit a third world country.

View on HN

Well your reasoning is a bit of a shortcut, and Ruth Chang[0] goes all the way down the rabbit hole, without having to resort to third world country fallacies.

[0]: http://www.ted.com/talks/ruth_chang_how_to_make_hard_choices

Thanks for this. The gist of the talk is that each hard choice gives us an opportunity to shape who we are and decide who we want to be. Hence we should look at hard choices that way and not shirk away from them.

This type of argument does not explain how to solve hard decisions, it simply says that if you know impact of each choice is similar, it is a simple decision.

There are endless ways for a decision to actually be hard. One is where the type of impact between options is different and difficult to compare. The difficulty may arise due to: uncertainty, complexity, lack of knowledge, value judgements, brutal trade-offs etc.

i) "Should I stay to treat to the ebola victim or leave with my family?" is hard because it involves value judgements, predictions, risks. It questions who you are. ii) "Should our startup be tightly integrated with facebook?" is hard because the potential impact is massive and requires substantial investigation to understand.

The very attempt to compare options can involve costs and feedback loops e.g. should I pay for an investigation, should I prototype this, should I wait for more info etc. The actions you take to compare options can change the available options. A single decision becomes a nested web of other decisions.

In the most general sense, making a decision is like devising a strategy. Is creating a good strategy easy or hard?

p.s. I'm working on this problem

Can you provide any more information on what you are working on? I, too, am interested in this problem, so it would be great to see any attempts at a solution.

Email is in my profile, if you'd rather not share here.

> "neither of the two choices is clearly better"

"Clarity" is dependent on what level of information you already have and what analysis you've already done.

In some circumstances, you've already done whatever investigation, research, and thinking can reasonably be done. If the choice is still hard, then the choice is likely to be of similar value either way and you may as well choose arbitrarily.

But in other circumstances, there's still investigation, research, or thinking that can be done (at a level of effort which is reasonable based on the importance of the decision.) In that case, one choice or the other may become clearly better if you are patient and diligent in the decision-making process. So be as patient as the decision warrants.

Yeah, I don't think paulsutter's post holds up to much scrutiny, except in the limited case where you have to make a decision immediately, or in the near future before you have time to do any meaningful research or calculation on the problem.

A problem is difficult when its potential consequences are serious and the best option is not obvious. But the best option can be non-obvious for two very different reasons: either it requires a lot of information and thus a lot of work to gather that information (like choosing a medical treatment), or it depends on randomness or other variables outside your control and understanding (like choose your lottery numbers, assuming someone gives you a lottery ticket).

For the latter case, it's pretty clear that you can choose randomly and shouldn't worry about your decision. For the former case, it's reasonable to "worry" in the sense that you should feel an urgency to gather the required information efficiently. Of course, it's not really that simple, because you have to optimize the cost of gathering information against the potential cost or gain from the decision itself. Also, the two reasons from the last paragraph are more like a continuum, because "randomness" is often effective randomness, i.e. you can't feasibly predict the outcomes despite them being deterministic.

If the research has a cost (and you know this cost), then you actually have three options: original option A, original option B, and research + correct option (or knowledge that both options are the same). If you can at least bound the costs of options A and B, you can make a pretty good guess as to whether the research is worth doing. If it's not, then choosing at random is the optimal choice, and you should not regret it.

Often when I am faced with two choices that I don't like I look up alternatives.

It's difficult or impossible to tell if you made the right choice though, so you have no way of knowing if you should regret your decision. If I visit the store and there are two boxes of cereal, or two deodorants, or two digital cameras, and I buy one of them, I start to feel unsure.

It doesn't matter if the cereal I bought tastes good or bad, I'll always wonder if the other option would have been better. Until I taste the other cereal, I'm left questioning my decision, and being unhappy.

The answer is to stop caring and satisfice. Optimizers are never happy.

I don't know if it's that simple, it's only human to question things in life, and it's one of the main reasons we've come so far in the first place.

Being unhappy and unsatisfied is the main reason we have the technology that currently exists. If people were satisfied traveling by horse, the automobile wouldn't exist. If people were satisfied with 480p, the HD television would never have been a worthwhile venture. Being continually unsatisfied is what causes us to strive higher.

If people were satisfied with horses and 480p, we'd be just as happy with those as we are with cars and 1080p. In fact, these decades and centuries of technological advancement hasn't, empirically, resulted in any enhancement in happiness anyway, because we're trapped on the hedonic treadmill: http://en.wikipedia.org/wiki/Hedonic_treadmill

I suppose I can reluctantly agree with that, although self-reported happiness surveys and all other attempts at interpersonal utility comparison are problematic. But even if I agree with that, what's the conclusion? Surely the people with horses become less happy when cars come into existence, at least until they upgrade to a car. So unless you can ban or otherwise prevent any new technology from being created, I don't see what meaningful conclusion or recommendation can be made from this claim.

Yes, but we wouldn't have cars or 1080p in the first place. Why would you develop 1080p technology if you were perfectly happy with 480p? Who would support your research and business by purchasing a 1080p tv, if they were already satisfied?

I never said any of this technology enhances our level of happiness. I just said as a species, being constantly unhappy could be one of our greatest strengths. Otherwise, we might be sitting in a cave, smiling at each other until an asteroid removed us from existence.

Who would support your research and business by purchasing a 1080p tv, if they were already satisfied?

Regular people may have been satisfied, but the military wasn't, who drove almost all of the research into the 80s. The consumer market is just a follow-on to that, where choices were presented to "me" as better.

I think you should read philwelch's post again. You seem to see development as an intrisic good thing, whereas philwelch does not (or at least questions it).

There is a world of difference between freezing in a cave and developing 1080p/2160p/4320p.

I think people living in the modern world, with all of our technology are no more happy than a caveman sitting by a fire with his family.

However, to survive as a species in the long term, we need to leave this planet. To do that, we need a number of advancements in technology.

So, if we can agree the survival of the species is in our benefit, then we can agree development is considered a good thing, even if it doesn't equate to an increase in happiness for the individual.

Except for medical technology : without it, the cave-man would have far fewer years of expected life-span to enjoy sitting by the fire with his family. And many of those years would be filled with ill-health, etc.

What's the damn point of surviving as a species on other planets? Because at that point, we'd be a different species anyway.

I've seen the "satisfice rather than optimize" meme a lot recently. I think it's a pretty weak argument. The problem is that there is still a potentially massive amount of calculations and predictions (about costs, opportunity costs, trends, etc.) required to choose a reasonable acceptability threshold. If you take more time to gather this relevant information you can probably choose a better acceptability threshold. The meta-analysis is still an optimization problem with an additional variable to optimize over.

It's not one argument. In its simplest form, take the digital camera megapixel race. Who in their right mind does anything useful with 20 megapixels, where most are buried in the noise levels for all common applications? Does that really need to be "optimized" further, even if it means quality and storage both suffers? A lot of real problems worked on in the industry could be described like that.

But if the satisficer's acceptability threshold is 25 megapixels, they're still not happy, and if their acceptability threshold is 0.5 megapixels, they'll be using a very noticeably poor camera.

Opportunity cost only exists if you're an optimizer.