The Sunk Cost Fallacy | Hacker News Reader

The Sunk Cost Fallacy

thedecisionlab.com

73 pointszameermfm69 comments

It’s easy to see the irrationality of the sunken cost fallacy when all the inputs and outputs can be quantified exactly and completely.

Real life is often not as clear. We don’t know whether our costs are sunk or not.

Has the athlete who trained 7 years to get to elite status wasted her time, or is she on the brink of a series of pb’s that will rocket her into contention for gold, two months out from the olympics? It’s hard to tell.

Should Vincent Van Gogh have stopped painting and studied a trade or a profession that would at least pay a living wage? Should his brother Theo have kept supporting him? It depends on when you do the reckoning. In Van Gogh’s case, the Art was never worth anything during his whole lifetime.

How much did Steve Jobs blow on NeXT? On Pixar? On the Newton? It depends on when you do the reckoning, and it depends on what’s coming around the corner.

The only thing we know for sure is that quitters don’t win the long games.

The sunk cost fallacy only applies when you can see the likely outcome is no longer beneficial and continue anyway.

“ Has the athlete who trained 7 years to get to elite status wasted her time, or is she on the brink of a series of pb’s that will rocket her into contention for gold, two months out from the olympics? It’s hard to tell.”

This is not a sunk cost fallacy. You need to make a follow through even though you know its a loosing proposition. The training is a sunk cost, but the fallacy is making a decision based on that sunk cost.

Isn't the notion that you have to follow through a prime example of the sunk cost fallacy?

All costs are always ”sunk”. The fallacy is to think that just because you’ve spent a lot of money, time, or effort, that that in itself warrants that you continue.

In your athlete example, as long as she makes her decision based on risk and reward in the future, then that has nothing to do with sunk costs. If however she thought that her outlook as an athlete was bleak, but figured ”I’ve already dedicated seven years to this, might as well continue” - that would be the sunk cost fallacy.

One argument is that the total revenue of Concorde did not cover its total costs, therefore the program should have been cancelled.

But it's more subtle. One interpretation is that past costs don't matter, therefore decisions should be based on expected value of further investments. In a slightly round-about way, more "sunk costs" means higher leverage on future investments.

Whether it's a fallacy depends on high well you can estimate the future expected value, and how it depends on the current state.

Actually, understanding sunk costs may have saved the Concorde rather than killing it. I'm using entirely made up numbers just as an example -- let's say that halfway through development, new estimates of total project costs were $1 billion and total revenue was only projected to be $750 million. Shut it down? Nope, because the first half-billion has already been spent. At this point, the options are to stop and get zero, or spend the second half-billion to see a 750m return. So work continues.

> spend the second half-billion to see a 750m return

or spend the 2nd half-billion on another project that returns more than 750m. Decisions should be made holistically, and consider all opportunity costs.

Of course, all past decisions are sunk. But knowing this doesn't help you decide if they lead you into a situation where your best choice is continuing to make them because the payout is close now, or if they were a bad idea and the best is to abandon their payout and pursue something better.

The one thing that is certain is that those decisions have biased your options into the one they support. Deciding if that bias is strong enough is the hard step.

No. Cost can be in the future. That's the entire point of Cash Flow Analysis in finance and of planning in any other area.

I think the comment you are responding to means that past costs are always sunk.

Thanks! Yes, in this context, this is what I meant.

I've heard a good example for the sunk coast fallacy, i'll share (because i don't think you're right with your examples).

You see a great-looking one-star restaurant, quite pricey, but you want to try. You pay 100$ for your meal, and sadly in the middle of the meal, you start to feel sick. You really can't taste anything, and you know that the more you eat, the worst you feel.

Saying to yourself "hell, i've paid 100$ for this, i WILL enjoy it until i'm done" is the sunk cost fallacy. Finishing this meal will at most give you 300cals.

That's a clear example of the sunk cost fallacy. What explains it? Why do people persist in investing more to make themselves suffer more?

In the athlete and entrepreneur examples, the outcomes are not as clear or predictable as your one start restaurant example. Things might work out, or they might not. It's impossible to tell.

