Bay 12 Games Forum

Yeah, because it didn't. He got where he got, and then he stopped doing real medicine. This sort of thing is part of why dentistry and especially orthodontics is the mess that it is: because we don't vet proposed procedures with enough rigor. The solution is not to welcome more insufficiently-researched pseudomedicine. The solution is to actually do the longitudinal studies required to establish how to predict long-term outcomes and set standards of care based on that as we refine our understanding, and keep doing that continuously.

As to your larger point, you're absolutely right: it's absolutely a reflexive dismissal, because that's required of scientific inquiry. It's part of the fundamental tradeoff surrounding the practice of science: findings can be helpful even if they're wrong so long as we follow correct procedure, but they're worthless even if they're right unless and until we do. We can't do anything with them; there's necessary metadata for us to integrate new information into our existing understanding in a rigorous way that anecdotes just don't have. The rest of the process from grants to publications is built on that, and with good reason: this is the best tool we have to believe as many true things and as few false things as possible, but it's very fragile and hard to tell when it breaks.

If I had come across this idea on a review board or something as a theoretical possibility, I'd certainly be skeptical, but it'd have the necessary surrounding information for the proposal to be evaluable on its merits, and I certainly wouldn't be dismissive. (I'd also be deeply confused as to why it got handed to a computational biologist for review, but it's not impossible.) It's like the difference between the statements "pi is less than two" and "pi is green." The first one is wrong, and we can prove it. The second one is pointless because we can't; the only possible answer is "come back when you have something scientific."  Going before the public and saying "maybe pi is green, I don't know, I'm just asking questions" is not something anyone with any credibility should do.

Yeah, it's a fairly classic conspiracy theory -- and, with apologies to smjjames for the slight derail, it's illustrative of why conspiracies of this type work and how they embed themselves into politics, so as a case study in modern American political dysfunction it has some merit here.

At the core of every conspiracy theory there's a gap, if you will, a hole in the shape of a demonstrably probative data set that would demonstrate the falsifiable claims of the conspiracy to be true and that no one in the conspiracy actually has. The rest of the conspiracy canon consists of a series of tactics to encourage you to ignore, dismiss, or minimize any concerns about that lack of data. As seen here, generating a narrative about how persecuted the theorists are is an extremely common way to combine forbidden fruit with an entertaining story. You want to know What They Don't Want You to Know, and everyone loves a good underdog story, so hey, two birds with one stone. It also has the effect of encouraging you to identify with the conspiracy and the conspirators simply by listening to them. You are, after all, one of the insiders now, and all the sheeple just don't know what you know. So, how valuable would you prefer to believe this knowledge is -- and by extension, how much better than the sheeple you are for knowing it to be true? And thus does it become more preferable to believe, and the belief that all contrary evidence is fake naturally follows.

In reality, this sort of claim -- cheap, probably harmless, and simple -- is well within the wheelhouse and budget of a number of government-funded science initiatives across the country. A grant for this would be really easy to sell, too, just on a public health basis, and the tricks to get something like this through an IRB aren't hard. Sure, he can't charge for it, but how much better if he could compensate them, like in a proper study? He could even pay himself a salary for it out of the grant, including just for analyzing the dental X-rays. Then he could write a real paper, stick it in a proper peer-reviewed journal, and actually have accomplished something worthwhile. That is not hard for anyone with any business claiming to know anything about medicine.

Instead, he's done the standard pseudoscience end-around run and presented anecdotal "findings" to the public without putting them through peer review. That's really dangerous; every time we let a doctor present unreviewed bullshit to the public, we make it easier for them to ignore real medical and scientific expertise, and people suffer for it. It is also possible, as Mew has demonstrated, to put together enough of a scientific facade to fool laypeople without a shred of real proof, and sucker a lot of people into trying a lot of unhelpful things. Loudly and utterly destroying the career and credibility of mistakes like him is not only just, it is literally the least we can do if we want science to have any meaning at all.

it's basically broscience of health. But the idea that little shit you do or don't do adds up over time is a fundamental rule of life. And stating that seems like a weird thing to try and revoke someone's license over.

No, peddling broscience is a totally valid thing to destroy a man's career over, because how the little shit adds up matters immensely.

How is that better than just shaking the whole thing, though, particularly when you may well have dice piled on top of each other?

Is flash player even still a thing anymore?

Last stable release was six days ago, so yes, I suppose it is.

No, Max. I think they do mean far enough. Look up near-field radiative heat transfer. It can increase radiative heat transfer orders of magnitude above what one can get from black body radiation, providing the two bodies are close enough (on the order of hundred of nanometers or less).

