The Grey Tribe

The very first experience it has with the base crew is that they save it from people shooting at it, give it warm hugs, and kill the people trying to deatroy it. After that it attacks and impersonates an unknown (at the time) member of the crew. After that it gets surrounded by dogs who are angry but too scared to approach, then it changes, then it attacks the huskies, and only then does anyone in the base camp treat it with hostility.

The entire first act of the movie is devoted to establishing that they know each other with an Intimacy so deep they can anticipate one another's actions and attitudes. They have been in an isolated arctic base for months and months where they can barely leave the same building. They are in one another's personal space throughout the movie. It's a vital plot point that the Thing can immitate people down to memories and personality traits. It's a vital metaphorical point as well. It's so deeply and fundamentally superficial and factually incorrect to call them unfamiliar with each other that it implies total inattention to what is happening on screen.

The Thing is, at heart, not a movie about any singular decision or behavior creating a bad outcome. Baked into the 1982 movie is failure, death, entropy, inevitable loss. It's not a movie that's meant to have a right solution, or a right decision - but when someone comes at this very bleak story without a good grounding in horror, there's a kind of urge to treat it like a puzzle. If only they were closer. If only they communicated.

That's not meeting it where it's at, because it rests on a situation where none of those elements really exist. People acted the best they could in the circumstances with the tools and information they had - and it simply was not enough. Nearly everyone dies. Even with the ambiguous ending, whoever is human is going to die, because it's winter in Antarctica and he is hundreds of miles from anywhere with no shelter and no food and no transportation. That's the sort of horror it is, the idea that when faced with extinction humanity's best efforts won't succeed. Creating an interpretation where if we had "just" this or that is shying away from the bleakness. But at the same time, not facing up to the idea that some things really might not be solvable, that the worst can happen in spite of it all, is a necessary skill. Not one we need to indulge in constantly, but we should have that knowledge.

In brief, fandom seems to be predominantly comprised of people who only want to talk about what isn't in the media. There is a fandom for The Thing on tumble, and it's very much about ideas that come from the movie, but not about anything innthe actual film. The few times I've had the misfortune to enjoy media with a large fandom, I find almost exclusively shipping, fantasies about specific characters antithetical to the subject matter, and an almost oppositional distaste for any of the actual intended or unintended meaning in the media. None of those are a problem per se, because the world is a big place and it doesn't matter in the grand picture what people want to pretended about a book or movie or TV series. This is mostly a me problem, where I can be perfectly fine and ignore what comes across my dashboard, but this is also a me post, which means I can complain about the phenomenon.

Anyway, where I find it frustrating is the encroachment of fandom focus on what isn't in the media into my little wading pool of enjoying digging through the ideas that are in the media, when a "what if it was completely different" idea inspired by the media gains so much traction that it begins to be treated as a reflection of the text, sometimes even exiting its primary fandom, like The Thing, into general circulation. Which is how you get people discussing the "creature is misunderstood" not-present ideas of The Thing by fandoms, while effectively ignoring the vastly more significant elements of the film relating to the Reagan era 1980s, cold war paranoia, and so on. Once a fandom non-textual idea propagates far enough, it not only occludes the ideas inside the original media, in my experience it almost invariably excises anything that was challenging about it. For example, how rapidly fandom's latched onto the idea of Remmick in Sinners as a misunderstood tragic figure - rather than anything to do with what the film is saying about systemic racism. It feels like a companion behavior towards people whose take away from media like Breaking Bad, Fight Club, or Obsession is that the male protagonist is meant to be viewed in a positive light - because that is an easy and more comfortable option than for example treating any character who isn't a guy or isn't white or isn't straight as a vital, human person at the center of the film.