Almost every athlete, artist, entrepreneur and warrior has dark moments in which they feel doomed to failure. Some of them quit, and we never hear about them. Others keep going regardless. For a few of them that make it through to the other side, the reward is a success that could never have happened without blind faith.

The blind faith is like a flywheel that powers the venture through uncertainty. It's a powerful asset that not everybody has enough of. But it's also a double-edged sword that can result in the sunken costs fallacy.

In the case of many ventures, the only people who can tell the difference between a sunken cost fallacy and a heroic tale of persistence are the historians.

Your comment makes me think of Theranos. Imagine if they could string us along long enough and meanwhile the researchers there figured out some breakthrough. Would we see the con artist as a visionary?

Famously, first three rockets made by SpaceX failed and the one that got into the orbit first was literally the last one they had available funds for. Musk would have been considered entirely rational if he pulled the plug after the third crash, thus saving himself some millions.

Maybe a certain level of irrationality is necessary to push things beyond usual limits, even though that effort often fails.

That example seems more like Musk put a very, very high value on entering orbit. It's not irrational to continue trying to put rockets into orbit if you value that accomplishment more than millions of dollars.

Why? After three failed rockets they knew much more about building rockets, and they could with much more confidence assess their probability of success, so at that point it was more rational to spend money that it was before the three failures.

This thread is confusing to me. I thought everyone knew about sunk-cost fallacy, but all I keep seeing is people referring to "sunk-cost fallacy" like it's the antonym for "persistence" or something.

Sunk cost fallacy isn't saying: don't be persistent in chasing goals, or give up early. It's specifically saying the costs to get you to your current point are already lost and should not factor into whether you chose to continue forward or not. Only the potential future cost and your estimate of likelihood of future success should. Not "how much you're already invested".

Is that even a realistic frame of mind for human beings? I think previous experience will weigh on you, and losing money is a painful emotional experience.

Yes. And I think the world works that way sometimes.

The only thing we know for "sure is that quitters don’t win the long games."

Can you cite some evidence proving this statement?

It's a tautology: a quitting a game means you don't win it. It works in the specific game context.

But you may as well quit one game to join another which has more value for you.

Depending on the viewer/commenter's perspective, you quit, you join, you switch gears or just whatever.

... and those who don't win and don't quit are ...

ignored. (Original saying ends with "idiots".

> Vincent Van Gogh have stopped painting and studied a trade or a profession that would at least pay a living wage?

He died poor and unhappy. He became famous only after suicide and all the mental health crap came out.

So probably yeah, his life as was sucked.

Often times there is tremendous uncertainty around how much of the cost is actually sunk, and whether there is some liquidation/going concern value that can extracted

>Real life is often not as clear. We don’t know whether our costs are sunk or not.

I'm having trouble thinking of when it's not clear.

>Has the athlete who trained 7 years to get to elite status wasted her time, or is she on the brink of a series of pb’s that will rocket her into contention for gold, two months out from the olympics? It’s hard to tell.

>Should Vincent Van Gogh have stopped painting and studied a trade or a profession that would at least pay a living wage? Should his brother Theo have kept supporting him? It depends on when you do the reckoning. In Van Gogh’s case, the Art was never worth anything during his whole lifetime.

How they spent their lives so far was a sunk cost. That's clear. However, the skills they gained from those years should influence a rational choice. The years themselves are irrelevant. Even if they were perfectly rational, the athletes of today are likely to be athletes tomorrow, because it's easier for them to be an athlete tomorrow than an accountant.

The number of years they've sunk training should only come into play if it helps estimate how much further to their goals. Which means they should follow the opposite of the sunk cost fallacy: the less they've sunk so far, the better. Take your example athlete. If she had instead trained for 14 years to reach that same level of skill, even at the same age, shouldn't she see the Olympics as a less likely goal? It took twice as long to get where she is, so chances are it would also take longer to reach Olympic level.

> I'm having trouble thinking of when it's not clear.

I phrased that badly.

Past costs are always sunk. What's often not clear is whether the costs have yielded a positive return or a loss.

