Sixty-one percent of this meeting cannot be checked

Here's a sentence a reasonable, hard-working engineer might say in a standup: "Yeah, still moving on that." Now do something slightly unfair to it: try to imagine the future event that would prove it false.

You can't. Not because it's definitely true — but because it doesn't point at anything. No function, no file, no ticket, no date. It has the grammar of an update and the epistemic content of a shrug. And here is the uncomfortable part: this is not a rare, degenerate sentence that only bad communicators produce. In a normal standup, it is closer to the median utterance than the exception.

The philosopher who ruined "that sounds about right" for everyone

Karl Popper spent a career on a single, sharp distinction, laid out fully in The Logic of Scientific Discovery: what separates a scientific claim from one that only sounds like it is not whether it's true. It's whether it's falsifiable — whether there exists, at least in principle, some observation that could prove it wrong. "All swans are white" is scientific, because one black swan sinks it. "Things happen for a reason" is not scientific, because no possible event could ever contradict it; it will comfortably absorb anything reality throws at it and claim victory either way. Popper wasn't being pedantic. He was pointing at the exact mechanism by which a claim can feel meaningful, sound confident, get nodded along to in a room full of smart people, and carry zero actual information.

Put a standup through that filter and something uncomfortable happens to about half of what gets said. "Going well." "Still chipping away." "Should be fine." "No blockers, just slow." None of these are lies. All of them are Popper's unscientific swans: statements engineered, usually with no ill intent at all, to survive contact with any possible future. Nothing that happens between now and next Tuesday can prove "still chipping away" wrong, and a claim that cannot be proven wrong is not a claim about the world. It's a sound the meeting makes to keep moving.

Popper's own target, worth remembering, was grander than office chatter — he was drawing the line between science and pseudoscience, arguing that Freudian psychoanalysis and Marxist historical prediction shared a specific flaw: both could explain any outcome after the fact, which meant neither one had ever risked anything before the fact. That's a big claim about entire intellectual traditions. It scales down to a standup disturbingly well, because the mechanism is identical at any size: a claim that can absorb any future is not describing the future, it's decorating the present. Whether the topic is the unconscious mind, the arc of history, or the state of the retry logic, the test is the same one-line question, and most standup sentences quietly fail it without anyone noticing they were even being tested.

What makes a claim checkable, concretely

The difference is almost always one missing ingredient, and it's worth naming precisely because it's so easy to add and so rarely added: a subject. Not a vibe, not a project name, an actual thing that exists and can be examined later — a function, a file, a specific department's workflow.

• "Going well" has no subject. Unresolvable by construction.
• "The charge() timeout fix is going well, should land Thursday" has a subject, charge(), and a date. Fully checkable — a commit either touches it by Thursday or it doesn't.
• "I'm blocked" has no subject. Blocked on what, blocking what?
• "I'm blocked on the staging DB credentials, which is holding up the export job" has two subjects and a relationship between them. Checkable — did the credentials get resolved, did the export job move?

Notice that turning an unresolvable sentence into a checkable one costs almost nothing in breath. It costs something else, though, which is the actual reason vague verbiage is so popular: a subject is a commitment to being provably wrong later, and "going well" is a commitment to nothing at all.

It isn't just status updates that go unresolvable

The temptation is to think this problem lives only in progress updates — "going well" and its cousins. It doesn't. Every kind of thing a person says in a standup can fail the same test, and it's worth walking through the other shapes it takes, because they're less obvious.

1. A risk with no target. "That refactor could be risky" names a feeling, not a subject. "That refactor could break the checkout flow" names a subject — checkout — and can be checked against the fault rate on checkout after the change ships.
2. A question that answers itself if anyone looked. "Does anything still call the legacy export path?" is a real question with a real, checkable answer sitting in the call graph right now — asked out loud in a meeting because nobody thought to check first. It's not unfalsifiable; it's just unresolved by the room, when it never needed to be.
3. A decision with no name attached. "We should probably drop that cache at some point" floats in the room, agreed to by nobody in particular, owned by nobody in particular, and is functionally identical to never having been said, six weeks later, when someone asks whether it happened.

None of these are lies, and none of them are laziness in the way that word usually implies. They're the standard output of a fifteen-minute meeting where speed is the only thing being optimized, and specificity is the first casualty of speed every single time.

The number, and why it's the finding, not a scold

This is where a real observation is worth more than a stern lecture. Across the standups CodeNSM has processed as claim extraction, a striking and consistent share of everything said in a typical fifteen-minute standup — in our fleet telemetry, often in the neighborhood of six in ten sentences — names no checkable subject at all. Call it what one pilot customer's dashboard actually showed on a representative Tuesday: sixty-one percent of that meeting could not be checked against anything, ever, no matter how much telemetry existed. That number is CodeNSM's own observation from its own extraction pipeline, not a research finding, and it will vary team to team, week to week — but the order of magnitude holds up often enough to be worth taking seriously.

Sit with that for a second, because the instinct is to read it as an indictment and it isn't one. Popper's whole point was never "unfalsifiable statements are bad" — plenty of meaningful human communication (encouragement, context, humor, the social glue that makes a team a team) was never trying to be a scientific claim in the first place, and shouldn't be held to that standard. The finding isn't that people are failing to be precise. It's that a standup silently mixes two entirely different kinds of speech — checkable claims and social noise — and gives them the exact same fifteen seconds of airtime and the exact same nodded acknowledgment, with nothing distinguishing which is which. Popper gave philosophy of science a way to sort the pile. Nobody had built the equivalent sorter for a standup.

An unresolvable claim isn't a failure to be graded. It's a fact about the meeting, and it's the single most useful fact a team can be handed — because it tells you exactly how much of your status reporting was never able to be wrong, which means it was never able to be right either.

What changes once the number is visible

Here's the part that makes this worth measuring instead of just grumbling about. CADENCE, the standup module of CodeNSM, treats "unresolvable" as a first-class outcome alongside "came true" and "did not" — never hidden, never quietly folded into a failure, because absence of a checkable subject is not evidence of anything. It is simply reported, with its own count, the same way a survey reports "no answer" instead of guessing.

And the behavior change, where it happens, isn't enforced — it's earned by visibility. Teams that see their own unresolvable share tend to start doing, unprompted, the cheap thing that fixes it: naming the function instead of saying "that thing," naming the date instead of saying "soon." Nobody has to be told their language was vague. The number just makes it visible in a way that a room full of nodding colleagues never could.

So the next time your standup wraps and everyone seems satisfied, ask the uncomfortable Popper question about your own sentence before you ask it of anyone else's: what, specifically, would prove you wrong? If you can't answer, you didn't make a claim. You made a sound the meeting was willing to accept as one.