6,300 exoplanets, zero Earths: the word doing the heavy lifting is "like"

The measurement chain is short and it stops early. A transit gives you a radius. A radial-velocity signal gives you a mass. Orbital distance gives you an insolation figure, which the source describes as yielding only a vague idea of temperature ^[4]. Nothing in that chain touches the surface, because the mix of gases in an atmosphere, and whether there is an atmosphere at all, can change surface temperature wildly ^[5]. So "Earth-sized" and "Earth-mass" are conclusions ^[3]. "Earth-like" is an extrapolation across the exact gap the instruments do not cross.

Venus is the control experiment, and it is 40 million kilometres away rather than 40 light-years. Slightly smaller than Earth, slightly less massive, not much closer to the Sun ^[6]. Its surface is hot enough to melt lead and its sky presses down at 90 times Earth's sea-level pressure, the result of a runaway greenhouse effect in an atmosphere of nearly all carbon dioxide ^[7]. Run Venus through the exoplanet pipeline and it comes out the far side as a rocky, roughly Earth-sized world in the warm part of its system. Every number would be right. The habitability inference would be wrong by 400 degrees. The author's conclusion is blunt: Venus is very Earth-sized and absolutely not Earth-like in any real sense ^[8].

The catalogue itself argues against the assumption that similar dimensions imply similar worlds. The first exoplanets found were radiation-riddled bodies orbiting a pulsar ^[9]. Then came gas giants in orbits so tight they nearly skim their stars, and others so distant their formation is hard to explain ^[10]. Some rocky planets have molten surfaces; others are frozen solid ^[11]. A sample that diverse is a warning about how much of a planet's character sits in variables the discovery method never records.

The incentive problem is visible in the reporting cycle the source describes: any exoplanet with a vaguely familiar terrestrial characteristic draws press conferences, news stories and headlines speculating on what the world may be like ^[12]. That is not a neutral translation error. It spends the credibility of the field on results the field did not produce, and it prices in an achievement (a true twin, which the author calls a huge landmark that would put a framework around our existence ^[13]) that has not happened once in more than 6,300 detections ^[1]. Anyone budgeting for the instruments that could actually close the gap, atmospheric characterisation on small planets, is asking funders for money to answer a question the press releases have implied was answered years ago. The honest version of the pitch is harder to make and more likely to survive contact with the data.

Note what is and is not being claimed here. The author's position is that Earth-like has never been true of any exoplanet so far described that way, and he flags it as opinion ^[2]. Earth-like worlds may well exist ^[14]. The catalogue grows practically daily ^[1]. The complaint is about a word that has run ahead of its evidence, and about the fact that no one currently has to pay a price for using it.