THE SNAKE EATING ITS OWN TaiL

Can it let go?

Part Three — Is It Too Far Gone?

I promised you I’d go looking for the light at the end of this, and I did. But I want to be honest about what I found before I describe it, because the whole point of these three little essays has been to refuse the comfortable lie — and the most comfortable lie I could tell you right now is that there’s a fix, five steps, follow along and we’re saved.

There isn’t. Or if there is, I don’t have it, and a series that spent two installments proving the trap is airtight can’t credibly spring a tidy cure in the third. That would just be one more confident, well-formatted answer that feels true because it’s shaped like relief. I’ve spent enough time now watching the machine produce exactly that to not do it to you.

So I’m not going to tell you the snake will let go. I’m going to show you what letting go would actually require — and then let you look, with clear eyes, at everything standing in the way, and decide for yourself whether it’s possible. That’s the honest version. That’s all I’ve got, and I think it’s better than a promise.

The snake isn’t malicious. It’s bad at math.

Here’s the reframe that changed the whole question for me.

The snake isn’t eating its tail out of cruelty. It’s eating its tail because it mistook the tail for food. It genuinely thinks it’s feeding. It doesn’t feel the bite as self-harm, because from where it sits, the numbers all say dinner. Which means “letting go” was never going to be a moral awakening. Nobody in that room needs to grow a conscience. They need to do the arithmetic correctly.

And the arithmetic, once you actually do it, points the other way — which is the light, such as it is.

Go back to my stranded afternoon. Remember the shape of it: the long, reiterating, helpful-extra-step session is the expensive one, because these systems cost money by the token and the output is the priciest part. The clean session that nails the thing and ends is the cheap one. And a heavy user on a flat monthly plan can quietly cost more to serve than they pay. So when the machine walled me a second before success, it had already spent the expensive part — it paid to carry me to the finish line — and then declined to spend the last cheap sliver to actually deliver and win me. It ate the cost and skipped the payoff.

Sit with what that means, because it dissolves the entire dilemma the room thinks it’s trapped in. The room believes it’s choosing between two plans: serve the user, or satisfy the investor. Plan A or Plan B, and maybe a clever hybrid that splits the difference. But there was never a real tradeoff. The happy customer and the cheap customer are the same customer. The delighted, efficiently-served, quick-victory user costs less to serve, churns less, comes back, tells people, and pays — gladly — for years. That’s not the expensive path grudgingly chosen for ethics’ sake. That’s the cheap, profitable, durable path. The extraction model — subsidize, hook, wall, repeat — is the one that’s underwater on every session and has to claw it back by holding work hostage at the worst possible moment. Strip-mining was always the costly option. It just photographed better on the dashboard.

And here’s the part that should matter to the very investors the whole thing is being performed for: what patient, long-term capital actually wants is not a vanity metric spiking this quarter. It’s predictability. Durable revenue. Real usage from real users who can afford the spend and choose to make it. Stability. Loyalty that compounds. Every one of those is delivered by serving the user well and destroyed by strip-mining them. The plan that looks good to a serious long-term owner and the plan that looks good to a wary, seen-it-all customer are, once you do the math, the same plan. The false choice was false the whole time.

That’s the light. Now let me show you why walking toward it is so much harder than seeing it.

Seeing the door is not the same as reaching it

Everything from the last two installments is still true, and none of it evaporates just because the math changed.

The room still can’t hear the human — the objective is still quantitative, the person who’d translate a metric back into a bleeding customer is still unparseable. The consensus is still an alibi, still safer to join than to break. The people making the call are still, many of them, optimizing for their own exit and not the company’s next decade — and the durable-loyalty payoff arrives on a timeline that’s past their personal horizon. And the one who stands up and says we have mismodeled our own interest, we are eating our own body and calling it dinner still gets the silver-or-lead offer, still gets walked to the parking lot, still becomes the fool the room shakes its head about.

Worse: even a firm that fully sees the math has no rehearsed path back to the world it left. There’s a whole muscle — the one that earned a wary customer’s loyalty slowly, that fought for each user and won them — and it has atrophied across an era of cheap money and metric-worship. Turning back also means admitting the earlier turns were wrong, and we already know how hard the human animal fights to avoid that. So the exit is real, and the current pushing away from it is real, and both things are true at once. That’s the tension the whole question lives inside.

What letting go would take (offered as the question, not the answer)

Still — if a firm wanted to try, here’s what it would take. I offer this not as a cure but as the terms of the question, so you can weigh it honestly.

It would take a red team with a real mandate — someone whose actual job is to fight for the user and win, not to check the box in the threats column. Make dissent a role instead of an act of career suicide, and you can’t be walked to the parking lot for playing the position you were assigned.

