The Doubt Is a Redaction.

Three years before ‘Oumuamua made “interstellar object” a headline, a half-metre rock may have beaten it to the record — by hitting us. On 8 January 2014 a fireball burned up over the Pacific near Papua New Guinea. It sat unremarked in a US government catalogue of fireballs until 2019, when Amir Siraj and Avi Loeb noticed its speed and made an extraordinary claim: it was moving too fast to belong to the Sun. If they’re right, it was the first interstellar object ever detected, and the first we could, in principle, pick up off the seafloor. Whether they’re right comes down to a single number nobody will release.

Where we land: held open. Probably interstellar on the reported data, but the error bar that would settle it is classified. The doubt isn’t in the sky — it’s in a redaction.

1 · The line between “ours” and “not ours”

There’s a clean, exact boundary. At Earth’s distance from the Sun, anything gravitationally bound to the Sun must be travelling slower than the solar escape speed — √2 times Earth’s orbital speed, which works out to 42.1 km/s measured relative to the Sun. Faster than that and the Sun simply can’t hold it: it came from outside, and it’s leaving. This meteor — catalogued as CNEOS 2014-01-08, nicknamed IM1 — had a reported speed of about 60 km/s relative to the Sun: well over the line, with a “speed at infinity” (how fast it would still be going after escaping the Sun entirely) of 42.1 ± 5.5 km/s. Taken at face value, that’s not marginal. It’s emphatically interstellar.

2 · The number that was never published

Here’s the catch, and it’s the whole story. That ±5.5 is Siraj and Loeb’s own estimate. The CNEOS fireball catalogue — fed by classified US defence sensors — lists a velocity but no uncertainties at all. And the interstellar verdict is nothing but a statement about that uncertainty: how many error bars the speed sits above 42.1. Sweep the unpublished error bar and watch the confidence you’re allowed to have slide:

Run the sweep and the odds tell the tale. If the true uncertainty really is 5.5 km/s, the chance it was interstellar is 99.94%. Widen the error bar to 10.0 km/s and it slips to 96.32%; at 15.0, 88.34%; at 20.0, 81.44%; at 25.0, 76.28%; even at a bruising 30.0 km/s it still holds 72.45%.

Read honestly, that cuts both ways. Even a pessimistically large random error keeps it more likely interstellar than not — the reported data genuinely favours the claim. But no honest confidence can actually be stated, because the real error bar is classified, and no outside scientist can reproduce the result from raw data they’re not allowed to see.

3 · The deeper objection, and the mud

Random scatter isn’t even the strongest doubt. A 2022 analysis argued that the fireball’s jointly extreme properties — very high speed and improbable material strength — are better explained as systematic measurement error than as real values: the same sensors, never built for this job, may simply have over-read the speed. A systematic bias isn’t a random error you can average away. Then in 2023 a dredging expedition pulled metallic spherules from the Pacific floor and floated an interstellar — even technological — origin; independent researchers attribute them to ordinary micrometeorites or industrial ash. None of it is decisive, in either direction.

An official can vouch that a number is “accurate enough.” That is not the same as publishing the error bar — and science runs on the second thing, not the first.

Why our math sees more

The surface story is a binary: “first interstellar meteor!” versus “debunked.” We’d rather locate the claim exactly — it is one inequality, v > 42.1 km/s — and then ask what that inequality actually rests on. The answer is an unpublished uncertainty, so the honest confidence is not 100% and not 0% but uncomputable, and the result unreproducible. Saying so precisely — naming the redaction as the thing standing between us and the answer — beats either headline.