Agree, the NPS aggregation is mostly BS, but getting a signal from median/mean, e.g. by comparing trends across some dimension, and reading complimentary feedback is quite useful.
No need to constrain yourself to FOL! Why not use some weird non-classical logic? ;-) Joking aside, rules engines typically use some form of defeasible reasoning, i.e., some not fully formalized non-monotonic logic, and it seems to be a pointless task to formalize everything. Still, I think that even from a practical perspective, looking at them through a formal lens can potentially help improve them, see this paper for an example: http://www.inf.unibz.it/~calvanese/papers/calv-etal-IS-2018.....
Generally (with NFTs), the supposed 1-1 mapping of art piece (or other artifact) and token happens on meta-level, right? So for sure, an artist can issue many tokens that represent the same art piece (and don't tell me the token is a hash of the image or whatever; it still can be sold on a different technology, or noise can be added to the image, et cetera). Because of this object-/meta-level mix up, it seems that NFTs guarantee pretty much nothing, they are just great at facilitating speculation, because the whole thing takes so little effort...
Judging the quality of something primarily based on its prevalence is a questionable idea as well (no offense). The better argument could be that some of these languages are the result of violent conquest, the subjugation of (indigenous) women and children, and possibly also of the scarcity of meaningful communication between genders.
As somebody who's about to finish a CS dissertation, I sometimes think the best way to do CS research is to get a somewhat interesting software engineering job that allows for a 4-day week and to do research as a side project. Most likely, this gives one more research time than staying in academia (with the exception of the first Post Doc, perhaps).
Edit: This works for most of CS, but not for most of Physics, I assume.
Still, one also must acknowledge that the hammer of using simple formal models to describe human decision-making has led to a crisis as well (in microeconomic theory). So it's best to do it like Tversky and Kahneman and build formal models that are then empirically tested.
Does anyone here have thoughts on JOSS? I've reviewed for them once, had the impression the editors take their job seriously, and think I'll review again in the future. The review approach that the journal facilitates has a strong focus on engineering aspects, i.e. it addresses a weakness of other venues, where it often does not matter how messy, unstable, and poorly documented the code is (or even if it compiles). On the other hand, the JOSS reviewers are typically not experts on the problem that the software is solving.
Mostly agree, but "The few geniuses stay and do research" is probably not true. I don't like the term "genius", but many of the best and brightest are put off by academic elitism and bureaucracy and leave, while at least some not particularly gifted individuals stay because they are more comfortable to play the academic game. One could even go so far to say that academia needs both of these types, but arguably, there are too many "game players" at the moment.
(And of course, this is an oversimplification, and most academics try to play the game AND do useful research, but to different extents.)
I disagree. We can't argue with formal precision here (unfortunately) but we should be permitted to try analyzing, anyway. Roughly speaking, the "objective" observations made in traditional mainstream news outlets tend to be correct, but the more an article strays from those objective observations (and becomes opinion, value judgement), the more contestable the inferences that the writers draw from these objective observations becomes (and the more divisive the article tends to become, in particular in the US).
Agreed. One (anecdotal) observation I've made is that bullshit meetings are harder to bear and to defend when everyone's online because people can protect themselves. It's easier to not join if nobody can approach you physically. It's easier to disengage. IMHO, if the culture is reasonably good, the organization and most of its employees profit from WFH. But if there are serious culture issues, it's probably really hard to recover.
What we consider "welfare" is typically not handed out based on merit, but based on need (and, from a global perspective, also based on privilege). The idea of a welfare state is not meritocratic. Many societies strive to be meritocratic with egalitarian constraints: anyone should be able to have a decent life (egalitarian), but to have a great life, one needs to contribute (meritocratic). Interestingly, inheritance neither fits the egalitarian nor the meritocratic perspective.
I think when looking at least at large scale applications in the context of games (and not just super-expensive showcases), like in the Stockfish chess engine, we see that it's not primarily about depth, it's about architecture design. Reference: start here https://stockfishchess.org/blog/2021/stockfish-14/ and go down the rabbit hole...
In the broadly useful domain of recommender systems (which typically make use of some type of RL-like feedback loop, but can be implemented using simple clustering approaches), at least in 2019, neural network-based approaches didn't seem to fair too well, either: https://arxiv.org/pdf/1907.06902.pdf (arXiv pre-print, but this is an award-winning paper).
Since then, it seems that researchers are moving away from getting deeper and deeper (the low-hanging fruit), and try to be more creative instead: new architectures, combining symbolic (logic-based) and sub-symbolic (ML-based) AI, etc.