If you bought bitcoin at $50K, and the price today is $30K, your investment has yielded a loss.

If you bought Tesla at $100 and the price today is $700, your investment has yielded a profit.

If you've trained 7 years for the 100m freestyle, and today you're ranked fifth in the world, your training investment has gotten you to fifth in the world. Beyond that, it's difficult to assess the value.

What are your aims? What are your prospects?

Let's say your aim is to win an Olympic medal.

Should you train for another year in your quest to win an Olympic medal?

If you train for another year, will you win an Olympic medal?

The only certainty is that if you stop training, you won't win anything, and your investment has been in vain. Your costs are sunk with no return.

It's that certainty of failure by quitting, vs the uncertain but non-zero prospects of winning if you carry on and invest more, that nudges you towards the possibility of sunken costs fallacy i.e. throwing good money or time after bad, even past the point of no return.

20 years in Afghanistan - screaming out example.

"We can't leave because of troops who died would have died for nothing". That's precisely Sunk Cost Fallacy!! They are already dead and nothing you do can change that - thus it's not relevant to FUTURE or PRESENT decisions. Only future prospects based on the current situation matter, namely, are the prospects for eliminating terrorism or inculcating democracy into the future good NOW? No? Then cut it off to avoid Sunk Cost Fallacy.

Most people realize that was necessary but again Sunk Cost Fallacy seems to have affected the withdrawal: only the past was consider but not the future costs of leaving "badly".

Afghanistan seems to illustrate both continuing to invest too long (20 years) and then once the decision was made (by Obama) to continue investing enough (some say too much) to prop up a democratic Afghan regime, and then not investing enough to ensure a smooth withdrawal after the regime started to collapse.

The sunk cost fallacy is also where we get tremendous motivation. A story of a Navy SEAL who spent months trying to find a doctor to get an expensive and dangerous heart surgery so he could qualify for SEAL training comes to mind.

It's also worth noting that sunk cost is often strongly correlated with remaining cost. If you've spent 4 years in a 5 year PhD program and start questioning your reasons for being there, you could frame it in terms of the sunk cost fallacy: "just because I've been here for four years, doesn't mean that I should continue on". On the other hand, there's a rational argument for staying: "at this point I only have one more year to go and if I leave now I leave empty handed but if I just stay one more year, I'll get a PhD".

If I understand correctly the sunk cost fallacy your example lacks an objectively better alternative, like an opportunity related to one of your hobbies, unrelated to your studies, which could land you in a great job in some years. As you stated it, there's no doubt, staying in the PhD is the only rational choice.

There are many, many corporations that rely on this (Many “cost plus” government contractors, for example).

A common pattern, is a deliberate underbid, in order to lock in the contract, then a series of additional charges, as the project progresses. A contractor tried this recently, on a friend of mine. Luckily, he consulted me, and I was able to help him to steer clear of that tar pit.

The 419 scam works the same way.

Cerner and the VA

That's not necessarily a sunk cost fallacy (at least as I understand it). If I add additional charges, the smart thing to do isn't necessarily to fire that contractor. This is because firing them, then hiring a new contractor to finish the job, is possibly more expensive than paying the charges. Asking for more bids, choosing a new contractor, then getting them up to speed can be costly.

The rational thing to do would be to weigh your options at that point. Do I believe that a new contractor can get up to speed and finish the project with a smaller budget than the current team?

Lots of very different strawman situations are being described. Sure if you've built expertise and experience in an area of effort (athlete; scholar) then you have to balance that against the cost of starting over somewhere else.

But if you have only spent time and money, and have nothing to show for it? Then 'sunk cost' is indeed a fallacy.

Nothing to see here folks; move along.

Some of the examples here strike me as wrong.

For example: being forced between a $100 trip and a $50 trip, both of which are prepaid, if both can't be done because they are on same day. It's not a fallacy at all to choose the $100 one - even if we are not sure in which way exactly, we are still living in a free market so if the cost was 2x without any beneficial differences, it wouldn't sustain itself - so that trip is likely to be better in some way.

