The Sunk Thought Fallacy
This inaugural post asks whether we should expense or capitalize the time we spend thinking.
I intend this newsletter to adhere to a few policies.
Posts should be:
Short.
Mostly about my professional interests: investing and entrepreneurship, especially within the SMB segment.
Free of self-indulgent digressions, metacommentary, and references to dead writers.
That being said, this post is:
Several thousand words long including footnotes.
Mostly about my personal preoccupations: rationalization and regret, especially as relates to decisions about where we spend our time and attention.
Full of self-indulgent digressions, metacommentary, references to dead writers, and showy parallelisms.
It’s also an attempt to give readers a sense of what's to follow. Let’s start with the title. What is a “sunk thought,” aside from a string of letters corresponding to a URL that was available when I was setting up this site?
Non, je ne regrette rien
Well, it’s also a bad pun based on a very tenuous homophone. A sunk cost is one that has already been incurred and can’t be recovered. Sometimes we pretend otherwise. This is the sunk cost fallacy.
To illustrate the concept, economics textbooks are required to provide an example about some poor soul, often named Jane, who falls ill the night of a concert and chooses to suffer through the show rather than “waste” the tickets she’s already bought. Sunk costs belong to the past. Ignore them, we’re told.
Easier said than done. Concert tickets are expensive. Moreover, the imagined version of the future they grant access to - what will surely be the best performance in music history, some combination of Queen at Live Aid mixed with James Brown at the Apollo - is as hard to let go of as it is easy to keep telling ourselves might yet be attainable.
The textbook example posits that Jane is certain she would enjoy the concert less than the alternative of staying home. She’s plotted her own utility function with 100% accuracy. But there’s a difference between knowing what you want and knowing what to do about it.
In real life Jane drags herself to the concert, because to stay in bed would be to admit defeat - to accept that her dream night out isn’t going to happen. And when she returns home well after 1:00 AM - the headliner came on an hour late, then proceeded to run through several extended renditions of tracks from his new album - she can just about persuade herself that she’s glad she went. Eager anticipation has been converted into fond recollection, with the actual experience of attending the concert best understood as a sort of transaction cost.
Did Jane make the wrong decision? The answer is quite obvious: who’s to say? We’re talking about a concert, not whether to pour more money into an unprofitable factory (the other example econ textbooks are required to provide). Jane’s utility function is whatever she wants it to be.
In the 60 Minutes interview that launched a thousand Medium posts, Jeff Bezos recalled applying a “regret minimization framework” when deciding whether he should leave a senior position at D.E. Shaw & Co. to work on the startup that would become Amazon. It’s a reassuringly technical spin on the ancient idea of imagining yourself on your deathbed.1
An alternative regret minimization framework would simply be: regret nothing. Said differently: rationalize everything
Bezos projected himself at age 80, and from this perspective was able to look back and see which potential regrets had the greatest potential to linger.
An alternative regret minimization framework would simply be: regret nothing. Said differently: rationalize everything. Contra Nietzsche (pro Wagner?), whose “formula for greatness in a human being” was to want “nothing to be different, not forward, not backward, not in all eternity,” the amor fati doesn’t make us great, only human, with all the flawed reasoning that entails.
The sunk cost fallacy describes how we muck up our future with residue from old decisions that we’re not quite ready to clean up. Along with the winner’s curse and the Grossman-Stiglitz Paradox, it’s one of those surprisingly poignant concepts that you occasionally find in economics. To me, what makes the sunk cost fallacy a compelling picture of human behavior is not just that it shows how we’re led astray by irrelevant considerations. It’s that once led astray, we’re still able to convince ourselves we’re on track.
Mistakes have been made. Maybe they’ll be expensive. Or - as two of the most hopeful words in the English language would have it - maybe not.
The Matching Principle
Sunk costs can be contrasted with incremental costs, but the best way to extend this metaphor (spreading it really thin in the process) is by borrowing some definitions from accounting. In particular, I’m thinking of the difference between costs that are expensed and costs that are capitalized.
