The Forge

Build your own thing, or join someone else's?

532 people have stood exactly here. 55 of them are still inside it — their capsules have not opened yet. The rest already know how it went.

57%would decide the same way again

16 mowhere the two paths cross

25 momedian time since deciding

The answer changes depending on when you ask

This is the finding that matters, and it is the reason a snapshot of opinion is misleading. Shortly after the decision, the people who joined are happier. Years later, the people who built it are — and by a wider margin.

Time since deciding	Built it	Joined
6 months	52% would do it again	61% would do it again
12 months	59% would do it again	61% would do it again
24 months	61% would do it again	50% would do it again
36 months	65% would do it again	42% would do it again
48 months	67% would do it again	49% would do it again

Regret about building it fades. Regret about joining accrues.

Where each path ended up

They chose	Would again	Mixed	Would not	Total
Built it	156	36	66	258
Joined	115	46	58	219

What they said once they knew

Written after the capsule opened — not while deciding.

“I built alone to avoid being disagreed with. It worked perfectly.”
Built it · 67 months on

“I waited until I was ready. Someone less ready shipped it.”
Joined · 66 months on

“Steady, and slightly small. I keep a folder of what I'd have made.”
Joined · 66 months on

“The company didn't survive. Everything I learned building it did.”
Built it · 66 months on

Common questions

What usually happens to people who built it?

In this archive, 67% of people who built it said they would do it again when asked around four years later. At six months it was 52%. Regret about building it tends to fade as time passes.

Is it better to building it or joining?

Neither, universally — and the honest answer changes with time. Early on, joined looks better (61% against 52% at six months). Later it reverses (67% against 42% at four years). The two lines cross at about 16 months.

How is this measured?

Every capsule records a decision and a confidence level before the outcome is known, then the person is asked afterwards whether they would decide the same way again. The percentages are Gaussian-smoothed over 477 answered capsules. Where too few sit under the kernel, no figure is shown rather than a noisy one.

Is this advice?

No. It is a record of what happened to other people, which is a different and more useful thing. Nobody here knows your situation.

You are about to decide something

Write down the real reason and how sure you are, before you know how it goes. In 30, 90 or 365 days it opens and asks you one question — and you find out whether your judgement was worth anything.

Seal a capsule