In a broad sense, it's similar to Förster resonance energy transfer in that regard. People try to think of photon wavelength as size, but magnitude is a closer equivalent, and the difference shows up when you start looking at how they move across very small scales.

And I deliberately did list MKULTRA in an attempt at steelmanning, because sometimes the tinfoil hat crowd are right. The problem is not what conspiracy theorists believe per se, but rather the methods by which they come to believe it, because those methods are not the most useful in understanding the world. In fact, they're worse than useless: they're actively toxic to having a framework for processing information that leads to believing as many demonstrably true things and as few demonstrably false things as possible.

The focus on plausibility over provability is a prime example. For one thing, the universe is provably implausible, but more importantly, your understanding of what is plausible is internal and subject to shifts in how you think. If you always think they are out to get you, that will feel more plausible to you, and you will start to see more evidence for it even if you have to bend it to fit. Moreover, other assertions with similar traits will also seem more plausible, which is how crank magnetism happens. Go far enough, and it becomes an implicit assumption that "they" planted all the contradictory evidence and hid all the proof and bought all the experts, at which point one can believe anything. It gets easier to operate in the realm of "isn't that interesting" and "what would we expect to see? exactly this" without all the fiddly checks and controls and other considerations, and it eventually soaks into your identity. Spend long enough thinking conspiratorially, and eventually all the experts have to be wrong, or else you're a fool who has wasted a ton of time on nonsense.

"Hard evidence" is great, but without a rigorous framework for defining for what it is supposed to be evidence and what other evidence would influence the probability of that theory being useful, it's just information, and can fit whatever pattern you want. "Critical thinking" is a fine idea, but without logical clarity it devolves into just casting aspersions on the credentials and motivations of anyone who disagrees with you. What our culture is evolving to do is to find ways of making reflexive, emotion-driven distrust sound smart to other people who don't know better, and conspiratorial thinking is leading the charge on that.

It sounds absurd, and I'm sure some smartass is going to put this in the quotes thread, but just because something happens to be true is not sufficient reason to believe it. Unfalsifiable things can be apparently true, as can things that aren't quite supported by the data but are concordant with it, and it's that latter that conspiracy theories evolve to run on. Believe in things only to the extent to which efforts to falsify them have returned confirmatory evidence, and then only with precision that borders on paranoia, because that's the last point at which one can stop before the slow slide into believing absolute nonsense.

Not quite, no. Zeteticism is literally this:

That [...] data could then be used to make [...] models

There's no questions involved at all, just taking a bunch of data and building a mechanism out of it. To use your ball example, it's seeing that you can hit the ball in a certain direction with your club and deciding why that happens based solely on the congruence between your theory and the data you used to make it. Alternatively, it's seeing that the world looks flat to you and deciding that it is. That's why it is awful science: you can make any theory you want right down to "God created ad hoc physics to produce exactly these data" and have it accord with 100% of observation, so there's no predictive value.

Empiricism doesn't let you do that. You build a model (which is really just a mechanism with defined input and output parameter sets) based on theory and then identify a data set that could falsify it. Then you go do the experiments necessary to get that data and either disprove or fail to disprove the most relevant null hypothesis or equivalent. By the way, this:
 

Science is about testing observations. 

Is also wrong for this reason. We don't test observations. We test predictions.

There's also a more minor point to be made that those predictions are not always about causes. Predictions are a logical expression of a pattern, and patterns can exist  without a time component. It's totally valid to predict that if set X has Y traits it will also have Z traits because of some mechanism and go off and go look for an X that has Y but not Z or otherwise try to disprove that assertion.  Y and Z can even have spacelike separation and the process still works. Sure, there's implicitly a cause somewhere, but mechanisms need not always address it to be falsifiable.

This brings us back to Naturegirl's original suggestion, which was a neural activity-based fishing expedition. We do these all the time, especially in bioinformatics, but not to generate models (and, therefore, nothing of interest to behavioral scientists, at least in the way you've proposed) It's like GWAS in that way. The point is that, since an AI is built to classify an input set into an output set (sort of) and her original post didn't include an output set, it's not the tool to use to pull out trends. We have better tools for doing that, PCA being one of them.

The original idea was to get a whole bunch of neural data together and learn from it. That's not inherently unscientific, but there are very specific ways in which it can be a useful part of the scientific process and AI can't really help with them. What we do with a giant pile of data, be it genomic or neural or proteomic or anything else, is fish through it for correlations. You can then go look through existing models to find ones your data may be able to disprove, but you don't go hare off and propose one ex nihilo just from fishing. At most, it's a source of very specific preliminary data. Useful, but not as a deliverable.