Expenses are outflows in connection with producing a good or delivering a service. They aren’t bad, per se, but on their own they do have the annoyingly definitional feature of reducing equity. You incur an expense: now you have less useful stuff (assets), or you owe more useful stuff to others (liabilities).2
Expenses are sometimes described as offsetting the revenue they help generate. At first this appears to be a statement about quantities, but it’s really a statement about timing. Whether an expense helps generate a lot of revenue (e.g. Apple paying a store employee ring up customers for iPhones) or less than a little (e.g. Microsoft paying a store employee to stand behind the register and wait for someone to buy a Windows Phone), both are recognized in the period when the sale occurs. There’s no assumed future benefit from this kind of cost. For better or worse, what you see is what you’ve already gotten.3
And where you see it is on the income statement. It’s no accident that “the bottom line” is a widely understood colloquialism: the financial statement from which the term originates is itself a marvel of simplicity. Income statements can obscure as much as they reveal, but like meditation or alcohol, they do encourage us to live in the now. No payables, receivables, or inventories here. Just revenue coming in, expenses going out, and at the bottom of it all, a one line score to tell you how you did.
Intangible Investments
If only life were so simple. If only accounting were. The beautiful abstractions of the accrual basis have some flaws when you look closer. Cold, hard cash4 turns out to live by its own rules. By its own schedule too, coming before or after but seemingly never at the same time as booked revenue.
Things are just as complicated going in the other direction. Not all costs are treated equal. Some costs send cash out the door. Some don’t. But for purposes of this admittedly ponderous accounting metaphor, the complication that matters most is that some costs are expensed right away, and some aren’t.
Instead, they’re capitalized. This adds a new asset to the balance sheet. We might have less cash than before,5 but we have something useful to show for it. Something useful and lasting. Say, a piece of land to build a factory. Or a factory. Or a piece of machinery in the factory. Or even a pile of raw materials that we’ll feed into that machinery.
When spending results in a tangible long-term asset like the ones above, we call it capital expenditure and represent its cost by gradually marking down the asset’s value over time. This is depreciation. Though technically an expense meant to reflect the fact that the universe tends towards entropy and everything that ever has been will one day cease to be (i.e. a decrease in an asset’s remaining useful life), depreciation doesn’t bum people out as much as you might expect. Depreciation is actually pretty sweet: you can subtract it from earnings to reduce your tax bill, or add it back on with some other things to create fun acronyms like EBITDA.
Amortization is when the cost of an intangible asset is expensed over its useful life, and to the extent that this post can be said to be about anything, it’s this: intangible assets are really hard to value.
Accountants punt away the problem through the conservatism principle. Since the precise value of a company’s brand, technology, or operating practices is impossible to know, these “assets” are often kept off the balance sheet. The results can seem a little silly, as if a doctor measured a patient somewhere between five foot eleven and six feet tall, but recorded his height as zero inches so as to avoid getting it wrong.
Take the example of two superstar firms.
If you look at the asset side of Hermès’ balance sheet, you’ll see cash, some inventory, a few stores, and no mention of the brand value that has allowed a company selling handbags to achieve a 170 billion Euro market cap.6
Likewise, in 2021 Alphabet spent $31.56 billion on R&D - slightly more than the entire state budget of South Carolina. And while much of that spending was internal investment made to keep the all-important Google search engine humming along as perhaps the most profitable product ever created, Alphabet reports essentially none of that as an asset on the balance sheet.7
To understand both Hermès and Alphabet’s success, we have to look beyond the expense vs. capitalization dichotomy and see that many of the costs they recognize today will produce lasting benefits in the future. These companies are in the business of monetizing assets that are not only intangible, but also invisible.
Probable Benefits, Reliably Measured
In the year of our Lord Michael Mauboussin 2023, the promise and perils of capitalizing spending on intangible assets is well known.
I’d like to say that the argument I’m making here is that we should all try to behave like Google, spending vast amounts (of time and attention) to develop our idiosyncratic human capital, secure in the knowledge that the upfront costs will produce returns well after the point when accounting conventions (or less curious people) would suggest they do. But in truth I can only get behind a weaker version of that argument. We should all be wary of behaving like Worldcom, covering up our current problems by pretending we’re investing in the future rather than spending our days wastefully.