I think that in some highly developed societies, and in particular in Scandinavia, this trend is actually good for endangered languages. The lingua franca of Scandinavia is English. The region, which is comparable in population size to a large German federal state or a medium-ish US state, has four different majority languages (Danish, Swedish, Bokmål, Nynorsk, all mutually intelligible, which increases the acceptance of imperfect language) and dialects play an important role for local identity. Of course, as a Scandinavian one has to learn one of these four languages to some extent, but the growing importance of English -- as a consequence of almost universal English proficiency and an influx of skilled foreigners not only to the capitals but also to the periphery -- narrows the gap between these languages and minority languages like Elfdalian and Sámi. As a consequence, the appreciation of the general population of language as a form of local identity also grows.
Can you elaborate? I doubt your claim because "how much profit" a business makes is tied to the legal/financial system(s) the business operates in. To me it looks like we have the "object level vs. meta level" fallacy that is so typical for crypto enthusiasts: in the end, what governs our society is not crypto tech (object level), but meta level institutions. And even if these institutions were to agree that, for example, a smart contract is a legal contract, they could still reverse this decisions and hence move the agreement back to the meta level.
Doesn't it make things like this all the more meaningless?
We know that "unlimited vacation time" can, given a competitive company culture mean "essentially no vacation time".
First, isn't this then the same with wellbeing days(sure, it's the whole company that's supposed to be off, but who knows how it will actually work in practice).
Second, wouldn't the first step be to give all employees 30 days of annual vacation and enforce that they actually take them? (This is what many organizations do in Europe...)
As I remember, at least as a narrative, it is clear that having speed limits reduces traffic jams, congestion and accidents, and will actually help get people faster from A to B. Of course, no speed limit is part of the "brand" Germany, but I guess this could be maintained by some work-around. Happy to revise my opinion when presented with evidence to the contrary, but to me speed limits are to Germany what gun control is to the US.
To generate new, generally useful algorithms, we need a different type of "AI", i.e. one that combines learning and formal verification. Because algorithm design is a cycle: come up with an algorithm, prove what it can or can't do, and repeat until you are happy with the formal properties. Software can help, but we can't automate the math, yet.
I think if the course instructors are more or less experienced contributors and put a focus on meta-aspects such as respectful communication (valuing maintainer time), this could work. Of course, this also means that at the moment, many schools do not have teachers with relevant expertise, I suppose...
He didn't clean it up when starting in the role, he damaged the (employer) brand he was expected to protect/develop as part of his role, he got re-assigned to a different role, which is arguably a reasonable consequence.
Yes, there is an association and a Head of Diversity should be smart enough to not generalize based on this association; it's failure to satisfy the core requirement of the professional role.
If you only look at the numbers it's the same story, but with some knowledge of the JavaScript ecosystem, it's arguably unreasonable to claim that babel is just a random, trivial package that is pulled in as an arbitrary dependency.
They have more than 20 million weekly downloads on npm: https://www.npmjs.com/package/@babel/core, i.e., they are probably used by at least hundreds of thousands, if not millions of developers (not to speak of end-users).
Of course, in the JavaScript ecosystem, the same may be true for a 20 LoC package that provides a trivial function implementation, but that's a different story.
I mostly agree, but many scientific communities are rather self-centered/inward-looking, and when it comes to technological innovation like programming languages, it could be a reasonable compromise to align slightly more with a "mainstream" ecosystem and slightly less with the academic community. Of course, ideally, all design choices are compared to the community's state of the art, but this may be impractical if it implies a comparison to tens of obscure languages.
At first glance, the paper seems to be written by competent people, but the 'related research' section is rather shallow. How is what the authors have developed more than "yet another language"?
I don't want to be too critical, though, it probably can't hurt to work on something new without dragging along all the baggage of the past. Still, it feels a bit strange from an academic perspective.
I wouldn't be too judgmental here. From the authors' and editors' perspective, the implications are clear because the paper is not marked as a review/survey. Sure, it's not a good choice, but I also think it's unreasonable to expect a title like "this study does not find evidence...".
I think the ban hits the right institution, but I'd reason the other way around: is it really the primary fault of the individual (arguably somewhat immature, considering the tone of the email) PhD Student? The problem in academia is not "bad apples", but problematic organizational culture and misaligned incentives.
My guess is: a journal that does not focus on studies of human behavior and whose editors are a) not aware of the ethical problems or b) happy to ignore ethics concerns if the publication is prone to receive much attention (which it is).