Or a day-night-day-penumbra-umbra-penumbra cycle. The primary-facing side would get maximum light just before entering the penumbra, experience total darkness in the umbra, and get some light directly from the star and some reflected from the primary from the time it exits the umbra to the time it re-enters. Every day would have a full stellar eclipse at noon. Every night would see the primary cycle from waxing half through full to waning half. The space-facing side would have an ordinary day-night cycle with the same period as its orbital period around its primary.

"However, more recent research is somewhat more optimistic."

I am deeply skeptical of that "recent research", which AFAIK is "computer models, where we have zero capability to verify the computer models against even a single data point".

The incentives to keep kicking the computer model until it provides something publishable, with no countervailing force provided by real data to be explained by the model, are just too strong for me to take that too seriously.

Every new simulation is publishable. And every simulation that is more detailed than what came before will receive some attention after being published.

If physics has some bias they polish their models towards, it's on the direction of repeating the findings of the previous ones, not of changing them.

Physics models are often checked against other models (which might have had some confirmation) in simple cases, and there are a lot of other sanity checks you can make on them to ensure they are providing reasonable results.

Obviously, one shouldn't blindly accept them without experimental confirmation, but one shouldn't also be wildly pessimistic in believing calculations grounded in three centuries worth of validated equations.

Yes, I'm sure the simulations don't have results like making these atmospheres suddenly collapse into black holes due to errors in the simulation or something. They'd notice.

But these atmospheres have to stable across cosmological scales in a highly iterative, chaotic environment. There is a mathematical lower bound on the rate such simulations must leak bits as a result of those things. I don't think we have enough bits of real information to make up for that, as it would take rather a lot.

Against the mathematical argument, I set the human argument that, like I said, there's all sorts of incentives to report the one simulation that produced the result that there may be an atmosphere and maybe it could even sustain life! That's a lot of high-profile articles and probably a promotion for getting my department some public attention. Even if hundreds of other simulations all result in "Yeah, the atmosphere freezes out".

Between the math saying we can't really expect this sort of simulation to contain meaningful information and the human factors involved, I can't put a lot of confidence in a model which can't be validated against real data.

I do find it amazing how people who probably think they're really in favor of science and support it will jump up to defend a methodology that literally runs off of zero data. Can someone explain the scientific process to me clearly that involves having no data at any point?

well even our deep space observations are largely computed. nobody has "seen" these 2 planets at all ;)

what does seeing mean? does it have to involve photons? can gravity work as well? sensors and computation are always involved, nothing is directly seen.

How did we verify the earlier theories?

We check them against data.

I have no objection to computer models; I object to computer models that can't be checked against data.

Cosmologists run a lot of models, to do things like try to determine the effects of dark matter on the universe. Such models are intrinsically problematic since they have to have resolution on the order of hundreds or thousands of lightyears on a side, or have to work with universes much smaller than the real one, or something that means the simulation is by necessity literally several dozen orders of magnitude smaller than the real universe (consider both the timestep and spatial dimensions).

But they can check the result by looking up in the sky and seeing if it matches. This allows them to overcome the problematic nature of the models and use them to say real things.

One wonders what exciting cosmological papers proposing all sorts of amazing and outlandish theories as a result of some computer model of the universe have been suppressed by the fact that they didn't correspond to the sky at all. I guarantee you that some very amazing simulations have been run that produced incredible results of great interest, whose only crime was that they completely failed to match the universe.

We don't have a tidally locked planet that has an active atmosphere in our solar system to look at.