But our brains don’t come staffed with an internal audit department. There are a million ways to deceive ourselves, and none is easier than overrating the value of the information we consume. Everything we read - or who am I kidding, listen to - has to be adding up to something, right?8
Of all the types of loss we are averse to, the most painful might be when we’re forced to mark down some intangible asset that we’ve been carrying around on our personal balance sheet at more than fair value. Our sparkling wit, our brilliant insights, our hard-won wisdom … it’s not fun to test these for impairment. When you do, you might come to the same conclusion as Philip Larkin:
[that] once you have walked the length of your mind, what
You command is clear as a lading-list.
Anything else must not, for you, be thought
To exist.
And what's the profit? Only that, in time,
We half-identify the blind impress
All our behavings bear, may trace it home.
But to confess,
On that green evening when our death begins,
Just what it was, is hardly satisfying,
Since it applied only to one man once,
And that one dying.
Larkin begins that same poem by saying that, “Continuing to live … is always nearly losing.”9
This loss of interest, hair, and enterprise —
Ah, if the game were poker, yes,
You might discard them, draw a full house!
But it's chess.
I’d like to believe the game can be like bughouse chess. You lose some pieces, sure, but with the help of others you might pick up some new ones too. Sometimes you even manage to win a piece and feel the satisfaction of passing it on. It’s nice to feel needed.
My hope is that this site can be a bit more than what this inaugural post suggests it might be. (A “hardly satisfying” attempt to “half identify” the various strains of thought that “applied to one man once.”) I’d be thrilled if writing in public spurred me to discard old beliefs, and maybe pick up some new ones too. I’ll even try to pass on some useful information along the way.
Thanks for reading. It’s nice to have some company as you walk the length of your mind.
Or, in the case of someone who’s backed a startup in the life extension space, imagining yourself approximately 60% into a 130 year lifespan.
Expenses can reduce equity via the asset side of the balance sheet in several ways, and some are more annoying than others. A decrease in cash (or I suppose, prepaid expenses) hurts more than a decrease in the value of fixed assets resulting from depreciation, because the reduction in useful life represented by depreciation schedules is … what’s another word for fake?
If I were better with Substack formatting, I would mark this footnote with a literal GIANT ASTERISK. There are a lot of things wrong with the statement above. For one, companies will commonly plan for expenses to exceed revenue during a growth phase: the losses they report today might be a necessary part of building up to a scale where profits can be made tomorrow. For another, accrual accounting separates when cash is received (or spent) from when revenue (and hence expenses) are recognized. Since businesses need cash to survive, the philosophical implications of what constitutes an expense is perhaps less important than the practical one of, “how much cash do we have at this exact moment?” We’ll mostly skip over the topic of inventory accounting, which is its own thing. But it is worth emphasizing here that some expenses don’t have an immediate effect on revenue - that is, they don’t generate any - but could arguably be classified as a type of investment. R&D spending has to be “paid for” all upfront (i.e. expensed) rather than recognized over time, even though the intangible assets this R&D helps create - patents, codebases, etc. - are expected to generate revenue in the future. The point - of this footnote, of this whole post really - is that those future benefits are uncertain and that humans don’t do great with uncertainty, whether we’re talking about GAAP or assessing the value of the thoughts in our head.
Which as any crypto bro can and will tell you, is itself an abstraction.
Or not! We might have the same amount of cash and a new liability instead.
There’s an even larger French luxury firm whose balance sheet better reflects implicit brand value: the conglomerate LVMH reports more than 50 billion in intangible assets, mostly associated with acquisitions of brands such as Tiffany’s.
Alphabet shows under $1.5 billion in net intangible assets, much of which comes from acquisitions of non-core technologies such as Fitbit.
“Much reading robs the mind of all elasticity, as the continual pressure of a weight does a spring…the surest way of never having thoughts of your own is to pick up a book every time you have a free moment.” I read this in a book of Schopenhauer aphorisms that I often pick up in my free moments.
“Or going without,” he elaborates. “It varies.” Elsewhere he condenses the formulation into, “Life is slow dying.”
The Sunk Thought Fallacy
Probably off the main point of your blog, but I wonder, having just retired if the decisions I made over the years on the work I chose to do were more based on regret minimization versus rationalize everything. I think it is a bit of both. Life takes you in different directions and at times it can be wild ride. Now for the second act - regret minimalization or rationalize everything?