> realize every screen displays color slightly differently

> realize every camera captures color a little differently

> realize every format and image editor translates color

> realize every printer deposits ink differently

> realize every kind of paper accepts ink differently

> realize colors look completely different under different lighting conditions

> ok I'm going to do everything greyscale now

> black density changes as ink cartridge depletes and depending on viscosity after multiple prints

> "what color white are you using"

and it's tempting to write off all these examples as "pseudoscience" like UFO conspiracy theories or flat earth, that mimics the appearance of science while disregarding the process, but it's an awkward fact that you can apply the scientific method to insane hypotheses (psi powers) and technically tick all the boxes and maybe even reach a conclusion that is statistically significant in some abstract sense, so coming up with a definition of "science" that includes all the meaningful investigations and rules out all the silly ones is probably impossible; to some extent "science is when you write down what you're doing".

but because science is carried out by fallible human beings, many of whom have emotional attachments (or simply unexamined beliefs) and even 100% valid scientific results must be filtered through the biases and ideological strictures of society, it's always tempting to take it one step further and claim that science is meaningless and results are an arbitrary fiction, or a step further than that and claim that reality itself is meaningless, given that any way of interacting with it is unavoidably socially constructed in some way.

One camp wants to reform science. They are still concerned about the replication crisis, and they focus on methods, open science, peer review, datasets, registered reports, generalisability of experiments, p-hacking, and so on. These people use the language of science and STEM-based epistemology. Let's call them the reformists.

The other camp is – and there is no way around it – anti-science. They do not like "STEM bros", they don't believe in maths/numbers/statistics/quantitative research. They are much more into me-search, as long as it is unabashedly subjective, literary, and qualitative. In general, these people are opposed to reductionism, empiricism, systematising, or objectivity. Their epistemology is very different from the first group, and they will tell you that the first group is missing the point. Materialism is false. The project of science will forever be mired by human subjectivity. Let's call them the radicals.

The third camp is the establishment. These are your presidents of professional organisations, your Faucis, your Susans Fiske or your Carolyns M. Rouse. They will in general pay lip service to sweeping structural critiques from the second camp, and vigorously push back against the first. Let's call them the institutionalists.

Now back to women in particular: If you run into somebody who introduces herself as a "feminist scientist" today, she may be in any of these three camps. She may be an establishment spokeswoman who sees the ivory tower of science beset by skeptic randos with blogs and access to papers through sci-hub. Just as women made it to the top, methodological terrorists are destroying the ivory tower. Institutionalists hate actionable, constructive criticism.

She may be a radical, an adherent of "feminist physics", or an anti-science humanities writer who explains that women have other ways of knowing, or that the problem with computer science is the focus on machines and numbers.

Or she may be a reformist who thinks certain hypotheses had been underexplored and we need to broaden our horizons while also tightening up our methodological standards, and to get more women into STEM.

It is important to understand that there is a deep distrust between "women in physics" and "feminist physics". One side thinks the others are nutjobs who make them look bad, and the other side thinks you cannot reform science from the inside. If you suggest to a feminist woman in physics that you took her for a subscriber to Luce Irigaray's idea of phallogocentrism, she might get very annoyed!

There's another way to look at this, though. One camp believes that women and/or feminists have shared goals, shared interests, and shared experiences, and for this reason, you can elide or ignore the contradictions, or at the very least argue for solidarity or a temporary truce. People who should be at odds about science because they argue for fundamentally incompatible epistemologies will put their differences aside and agree to disagree, or even downplay the extent of the disagreement in public in some attempt at message discipline.

Multimodal AI systems have achieved remarkable performance across a broad range of real-world tasks, yet the mechanisms underlying visual-language reasoning remain surprisingly poorly understood. We report three findings that challenge prevailing assumptions about how these systems process and integrate visual information. First, Frontier models readily generate detailed image descriptions and elaborate reasoning traces, including pathology-biased clinical findings, for images never provided; we term this phenomenon mirage reasoning. Second, without any image input, models also attain strikingly high scores across general and medical multimodal benchmarks, bringing into question their utility and design. In the most extreme case, our model achieved the top rank on a standard chest X-ray question-answering benchmark without access to any images. Third, when models were explicitly instructed to guess answers without image access, rather than being implicitly prompted to assume images were present, performance declined markedly. Explicit guessing appears to engage a more conservative response regime, in contrast to the mirage regime in which models behave as though images have been provided. These findings expose fundamental vulnerabilities in how visual-language models reason and are evaluated, pointing to an urgent need for private benchmarks that eliminate textual cues enabling non-visual inference, particularly in medical contexts where miscalibrated AI carries the greatest consequence. We introduce B-Clean as a principled solution for fair, vision-grounded evaluation of multimodal AI systems.

In other words:

Simple machine learning classifiers for medical imaging are really effective. We love them, they're great.