When to stop farming a project

There is a marker for the moment you stop believing in a project, and it has nothing to do with the project.

It is the week you stop bringing it up with other people who farm.

I can find that week for something I then carried for another seven months. The interactions kept happening, small ones, a few dollars at a time. The belief had gone long before the activity did.

Everybody has an entry rule

Ask anyone doing this why they picked a project and you get a real answer. Funding raised with no token issued, a live product with users who are not farmers, a points programme running in the open, a team that has said something about rewarding early participants. People think hard about entry, and it deserves the attention it gets.

Ask the same person what would make them stop and you get a shrug, or you get some version of “when it is obviously over”.

Nothing is ever obviously over. That is the whole problem.

I have looked across four other people’s tracking sheets in the last year. Every one had a start date and a reason for picking the thing. Not one had an exit condition written down anywhere.

Sunk cost wearing the costume of patience

Patience genuinely works here. A handful of wallets used properly across a long stretch will out score a hundred thin ones spun up in a weekend, because time is the input you cannot manufacture, and I have made that argument at length elsewhere.

There is a condition attached to it that does not get said loudly enough.

Patience is what a live opportunity is owed. The same behaviour aimed at a finished one is persistence, and persistence costs exactly as much while buying nothing.

From the outside they are indistinguishable. Same wallets, same Sunday routine, same person quietly pleased with themselves for not chasing noise. The only variable is whether anything still exists at the far end.

Which is why “one more season” is the most expensive sentence in this hobby. It commits you to nothing. It sounds measured. What it does is spend another few months of gas and attention defending money that already left.

The bar rises and the pool stays where it was

The signals worth acting on are structural, and every one of them can be checked against something published.

The first is a ratio. Programmes raise their qualifying bar all the time, and early in a season that usually means somebody is trying to price out the cheapest farming. That is fine and it is normal.

What matters is whether the other side of the trade moved with it. Requirement doubles, stated pool stays flat, participant count keeps climbing. You have been asked for twice the work in exchange for a thinner slice of the same thing.

A bar that rises while the pool rises is a programme still buying users. A bar that rises alone is a programme managing what it will eventually owe.

The revision points backwards

Criteria change mid season. That is ordinary, and I have covered the reasons for it in its own piece. Teams design these programmes before they have any usage data, then patch them once real behaviour shows up.

Direction is where the information is.

A change that adds a new way to qualify from today onward is a programme still trying to grow. A change that reaches back across activity you already completed in good faith and quietly revalues it downward is a programme reducing what it owes.

One of those is a rules update. The other is a budget decision dressed as one.

Once is an accident. The second time it lands on the same programme, I stop reading it that way.

The silence is selective

Teams go quiet. A slow month on a blog means very little, and building a thesis out of a communications lull is how people talk themselves into and out of everything.

The version worth noticing is narrower than that. The account still posts. Integrations get announced, the weekly update still ships, the ecosystem thread still runs. What has disappeared is any answer to the token question specifically.

The clearest form of it is a question that used to get answered. Somebody asks about timing in the Discord and gets moderator boilerplate, where the same question six weeks earlier got a direct reply from somebody on the team.

That shift usually means a decision was taken internally and nobody wants to be the person who says it out loud.

The crowd outgrows the incentive

This one can happen while every person involved behaves impeccably.

If the eligible population grows faster than whatever is being handed out, your share shrinks while you work, and no rule has to change for it to happen.

The shape of it is a programme that got popular. A quest platform lists it, a large account posts a thread, participation multiplies inside a fortnight. Nothing dishonest occurred anywhere. The arithmetic on your position simply got worse, and there is no mechanism by which it improves again.

Nothing is being built there any more

The slowest one to admit, because it means looking away from the dashboard.

Go and see what has actually shipped on the thing in the last six months with the points programme excluded from the count. Applications. Tooling. Projects that chose to launch there when they had options.

If everything of substance arriving is another incentive scheme, and the builders who were around a year ago have gone elsewhere, what is left is a farm with a token attached. Any eventual distribution out of that gets divided among people who look exactly like you.

Commit graphs are a crude proxy and they can be gamed, so do not treat one as a verdict on its own. A project whose only news is about points has already told you what it has left.

Slow and finished produce the same week

Here is why those five matter far more together than any single one does.

A slow project and a dead project feel identical from where you sit. Nothing happens. The number ticks up. You run the routine and you wait.

Slow is ordinary. Teams miss timelines as a matter of course, and eight quiet months followed by a launch is a completely normal sequence. Anyone who cannot sit through a dead stretch should stay out of this entirely.

The trouble is that finished feels exactly like slow, and it keeps feeling that way for as long as you are willing to read it generously. Nothing arrives to correct you. No project has ever published a notice saying the opportunity has closed.

So slow is the default reading, and it keeps that position until you make it earn it. The question worth putting to a project every few months is what would have to be true for this to be over, because nothing is going to ask it on your behalf.

The note that takes five minutes

The fix is mechanical and the timing of it is the entire point.

When a project goes on the list, write down three things. The date you started. What you have spent so far, in money and in rough hours. The condition under which you would stop.

That third one has to exist before any sunk cost does. Nine months in you are no longer capable of writing it honestly. You will produce a condition you already know will never fire, it will look perfectly reasonable on the page, and everybody does this including me.

A workable stop condition is a fact rather than a feeling. The ones I use:

• The qualifying bar has moved twice with no movement on the pool.
• The criteria have been revised backwards across activity I already did.
• Ninety days with no substantive answer on the token question.
• Participation has multiplied while the stated allocation has not.
• Six months in which the only thing to launch here was another incentive scheme.

Pick two. One on its own is too twitchy for an activity where quiet stretches are routine. Two together is a pattern.

And when they fire, you stop. You do not renegotiate. Renegotiating is the failure mode compressed into a single word.

What I did instead of writing one

I had a reason every week, and every reason was true.

The cost was small. I was set up there already. There might be a second season. It would be a shame to walk after nine months of it.

None of those were statements about the project. All of them were statements about my own history with the project.

The gas was the cheap part. What it really cost me was a slot, because I run about as many projects as I can genuinely keep up with, and one of those places was held by something I had already written off in private. There is real arithmetic behind what that trade costs and it deserves its own piece.

Things that are not signals

The opposite error is real and it is expensive too, so here is what I ignore.

A quiet month. A slipped roadmap date. A founder posting less because they are heads down building. The mood in a Discord, which is the least reliable instrument available to anybody in this.

I have been wrong in that direction as well. I cut something early once over a rule change I had decided was fatal, and it went on to do fine for the people who stayed. An exit rule will not be right every time, and a rule that would have been right every time is a rule fitted to the past.

None of this is financial advice, I am not going to tell you what any token will be worth, and I am not going to name a project and call it finished. I do not hold that kind of certainty about anything, and the people publishing calls like that are usually positioned somewhere else. What I can say is that stopping is a decision you are allowed to make in advance, on your own evidence, on terms you set while you could still think clearly. The rest of how I run this side of things is on the home page.

Get new guides and videos first — join the Telegram channel.