Not dropping out of a paid program with a lower success rate in favour of a free one with a higher success rate is a good thing: it very likely means that a free one is a cheap, bs thing that gives no real knowledge (certainly they won't teach you better for free, so if the pass rate is higher it only means that exams are simpler to please everyone). Plus, some time is already lost on the paid program, and time is money.

The ski trip one does seem bogus. The person obviously wanted to go on both trips since they bought tickets for both. Unless something has changed and they no longer want to go on the $100 trip in the future, then it makes sense to go on that one now to reduce future costs (spend $100 in future vs spend $50 on other trip in future).

Whether they enjoy the $50 trip more than the $100 one, or think they may do, is irrelevant if they still want to experience both (as one can assume by their having bought tickets for both).

Honest question - in the concert ticket example, the article said that “ $50 spent on concert tickets..should not be a factor in our current decision-making” (because it’s irrecoverable).

But what if you are still interested in this show? If you choose not to go today, then you would need to purchase another $50 ticket in the future, thus increasing your total investment to $100. But if you go today, despite being sick, your investment would only be $50. So the $50 already spent could be a logical input into your future decision making…assuming you still want to go, and are evaluating a second purchase or not

Is this still sunk cost fallacy?

The fallacy is going because of the resources (whatever they may be) spent, not considering (or fairly weighting) other considerations. If you think the concert is going to be sufficiently enjoyable (or unique, last concert by your favorite band) to go to despite being sick, having to deal with frustrating traffic, and likely getting sicker because of going, then go. But don't go because you spent $50 just to end up laid up in bed for a week after and never enjoying the concert because your sinuses were so stuffed up you couldn't even hear it.

It's important for a good analogy to have these bases covered.

It's not an analogy, it's an example.

And everything I listed in my comment came from the article:

> On the day of the concert, you feel sick and it’s raining outside. You know that traffic will be worse because of the rain and that you risk getting sicker by going to the concert.

So they did cover those elements. The point of the article is to discuss the sunk cost fallacy, which they did. The fallacy is choosing to go to the concert because you spent the $50 on the ticket, disregarding the other considerations or weighing them below the sunk cost in importance. If you go despite the issues they mention but because you think it's going to be worth it (in enjoyment, uniqueness, whatever) then you've made a decision that wasn't based on the fallacy. Which is what they discuss in the article.

In your question, it sounds like you only want to see this concert live once, so your choice is between seeing it now “for free,” but sick, or later for $50. That is not a sunk cost fallacy, since the current marginal cost is lower to see them now.

But there are many factors that can be a sunk cost fallacy too. If you hope to go again anyways, for example, then that future interest is not a factor now. Or if you pay more than $50 extra (or refuse to pay less than $50 extra) to get to the show tonight. Since if you incur a new marginal cost by being sick, you must consider that cost in relation to the expected future cost of the show, not the money already invested in it.

So to avoid the sunk cost fallacy, you should always be willing to pay an extra $50 (in expected total), even after spending multiples of that (unexpectedly), and then still not go if someone would charge an extra $51 unexpectedly.

I had this decision in 2007 to see Daft Punk or take care of a tree that had fell at my home. In hind-sight I'd much rather have gone to the concert. This analogy is terrible.

>The Sunk Cost Fallacy

Every company that has ever implemented SAP.

There's also a positive framing for the sunk cost fallacy.

Invest your money in something (e.g. exercise equipment) and you'll be more willing to spend time using it.

The ultimate expression of sunk cost fallacy is Bitcoin (and Proof-of-Work in general).

“We burned so much coal generating random numbers, it has to be worth a lot. Let’s spend even more.”

Aliens visiting Earth would have a great laugh at our sense of priorities.

The sunk cost fallacy fallacy: we have perfect knowledge, particularly of domains we haven't poured a lot of time into, and therefore can, with certainty, act perfectly rational at all times.

Sunk cost fallacy is economics 101, yet most people I know (including business-savy people) have not seem to have heard about it or fail to recognize it when it affects their decisions.

How things should be done:

Value (to the customer) > Price (sale price) > Cost (to the seller).

What's needed is value (which is subjective), not necessarily cost.

