aqsalose

Many of the issues sound like issues coming from using improvised civilian hobbyist tech and doctrine being in its infancy.

If current FPV drones are bit lackluster, it doesn't preclude 'next generation' that are purposefully developed for military use won't be useful. Also it sounds like the designation of "FPV drone" is specific to particular family of drones specific in current day and time, which may be something quite else next year. Like, obviously the next stage is a FPV drone with some capabilities of "reusable" drone or loitering munition author complains of (capability to hover easily)? Or "reusable" drone with FPV camera?

I think Third Republic France is a more apt comparison. Political fights about religion and content of education, check. Diverging media landscape aligned with party political identity and ideology, check. Major changes to civil service personnel after consequential elections (1879-1884), check.

Sounds sensible, bu the major unasked question it avoids is, was the current funding and organization structure of science in place when the past scientific achievements were achieved.

the impression I get from anecdotes and remarks is that pre-1990s, university departments used to be the major scientific social institution, providing organization where the science was done, with feedback cycle measured in careers. Faculty members would socialize and collaborate or compete with other members. Most of the scientific norms were social, possible because the stakes were low (measured in citations, influence and prestige only).

It is quite unlike current system centered on research groups formed around PIs and their research groups, an machine optimized for gathering temporary funding for non-tenured staff so that they can produce publications and 'network', using all that to gather more funding before the previous runs out. No wonder the social norms like "don't falsify evidence; publish when you have true and correct results; write and publish your true opinions; don't participate in citation laundering circles" can't last. Possibility of failure is much frequent (every grant cycle), environment is highly competitive in a way that you get only few shots at scientific career or you are out.

Yeah, the title is a bit hyperbolic. I have not used selection methods that much, but not too surprising they would have similar results to LASSO as selection or predictive method for people who think of it in terms of "feature development".

The distaste for step-wise selection comes from its typical use. If one reads Harrell's complaints quoted in the blog post carefully, quite many of them are less about the selection method but what analyst does with it, namely, interpretation of inferential statistics. When you see step-wise in the wild, practitioner often has used step-wise or other selection method and then reports the usual test-statistics and p-values for the final fitted model ... that are derived with assumptions that don't usually take into account the selection steps. It is quite unfortunate in fields where people put lot of faith in coefficient estimates, p-values and Wald confidence intervals when writing conclusions of their paper.

With LASSO and its cousins, the standard packages and literature strongly encourage the user to focus on predictions and run cross-validation right from the beginning.

From abstract (article is paywalled)

Although modest bivariate associations were detected with educational attainment (r = .17) and body mass index (r = −.17), almost all regression-adjusted coefficients were nonsignificant. No clear pattern of moderation was detected between delay of gratification and either socioeconomic status or sex. Results indicate that Marshmallow Test performance does not reliably predict adult outcomes.

I guess the question is whether the covariates that were adjusted for in the regression are true confounders and not, say, something caused by ability to delay gratification.

Well. You have to exist, which means you compete, which might mean you grow.

Why growth? At some point you would eventually hit perfect saturation anyway, the steady state where everyone already is buying your product to the extent anyone can buy it. I get that losing business is bad, and it's better to "overcorrect" to growth, but as long as you compete enough to keep approximately same market share against other competitors, selling inflation adjusted $30 buckets of bricks to each generation of kids with profit sounds like perfectly good business. Owner of the business would receive steady income selling the inflation adjusted $30 buckets.

I'd imagine you'd hit problems when the buckets of bricks you are selling are ~eternal and number of kids is no longer growing, so nobody needs new ones.

"it honestly comes from a place of ignorance, and I say that as basically a layman myself"

Here is an added complication: succinct technical communication can be efficient when communicating to peers who work on the exactly same domain, similar problems as you, and want digest your main ideas quickly.

On the other hand, for any particular paper, the size of the audience to whom it is directly relevant and addressed to can be small. The size of the audience who got to reading it anyway may be vast. (Maybe I am reading your paper because someone cited a method paper that in lieu of a proof or explanation writes just two words and citation to your paper. Maybe I am a freshly minted new student reading it for my first seminar. Maybe I am from a neighboring field and trying to understand what is happening in yours. Maybe I tried to find what people have already done with particular idea I just had and search engine gave your paper. And so on.)

During my (admittedly lackluster) academic career I recall spending much more time trying to read and understand papers that were not addressed to me than papers that were and where I enjoyed the succinct style that avoids details and present the results. (Maybe it is just an idiosyncratic trust issue on my part, because I am often skeptical of stated results and their interpretation, finding the methods more interesting). But that is not all.

I also noticed that genuine misunderstandings coming from "brief" communication of technical "details" were quite common; two different researches would state they "applied method X to avoid Y/seek Z[citation]" in exactly so many and almost exactly same words, where X,Y and Z were complicated technical terms, yet the authors would have quite different opinion what the meaning of those words were and what would be the intended reading and how and why X should be implemented.

