Groundbreaking Isn’t the Test. Coherence Is.

Isaac Matamoros, who advises on EU-funded innovation programs, published an essay this month asking a question that sits underneath almost every innovation decision and is rarely asked outright:

“What actually makes one proposal worth backing while another is quietly left behind?”

The instinctive answer is that the strongest proposal is the most original, the most disruptive, and the one that announces itself through sheer ambition. Matamoros argues that instinct is wrong, or at least incomplete.

Disruption is not, by itself, a selection criterion.

The quality worth looking for is coherence:

Is the problem genuinely understood, does the proposed solution actually respond to it, can the team behind it adapt when its assumptions prove wrong, and is there a credible path from hypothesis to evidence?

None of that guarantees success, but together, it earns an idea the right to be explored further. Groundbreaking tells you something is new; it does not tell you it deserves to go first.

We told a version of this story last winter, before we had his language for it. A valley spring fed a wooden channel villagers had built themselves; it was simple, and it worked. A mountain chief admired it and ordered it scaled into a Grand Aqueduct System, commissioned from a distant guild, built without the spring-keepers who knew exactly what would break and how to fix it. The aqueduct was completed on schedule and never worked, so the villagers went back to the channel. The issue was never information; it was conviction. Leadership had already decided the working solution was not impressive enough to bother scaling. The council was optimizing for groundbreaking, but nobody stopped to check whether the idea actually held together.

That test has to be applied at each place an idea travels through an organization, and we have spent the better part of a year documenting exactly where it tends to go unapplied.

It starts with whether the people designing the solution seek out what the people who will live with it already understand. Too often they settle instead for what the people who commissioned it assume. We wrote in June about a data and dashboard initiative where the operational team offered to walk the design group through the reality on the ground, and the offer was never taken up. Steve Blank’s line about startups applies just as directly to command-directed platforms:

“No facts exist inside the building, only opinions.”

Declining a free walkthrough of the actual problem breaks the idea at the first checkpoint, before a line of code or a procurement dollar is spent.

It continues whether the solution matches the capacity of the team asked to run it. A platform mandated from a well-resourced level of the organization lands on a team with none of that room, one that inherits a second job with no budget for either. A solution can be architecturally sound and still fail this checkpoint, because nobody weighed it against the reality of who has to carry it.

It shows up again in how the AI industry argues over which vendor should own the governed layer sitting on top of an organization’s data and models. That fight assumes the rest falls into place automatically once someone wins architectural control. Matamoros’s essay is a useful corrective here too: owning the most sophisticated layer is a claim about novelty, not evidence that what gets built actually fits the problem an organization has.

And it shows up last in whether there is a credible path to durability, not just to initial validation. The Pentagon’s Gemini rollout we wrote about earlier this month was groundbreaking by any reasonable definition, and the comment thread under the announcement was full of people who could not reach it from where they actually work. Of the people who eventually could, nobody had been assigned to own the bridge once it existed. An access solution that lasts exactly as long as one person’s institutional memory has not passed Matamoros’s fourth test; it has only delayed the failure.

None of this argues against ambition. The spring-keepers were not asking the mountain chief to abandon the idea of a mountain-wide water system; they were asking to be in the room before the guild broke ground. The question an organization should ask about its next AI initiative, or any initiative directed from above, is not whether it is impressive enough to justify the budget line. It is whether the problem is understood by the people closest to it, whether the solution matches the capacity that exists to run it, whether someone can name who owns it next year, and whether anyone with the authority to stop and listen is willing to do it before the aqueduct is finished.

Groundbreaking is not a selection criterion. Coherence is, and it has to survive contact with the people who were never in the room when the decision was made.

If you are weighing whether your organization’s next initiative actually holds together well enough to survive its own rollout, that is precisely what our fractional AI and technology support is built to close: someone accountable for “the coherence question” after the launch, not just before it. Reach us at info@stormkinganalytics.com.