Hacker News Reader

Comment by tantalor | Hacker News Reader

I don't think this is a paradox at all.

If an option is perceived to be significantly better, then the decision can be made quickly.

Otherwise more study is required to decide which is better, or if they are actually indistinguishable.

It's the same reason why more trials of an experiment are required to reject the null hypothesis when the effect of the experimental variable is very weak. This is analogous to considering the options in your mind until you decide the difference between them is too small to matter. That takes time!

Sometimes after deep thought, you may decide one of them is actually better, but not by very much. Obviously you could argue that time is wasted for such a small benefit, but it's not a paradox.

Introductory statistics tells us we can never prove the null hypothesis; all we can do is reject or fail to reject it. In other words, you give up on the attempt only after deciding extended study is unlikely to reject it.

View on HN

It is not a paradox but it can be irrational. When the two choices have similar utility the cost of distinguishing them is high and the benefit of making the correct choice is low. There is a crossover point where it does not matter which option you choose, putting more effort into your decision can only leave you worse off, and this article claims we tend to continue putting in the effort well past that point.

Sure, but how do you know a given decision is one of those decisions?

For me it should be intuitive. My test is "Will I accept either outcome if I flip a coin?". Is so, flip the coin. If not, pick the option that prevents me from flipping a coin.

But the original question remains, how do you know that until you have analyzed it?

For certain types of decisions, one can safely rely on prior experience. I know to a high degree of certainty that most entrees at a moderate to high quality restaurant have roughly the same value, therefore choosing quickly is best because there's little to discover and little to gain from the discovery. Many aesthetic choices or choices of mere preference are like this. Other classes of issues are known in advance to be both trickier and yield real, effective outcome differences, these are often big life decisions that act as life's value centers, e.g., which house, which partner, which employer, etc.

It's of course possible to be wrong in either direction -- this difference matters though i think it doesn't and vice versa -- but in general, we tend to be pretty good about taking seriously the things that really are serious, but pretty bad at not wasting time on things that don't matter much. This shows up in the paradox of choice, but really just imagine someone hemming and hawing over two brands of cereal in the aisle at Trader Joe's and you get the picture.

So the practical problem isn't really "how do you know until you analyze," because we more often fall into opposite the trap and waste time systematically attempting to optimize choices where there is little value to be gained. We should step back in more cases and remind ourselves that most differences make no difference at all. Save your time and energy for the ones that do.

A similar problem occurs on the big questions too, like when choosing an employer or university. On big questions often times much of the differences are unknowable.

Big decisions are often like icebergs. Most of the important stuff is hidden below the surface. No matter how much you analyze the tip you still have an essentially 50% chance of getting the decision wrong. It's really hard to make a major decision knowing that you have a 50% chance of getting the decision wrong, but over-analyzing the tip of the iceberg is not going to improve your decision making.

I think the process of making good decisions about "big" things should involve practicing making many more decisions about smaller things and evaluating the outcome.

There are a lot of insightful heuristics from decision making that can make big decisions much better/easier/etc.

I don't think people would spend any more time on big decisions if they made faster/fewer decisions with little "value" to gain, if anything I think their habit of making fast decisions would mean they spend less time/worse on the big ones!

For sure, sharpening analysis chops is a good reason, but in a world of finite time, and an overabundance of important decisions to make regarding one's life, there's no shortage of analysis fodder to begin with.

And no doubt that sometimes it's just nice and pleasurable and interesting to deliberate over anything at all, making small things grand, but it's also good to know when it's actually optional.

Flip a coin and it will make your decision for you.

Personally, I like the strategy of flipping a coin as soon as the preference is not obvious, and then picking the other one if you realize you don’t like the coin flip.

I love this approach, it's effectively a forcing mechanism for revealing preference.

The issue is it is hard to know whether the two (or more) choices are equally good.

If you have no way of knowing which choice is better then you would flip the coin of course. The problem is there typically is a way to possibly learn more about which is better, AND HOW MUCH, by studying the choices. The "How much" -part is crucial. One choice might lead to core meltdown.

Often such study does not provide any further information about which is better. But it might. You just don't know. Except if you are lucky and the answer is immediately obvious.

So if it looks like you have two choices A and B, it is more typically the case that you in fact have THREE choices:

1. Choose A

2. Choose B

3. Spend more time studying which is better, and how much, A or B

My algorithmic approach:

You spend a fixed amount of time, if no choice could be made during this period, then flip the coin.

The fixed amount of time could be as short as five seconds. Or the amount that makes sense to you, and you feel comfortable with. If it's one week for you, then it is.

>The issue is it is hard to know whether the two (or more) choices are equally good.

If the choice doesn't matter much, then both options are equally good.

E.g. "should I get this pair of shoes or that? I like both".

Just get one and get on with it.

"Should I go to Paris or Vienna for a short trip?"

Just pick one and get on with it.

Still people obsess for these kind of choices for needlessly long...

The case discussed, were people waste time they shouldn't, is (a) when the choice doesn't matter much or (b) when you do know that either option would be more or less fine (which is basically a variant of (a)).

If (c) you can't tell if they're equally good AND chosing the right one does have a BIG impact, then this case doesn't fall under the advice of TFA or what we discuss here.

Rather, it's an "ipportant choice" and people should think it over...

Yes of course if you "know" that the choice doesn't matter and you still obsess with it then it is your fault. I don't know how much that happens in practice, and I wouldn't blame people much for doing that because maybe they enjoy doing it.

