A team cannot reliably judge whether its own idea is new

How we check our own ideas

The person who builds an idea is the worst-placed person to judge whether it is new. This is not a personal failing. It follows from how ideas reach a founder in the first place: an idea gets their attention because they find it exciting.

Excitement does not show that an idea is original. It shows only that the founder has not seen the idea before, and a founder who had seen it would not find it exciting. So we do not ask the people who build an idea to judge it. We ask someone whose only goal is to find a reason to reject it.

Why builders miss this by default

The cause is not a lack of intelligence. It is the situation builders are in. Building something takes steady belief in it, and it is hard to keep that belief while also searching seriously for the fact that would prove it wrong.

Nobody decides to skip that search. But after months of building, a team stops seeing the search that would hurt it. Without noticing, it runs a search that confirms what it already believes. An outside reviewer with nothing at stake does not have this problem, because they have no months of work to protect.

What we did

We took twelve ideas from twelve different fields. A team at Syntaxe was genuinely excited to build each one. We gave every idea to an independent reviewer with one instruction: find the reason this already exists, or the reason it will not work, and keep looking until you find it or run out of places to look.

How the review works

The reviewer receives only the idea and the reason given for it. We deliberately withhold the internal reasoning that produced it. This means the reviewer starts where a capable outsider would start when seeing the idea for the first time.

The reviewer succeeds in only one way: by finding the fact that rules the idea out, or by searching every available source without finding one. Each idea then receives one of three results:

• Survive: the reviewer found no reason to reject it.
• Sketch: the idea overlaps with existing work in a real but not fatal way. It is worth returning to once the overlap is understood, but not worth putting time and money into as it stands.
• Kill: the idea already exists, or it fails for a concrete reason.

What happened

All twelve ideas had already passed our own internal check. The teams behind them were confident, and with good reason. The ideas made sense, and nothing in our process up to that point had found a problem that ruled them out.

The outside review marked all twelve as sketch or kill. Not most of them: all of them. In almost every case, the reason was existing work that the builder could not have seen from inside the process that produced the idea.

In several cases, that existing work was not hard to find. It had been published, sometimes years earlier and sometimes by a well-known research group, using different words for the same underlying method.

In a smaller number of cases, the idea had not been published anywhere. There, the reviewer found a specific reason the idea would fail soon after it moved from planning into practice. That kind of failure is visible only to someone who is actively looking for it.

What we learned

Enthusiasm does not prove that an idea is new. Confidence inside a team measures how thoroughly the team has convinced itself. It does not measure how much of the same ground others have already covered.

An idea is new only when someone whose whole job is to make it fail tries to disprove it and cannot. That is a much higher standard than “our team likes this”. It is the only standard that actually tells us whether an idea is worth building.

What changed

This is no longer a one-off exercise. It is now a permanent step in how we work. Every possible new direction at Syntaxe goes through the same outside review before it receives real time and money, on the same terms: an outside reviewer whose only aim is to reject it, given as long as it needs.

Most proposals do not survive. The ones that do have already faced the kind of scrutiny they would eventually meet from the outside world. They have simply faced it earlier, when finding a problem costs far less.

Another note on this site argues that capable people working on the same problem often reach the same idea, so the idea alone rarely gives a company a lasting advantage over competitors. This report tests that argument on our own work, with a result: twelve ideas checked, and all twelve rejected or downgraded.