In conclusion, I think many a scientific field would benefit from a style where authors were expected to clearly explain what they did and why (as clearly as possible).

However, I wouldn't then use version control software like Git for versioning analysis objects, as it is designed for text file source control and diffs.

(How one does a diff of a data object look like? If there is a natural text format to save it in, it still is usually quite messy, and Git doesn't really like Gb sized csvs.)

My preferred workflow is to version the source files in Git and store the associated data objects in a separate archive directory with meaningful name and the hash of commit of generating code as metadata attribute.

Now if you had a version control "IDE" software that would render changes in figures and other blobs nicely, then it would make sense to build a workflow around it.

In my experience, the trick is to move them without disassembly or with minimal disassembly (removing only moving parts like shelves that are planned to be removed) like any other furniture. Nothing weird with that: Most traditional furniture items made by a carpenter would be equally incompatible with disassembly.

I guess it's referencing the fact that education today is largely about having textbooks shoved in front of you until you're able to recite enough of it.

I would argue contrariwise, the education today is bad because the textbooks are devoid of content and nobody can recite any of the little they have. For my parents' generation it was not unexceptional for people to cite poems from memory. I have bunch of their middle school books, and it appears they read more and longer texts for middle school than some university students today. During my grandfather's time kids were expected to recite a chapters of textbooks aloud in front of class, and he also remembered good bunch chunks from the Bible.

Compared to that, fill-in textbooks we used when I was in school seem a bit underwhelming -- and I am in my 30s. Kids today use e-learning environment (makes direct comparisons difficult).

Today, very few people appear to read anything, let alone books, even fewer remembers anything. Thus conversations about anything factual seem often pointless. But it is not like one can blame anyone for that, it only makes sense: Why truly should I remember anything when I can flip out a smartphone and hit query to a search engine? But people reading the same Wikipedia article or repeating the same news cycle talking points at each other makes for a boring conversation.

I suspect interacting with the real physical world and its realities and to realize one can affect it would be good, no matter what career they'd pick later. Picking a career in software development has been a good choice for bright kids for several decades now. In the long term view, the past is full of "good career choices for bright kids" that at some point no longer were not.

The physical world is not going anywhere.

But I'm also a musician/artist and so I find some of these conversations odd. The problem with them I see is that they are oversimplified. To get better at drawing I often copy other works. Or I'll play a piece exactly as intended. Then I get more advanced and learn a style of someone I admire and appreciate. Then after that comes my own flair.

So I ask, what is different between me doing it and a machine?

You are a human. If you practice art as a hobby you can feel pleasure doing it, or you can get informal value out of the practice (there is social value in showing and sharing hobbies and works with friends). One could try to formalize that value and make a profession out of it, get livelihood selling it.

When all that "machinery" to (learn to) produce artistic works was sitting inside human skulls and difficult to train, the benefits befell on the humans.

When it is a machine that can easily and cheaply automate ... the benefits are due to the owner of machine.

Now, I don't personally know if the genie can be put back into bottle with any legal framework that wouldn't be monstrous in some other way. However, ethically it is quite clear to me there is a possibility the artists / illustrators are going to get a very bad deal out of this, which could be a moral wrong. This would be a reason to think up the legal and conceptual framework that tries to make it not ... as wrong as it could be.

It could be that we end up with human art as a prestige good (which it already is). That wouldn't be nice, because of power law dynamics of popularity already benefit very few prestige artists. So it could get worse. But could we end up with a Wall-E world where there are no reason for anyone to learn to draw any well? When a kid asks "draw me a rabbit", they won't ask any of the humans around, they ask the machine. The machine can produce a much more prettier rabbit, immediately and tailored to their taste.

But if I train my own neural network inside my skull using some artist's style, that's ok?

How well the network inside your skull can manipulate your limbs to reproduce good-quality work in some artist's style?

Our current framework for thinking about "fair use", "copyright", "trademark" and similar were thought about into existence during an era when the options for "network inside the skull" were to laboriously learn a skill to draw or learn how to use a machine like printing press/photocopier that produces exact copies.

Availability of a machine that automates previously hand-made things much more cheaply or is much more powerful often requires rethinking those concepts.

If I copy a book putting ink on paper letter by letter manually, that's ok, think of those monks in monasteries who do that all the time. And Mr Gutenberg's machine just makes that ink-on-paper process more efficient...

Sure, extroverted students have some advantage.

On the other hand, extroverted people have similar advantage in the real life. I myself am quite happy for every lesson where I was pushed to practice people-facing skills (presentations, demonstrations, etc). Even an introverted person can learn to talk about topic knowledgeably if they know it -- which often is valuable confidence-building experience to have. Despite the introversion, one can do it!