Based on my own experience as a SW developer I have found that more often it is hard to know beforehand whether a design-choice matters much or not, and how much.

I suppose that is also the reason for much of "Technical Debt". People make less than optimal design choices not because they are bad designers, but because trying to evaluate the benefits of each choice would be too time-consuming.

A friend was spending forever trying to decide which of two toothpastes to buy at Target. The cost of avoiding calculating which is better by just buying both was only $3. In cases like that I think it's irrational to spend so much time thinking about it.

A grandmaster logician descends from logic heaven to tell you it was actually rational and asks you to figure out why that it so. What is your response?

It might be one of these:

A) the benefits are higher than you think: e.g. little did you know it but they have an unusual mouth microbiome which means some toothpastes will work and the rest will turn them into a wimpering ball of cavities within the week. They don't quite remember /which/ ones though and are trying to figure that out.

B) the costs are lower than you think: this friend is funemployed and has nothing else they would rather be doing with their time, in fact they find a great deal of satisfaction from knowing they have "optimized" every area of their life by spending more time on their decisions.

Price is not only money for the purchase. Aferwards, it increases the effort of cleaning teeth, on the left using one, and on the right using the other, over the next 5 years, before making a final choice based on results at the dentist.

At which point both of them will have been upgraded with different ingredients.

> this article claims we tend to continue putting in the effort well past that point.

it's because people have the irrational belief that their thinking and reasoning more about a problem/decision makes their outcome better.

I think the common symptom found in software is design analysis paralysis. A software team end up using tonnes of time fretting about design choices (often as a committee), rather than just use that time to implement one of the designs. The (ir)rational idea that by reasoning about the design up-front, one saves the (potential) time lost in reimplementing a different design in the future!!

> it's because people have the irrational belief that their thinking and reasoning more about a problem/decision makes their outcome better.

i'm not sure this is correct. i bet it often comes down to whether we've found an [emotionally] satisfying answer/outcome.

Definitely trying out different alternative designs is a good idea. But it has its cost as well. But you are right it is usually better to try out something than just ponder all the possibilities.

This! Often (actually almost always) time it self is valuable for multiple reasons. Spending precious time in the "meeting room" instead of progressing towards (almost any) implementation can decrease the overall value and increase the risks.

Also some people strongly prefer action over deliberation, especially by a committee so there is also a psychological cost to the team to consider.

With software and a capable team, there can still be an illusion that decisions are cast in stone and can't be changed later, while the opposite is true.

Implementing a solution that is almost there and then making corrections can be both more satisfying and lead to very good results. This of course requires the psychological freedom an flexibility to quickly acknowledge errors in thinking and design and not holding on to "the plan" too tightly.

I’ve argued for just building POC approach at times and worked out great sometimes. I found the issue was usually with management. Many have been extremely reluctant to throwing away code. Even though writing POC and re-doing everything will take the same time as planning for ages and doing it once.

Proof Of Concept.

It can be worth agonizing over the decisions early on while the system is simple and easy to change. Later on it's often not logistically, politically or socially justifiable to go back and fix or improve an original bad design decision and re-base and retest everything built on it.

The result is engineers sat around muttering "I told you so" to themselves and wasting weeks debugging and hacking around bugs.

Sure. Now we need to know when it can be worth. If we do it all the time we're back to the starting point of this discussion.

I find that the agonizing rarely bears as much fruit as doing.

In some cases we've deployed, found horrible problems that were completely unexpected and switched course.

Additional time spent agonizing would just have delayed getting us to the point where we learned those things. The critical information we needed to make that decision wasnt going to reveal itself with additional introspection.

This isnt to day when a dev waves a big red flag in the beginning you should ignore them, though (ever). But, I find that waving that red flag doesnt ever inhibit delivery speed whereas multiple people agonizing over decisions where the information to determine the "right" one usually does.

While it may be easy to label it as irrational post hoc, it might not be clear beforehand without the expended effort in evaluation. There is no paradox, just an information asymmetry in time.

I reject the hypothesis that this is a matter of irrationality. The error in this line of thinking lies with the fact that you have implicitly equated "have similar utility" with "have similar visible utility".

The reason "obvious" choices are "easy" is not because of the difference in actual utility, but because of the difference in "obviousness". I.e. it relates your observation to the probability that it does indeed relate to a difference in utility of not.

Therefore, two options with "similar visible utility" require more effort discerning whether the "actual" utility is also small, or you're about to make a very big mistake based on superficial notions of "no visible difference".

I content that given knowledge of actual equivalence, the paradox goes away and choices become trivial again.

A good example is phone contracts (at least in the UK). You'll spend far more time weighing options if they're bundled with all sorts of dark pattern magic like "first 3 months free then £X plus promo vouchers in select stores", compared to when you only have the annual price to compare between two choices. But it wouldn't matter if that price difference is £100 or £1.

With the caveat that I'm just an internet commentator that doesn't know what I'm talking about.

> Sometimes after deep thought, you may decide one of them is actually better, but not by very much.

I feel like the value here is simply not for the decision at hand. It's that you learn something for future decisions.

When the options are very close, there's not much value to be gained for this decision, but it indicates there is something new to learn. When the options are obviously far apart, there's nothing to learn.

I've thought about this on and off in the context of poker. When you're playing 24 tables at the same time you simply can't waste focus in the moment on the close calls unless the pot is big. Instead, analyze the things you think are close calls offline.

Reply on news.ycombinator.com