Gwern had the ultimate take down of the sunk cost fallacy. https://www.gwern.net/Sunk-cost Outside of specific (usually government) projects, I tend to think of it as the sunk cost fallacy fallacy.

I find the example in the article a bit strange regarding the concert.

Wouldn't going to the concert in a sense "recover" the $50 if you valued the experience of going to the concert at $50 or more?

They provide several other articles for additional context, such as the Concorde.

No, you don't get the money back. In theory you should act as if you got the tickets for free. You are going because you find the concert interesting, you do not go because you have already spend $50.

Also you take into account future costs. If you want to go to the concert and there are several days that you can go, you choose to go to the one that you already have tickets. Attending any other day will cost you $50.

Another cost is the social consequences. Maybe you do not want to go anymore, but you know that your mother will be pissed off after she gave you the money. Or you may be judged as irresponsible by your friends if you don't go.

That you paid in the past do not matter anymore. But are there future consequences what you need to worry about.

From the article: “In economic terms, sunk costs are costs that have already been incurred and cannot be recovered.”

It wouldn’t be a sunk cost if if going to the concert was still valued at $50 or more.

This analogy is so very out of date. I've been able to sell my tickets for face value on the day of the event, using the 'net for over ten years now. And I could have made a killing if I really wanted to.

That example is one I have seen in nearly every article I have read on the sunk cost fallacy, and every time a thought occurs: "can anyone come up with a better one?" If I am truly sick on the day of the concert, I call my friends and give them the ticket. It seems trivial.

The best example i know is related, it that you feel sick in the middle of a meal at a one-star restaurant (let's say on the third dish).

That's a decent example, but I don't see why one-star vs five-star would make any difference. It's also depends if continuing to sink time/money into the meal (by having desert, say) would make you feel more sick or not. Unless you feel like vomiting, maybe you want to stay for that tasty desert anyways!

You'd only be committing the "sunken cost" mistake if you continued with the meal without having some good forwards-looking (social, perhaps) reason to do so.

It does seem a poor example, but I guess the intended point is that the logical decision should be "will I more enjoy going, or not going?" rather than taking the already "sunk" money into account. Of course in the real world, especially if the money was a significant amount to you, your enjoyment of staying home might genuinely be tainted by the wasted money, and that's just as legitimate a factor as anything else figuring into your "based on where I'm at now" assessment.

I think a better example, maybe appropriate to this forum, might be where you've been spending significant spare time working on a project with dreams of future riches, and it eventually dawns on you the payoff isn't going to happen (maybe the opportunity has passed, or whatever - reason doesn't really matter). The question is do you now continue working on the project to "at least get the satisfaction of completing it, given how much time has been sunk", or do you cut your losses and abandon it. Even this isn't a "pure" example since there's no reason not to continue if the modified goal (satisfaction) is worth the remaining cost to you.

It's probably the case that in most, maybe all, "sunken cost" scenarios there is some benefit to continuing to invest in it, but the "sunken cost" may distort your judgement of assessing that benefit. Even with Concorde, there was considerable cachet and advertising benefit in operating Concorde, even if it wasn't paying for itself operationally.

I've always been able to sell my tickets at the last minute and I always have sold at cost even though I could have sold for more. This analogy is bad. There is such a thing as a sunk cost fallacy but this concert example is poor.

If the cost to complete a project divided by its expected annual revenue is more than the P/E ratio of your stock, you're in a sunk-cost fallacy.

"We have so many lines of COBOL, we have to keep investing in COBOL"

The people in that situation go to concert, because they really really want to see concert and have fun. Most concerts happen only once then you have to wait a year for another chance.

Or once in a lifetime in some cases.

The Sunk Cost Fallacy is why we are still doing the lockdowns.

Can you elaborate?

Their grotesque implication is that older and vulnerable people who could die of Covid is a sunk cost?

You know after the first 2 vaccines didn't stop COVID. I'm sure a third one will do it. Keep destroying businesses and lives for the next decade, because you've already travelled so far down this path. Zero self reflection. Young people committing suicide due to the lockdowns don't matter.

Reply on news.ycombinator.com