If the professor - lecturer administering the test is any good, empty rhetoric won't help too much. If they are lazy, students one can try to give "answers" without showing what they don't know in written exams, too.

Hereditary monarchy is only one kind of monarchy.

Early on, the Swedish king was elected at the Stones of Mora. The Holy Roman Emperor was nominally elected by prince-electors (who most of the time elected a Habsburg).

And even withing a hereditary framework, there are other alternatives to retirement in addition to outright abdication. An elderly monarch could for all intents and purposes retire and a let the crown prince (and I suppose in current British succession order, crown princess) rule, appointing them as a co-ruler.

Coincidentally just yesterday there was a big news article in the largest daily newspaper about the problems teachers have with uncooperative parents. One memorable case was of the parents calling the teacher and informing them that the parents have agreed with their kid is exempt from reading books. In another, during a disagreement with a teacher, kid called their parent, put the parent on speaker, who then proceeded disparage the teacher in very low language in front of the rest of class.

The article is here https://www.hs.fi/kotimaa/art-2000009001096.html : it is in Finnish, but Google Translate manages to make sense of it in English.

The social standing and respect teachers have varies a lot in different social spheres, but it is certainly not as high and universal as doctors.

To piggyback on the OPs question, I for one think the part in parenthesis is actually most important:

(Data cleaning and management should also be learned)

There are many students and graduates who either didn't want to do research in the first place or didn't get that research grant or position and looking to get employed in private sector with their degree. Many universities and colleges have now also retooled some of their statistics degrees as dedicated "data science" curriculum who either know basics of ML/DL or have the prerequisite background to learn quickly.

However, in my experience (I am extrapolating from my own past job search experiences) while "understanding theory behind the algorithms" counts still for something, it is much less than one would think. Familiarity with the software technologies and practical implementation is what counts much more. This includes not only "data management", a phrase which makes it sound like the data simply exists somewhere and only needs to be managed (not unlike a Kaggle competition), but also the data pipeline management from generation/collection to analysis and communication of the results, and deploying the software the implements it all, and so on. I suppose (never been on that end of the interview table) given any two candidates to interview, it is very difficult to evaluate how deeply one understands theory of some algorithm compared to other if they both demonstrate some basic understanding (and what is the practical use of possible difference in insight from such differential, anyway?). Likewise, I assume it is somewhat easier to gauge whether someone seems to able start delivering results or contributing to their on-going work quickly if they have the relevant technical skills and/or domain knowledge.

he’s a villain with good PR

In the context of Watchmen, he isn't the only character who fits the description.

Here's an illustrative thought experiment: imagine you have a time machine. Now pick a worker at random from some time and place in the past 5 centuries, and carry them forward by 30 years. will they be able to earn a living?

I don't find Stross' thought experiment very convincing. One doesn't need to imagine time travel. A CS graduate from the 1990s who didn't timetravel directly to 2020s but got there regular way and didn't do anything to update their skills during those years would find themselves with equal difficulties in job market than the time-traveler. (edit: Or worse difficulties.) That is why it is a good idea to continuously develop ones skills.

However, on much shorter timescales, say, 5 years, one can make a reasonable guess what kind of degree is more likely to result in gainful employment after graduation than other. A degree doesn't equip one for a job, but a useful one results in one enough understanding of some field that one obtains, should I say, a fighting chance or more to equip oneself for a job related to the field. And having a job often results in better chances to learn more and further equip oneself for one's next job.

Then in a later part of the blog post Stross argues that as arts sector is today very profitable to the UK, it warrants continued government support for arts education. This strikes me a bit inconsistent with his earlier claim that prediction of the future need for skilled jobs from the current state is impossible.

A better argument would be that it is possible that arts are going to be more useful than STEM in the future, and it would be unwise to cease arts degrees. It has certain ring of truth to it. However, I came under impression that Stross is in favor of keeping the number of arts degrees at the same level or increasing their amount, but if we take "impossibility of prediction" seriously, there is no telling the current amount -- or higher amount, or lower amount -- of arts degrees awarded is any better in 30 years either.

I am not sure the education allocation is best done by the government giving commands how many artists and engineers are needed to be trained (or given subsidies to be trained, or whatever). If that choice is for each individual to decide without government planners intervening, they at least have some idea of their personal talents, wishes, and circumstances than either Rishi Sunak or Charlie Stross.

Britain's descent from the powerhouse of world-changing ideas to one giant housing estate and Tesco superstore is almost complete.

In my limited foreigner's understanding, Britain was "the powerhouse of world-changing ideas" during period that has fuzzy limits but starts maybe around Newton and continues until maybe Turing -- but after WW2, what was left the powerhouse was certainly eclipsed by the US, and after the 1980s, Asiancountries.