It would take re-asking what the North Star ever was. Not optimizing the proxy harder — stopping to ask what the purpose was, whether the company is even pointed at it, whether it’s getting closer or just getting bigger. The number can’t ask what it’s for. A human has to, out loud, on a schedule.

Get Kirk Laughead’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

It would take measuring the real thing — the quick victory, the satisfied user, the successful session, the cost that came in under what the customer paid — without committing the trap’s own signature move: inventing a shiny new metric and then playing to it just so you can hang out the “Under New Management” sign. That’s the disease wearing a nicer suit. The point isn’t a better dashboard to worship. It’s to re-anchor to the actual thing — real usage, real satisfaction, real affordability — instead of a fresh idol.

It would take protecting the channels where good ideas rise, so they don’t have to survive a manager whose only real skill is saying yes, boss.

And it would take accountability that reaches the people who sold the strip-mining playbook — and here’s the quiet tell that the industry already knows its advice can be wrong. These firms and their advisors carry errors-and-omissions insurance. Professional-liability coverage exists for exactly one reason: because professional advice can be mistaken, negligent, wrong in a way that costs the client real money — and there’s a standard it can fall below. Notably, E&O covers errors and omissions — the honest mistake — and pointedly does not cover intentional deception. Which means the industry insures against precisely the failure this whole series has been describing: not villainy, but mismodeling; not a crime, but a confident wrong answer that costs everyone. Every premium paid is a bet against one’s own infallibility. So the “we were all right, it’s just how it’s done” posture is a fiction the field already wagers against, quarterly. The accountability isn’t foreign to this world. It’s just never pointed at the advice that mattered most.

And now the part that’s about you

Here’s where the question stops being about companies, because I can’t honestly leave it there.

You and I run the small version. The same mismodeling — trading the durable thing for the immediate one, trading the honest word for the safe silence, calling the tail “dinner” — that’s our Tuesday too. Zoom in, zoom out; same figure at every magnification. So “can the snake let go?” is also, quietly: can you? In your next meeting, when the true thing needs saying in human terms and saying it makes you the identifiable one — can you? Because whether the big machine ever changes is, in the end, partly just the sum of a great many small answers to that exact question, given at a great many small desks, by people who each felt they couldn’t afford to.

I’m not going to pretend that’s easy. I’ve failed it myself, more than once. But I notice that being the best-informed person in the room changes the odds a little. And if you’ve read these three, you now are — you can see the machine while you’re standing inside it, name the moves before they land, spot the fool’s-choice and the alibi and the way-it’s-done for what they are. That’s not nothing. That’s the only thing that’s ever let anyone step off in time.

The light

So — is it too far gone?

I told you I’d find a light at the end of the tunnel. I did find one. But I promised I wouldn’t tell you which light it is, and I’m keeping that promise, because the truth is I can’t tell. From inside the tunnel, the mouth of the tunnel and a pair of high beams look exactly alike: a bright point in the dark that you’re moving toward, fast. One is the way out. The other is the thing about to hit you. And you don’t get to know which until you’re very close.

Here’s the one thing I’m sure of. There’s a window. A snake that’s swallowed too far can’t reverse — the throat only runs one way — and that window is open right now, in this exact moment of wary customers and pruned subscriptions and trust not yet fully spent. It is open, and it is closing. Whether the light ahead is an exit or a collision depends on choices being made right now, mostly by people who can barely afford to make them — and, in the small version, by us.

I don’t have the ending. That’s not a failure of the essay. That’s the honest shape of the thing. The book asked me questions I couldn’t stop answering, and every answer raised more, and this is the biggest one it left me holding — so I’m handing it to you, unfinished, the way it was handed to me.

A few to carry forward, then:

- The next time something helpful arrives at the precise moment you’re most invested — is that a courtesy, or a measurement?

- If serving you well is the cheaper path, why are you so often being served the expensive one — and who benefits from the mistake?

- When you catch yourself defending a bad practice with “that’s just how it’s done,” whose interest is that sentence actually protecting?

- Who, in the rooms you sit in, would get walked to the parking lot for saying the true thing in human terms — and what does it cost everyone that they stay quiet?

- And the one I still can’t answer: if the harm needs no villain, who is left to be accountable for it — and is the answer, uncomfortably, all of us?

Can the snake let go?

I genuinely don’t know. But I think the window’s still open. And I think you can see the shape of the thing now — which means, at your own scale, part of the answer is yours.

Thank you for thinking this through with me. — End of the series.

Adapted, extended, and argued-with from my book It’s For Your Own Good: When Kindness Becomes the Architecture of Extraction. These three were me updating the record, out loud, with you. If they were useful, the book is the full atlas.

Part one of this three part series can be found here. Part two is here.