[dodo53]

it feels a bit like: damn those scientists/engineers trying to solve complex problems with a one one-size-fits-all analytical/reductionist toolbag; some problems are really hard! What we need is to use our one-size-fits-all Structured Dialogue Design!

[Confusion]

I don't think that's what he is arguing. The problem isn't that scientists aren't solving problems well enough: the problem is that those solutions don't get technologized, combined and put to use, through lack of communication, cooperation and coordination between scientists, other scientists from other sciences, industry, politicians and some in between.

A staggering amount of knowledge is being produced and most of it sits gathering dust. We've already reached the point where many new inventions/discoveries turn out to be rediscoveries of something already conceived of, or even thoroughly figured out, in the sixties. This happens all across the sciences, because there is too much relevant information for a single person, or even a single university department, to know of. We need tools to help with that.

My God, as I'm writing this I realize this is the most powerfully insightful thing I've read this year!

[derefr]

To put it another way: scientific positivism allows one person, working alone in a laboratory, to discover something fundamentally new, by simply ratcheting along in tiny increments from what has already been discovered. This can be good—something new inevitably gets discovered over time—but it also has a fundamental weakness, in that most scientists toil in obscurity with the public never understanding the relevance of their work, and thus never applying it outside their original domain.

I'm not sure the output of scientific positivism can be well-indexed, to the point where anyone can find all the relevant pre-existing work they could build upon in their own: as there is no one big point where something new is discovered, but rather a lot of little facts and confirmed/refuted hypotheses that snowball until a meta-analysis of several studies can actually say something for certain, there's no one thing for a scientific expert system/communications tool to return to you. I expect, to have such a thing at scale, we'd need an actual human-level AI that "read" and understood the significance of every study in every field, and could see all the cross-correlations.

Before the advent of scientific positivism, though, we had a sort of ritualized science, where we would get large groups of people all studying one thing or another, out looking for proof or disproof of whatever particular thing they currently believed, leaping around wildly in hypothesis-space rather than just edging forward and taking whatever facts came along. There would be scientific belief "movements," in the same way that there are artistic "movements." Because it was ritualized, science was able to be made an entertaining talk of at the time, similar to celebrity gossip—everyone would have their own opinion on whether the currently-researched belief held true or not, and would debate it constantly, increasing public awareness of the subject; a "named hero" scientist would later come with a sweeping experiment and prove one or the other group right, and would be heralded by that group and have some unit of measure, chemical element, or heavenly body named after them.

Ritualized science didn't necessarily advance human knowledge "as a whole" very quickly—the iterative assembly-line process we have now seems to work quite well for that—but it did seem to get each new scientific fact thoroughly embedded in the public consciousness, because of what is basically good social game design. Perhaps we need some more of that, some hybrid model where scientists can still be "heroes" with "rivals" in the public eye, entertaining and informing in conjunction and raised up with social status, rather than simply workers for government grants raised up only with citations in journals?

As a screwy tangent: perhaps this could even be a facade on top of current science, a sort of staggered release of scientific knowledge in high-assurance bursts, with the rest of the "development" going on in some scientific "closed alpha" where the public wouldn't be constantly bombarded with overzealous summations every time the ratchet was turned (basically a justification for Yudkowsky's "Bayesian Conspiracy.") When a scientific hypothesis was made theory, it would be handed off on stone tablets to a researcher well-trained in rhetoric with a nice-sounding last name, and they would become, to the public, "the one who discovered the theory of X." Only the conspiracy would know that it was the work of thousands.