Maybe one can stretch it bit further after the WW2 if one thinks that popular culture production like Beatles is a worthwhile substitute. [1]

How was the education in Britain organized during that era?

[1] I don't; AFAIK income distribution in popular culture production is very winner-takes-all top-heavy, much worse than the software income distribution often denigrated as favoring the 10X developers. 10X coders may make much more than a marginal software developer (I am imagining soon-to-graduate CS student who would-be entry-level dev who has difficulties getting the first interview), but I believe it easier for the marginal software developer land a software job that pays the bills than for a marginal would-be musician to land a music job that pays the bills.

Then these real world problems don’t actually warrant deep learning ?

It is an important lesson to be communicated.

I'd like to present a conjecture: everyone thinks their data is big until they have worked on much larger dataset. ("We have 10k samples, it is quite big!" -> "We have 1m data records, is quite big!" -> "Our process outputs that much per day")

There is not definite answer. Reviewing a paper is difficult, been for people who do it a lot. I think I got better at when I was a part of research group and wrote papers myself, collaborated, and observed others writing papers: it is good to be remember that there is a substantial, arduous process of which the published paper is only the output. One difficulty is that it doesn't necessarily translate to different domains: the process can be quite different in different fields.

I would imagine that those who actually spent some of their childhood in a war zone would have a very different perspective.

Not sure if it would imply less war. Around the Napoleonic wars, many officers started their careers as teenagers or pre-teens. Nelson joined the navy as about 12-year old, which wasn't unexceptional age during the Napoleonic times. Napoleon was 10 when he was enrolled in a military academy. He was admitted to Ecole Militaire around 15 years old, graduated in one year, and got a commission as 2nd lieutenant.

True, Orwell has a worldview (though it evolves over time) and it connects his thoughts if you pick up a collection of his essays and study them systematically. But one does not need to agree with everything to find agreement with some particular elements and thoughts.

Depends on the crime. For serious crimes I think it is quite common?

Like, Finnish law claims jurisdiction over all crimes committed against or by Finnish citizens committed abroad, no matter the location or acts legality per local law, if the crime could result in prison term of 6 months or longer according to the criminal code of Finland. We also claim jurisdiction over some particularly heinous crimes committed abroad no matter the nationality of victim or perpetrator. I think it is very rare to see it applied, because crimes according to Finnish law are often crimes abroad, too (I mean, the location of country often takes precedence?). And I don't think our laws grant our authorities power to act outside territory of Finland (like the US sometimes does).

But if a Finnish citizen traveled to country X, did something bad enough to mandate a long prison sentence, came back to Finland, was found, and for one reason or other were not extradited back to X for trial there, they would be tried here.

Similarly, in my work with industry partners, some of the most rigorous methodological discussions I've ever had have been with them.

I am willing to believe this depends a lot on which industry you are working in, what are you selling, and who are your clients.

The academia is characterized by world's top experts in a narrow niche investigating speculative problems few people have any idea of. More often than not, that research turns out to be a dead end, "wasting" years of work.

I believe this is very much dependent on where and with whom you are working with. By definition, not everyone is a top expert. Even rarer to publish a top paper. Sometimes entering a narrow niche field makes it possible to work in an insulated silo where niche's favorite problem statements and research programs can escape critique from experts of other disciplines.

As a practical, though not quite disastrous example, I did an applied maths MSc with focus on ML and data science, and then spent some time in bioinformatics oriented data science grad program. It was only after I entered the pharma industry that I found a field where it was expected to have serious interest in doing ones best with causal inference while acknowledging its limitations ... with methods which apparently have been bread and butter of econometrics and maybe some biostatistics for decades. On the other side of fence, typically only people exposed to particular statistics textbook or ML fields are interested in running LOO/cross-validation model validation checks for their model fits. I see some more communication between the disciplines could absolutely improve the work of everyone involved. And these are big fields. Small niche fields with niche problems where everyone publishes in a niche journal can become worse.

Sounds like, if you want a capable materials, mechanical, chemical and electrical engineer to write your pull requests, you'd need to pay them a salary they request. (Them in plural, because it is unlikely to find a single individual good at everything.)

Software people like to say that software engineers is super complex and difficult. On the other hand, an enthusiast occasionally makes great FOSS contribution by filing a pull request. For some reason, that is?[1] quite rare in many other forms of engineering. If it is only because of capital cost differences of building things in physical world vs building in software world (which affects stuff like learning by experimentation), maybe we should acknowledge they are a part of reason why building things in physical world is complex and difficult.

[1] Or looks rare, I may be mistaken.

Only thing I like to buy from Amazon are ... books.

Theoretical parts of computer science is connected to discrete mathematics, sure. But that is only a subfield of mathematics and mostly happens already at CS departments, so you'd get a CS degree anyway.

It is also possible that aptitude for math is related to aptitude in software engineering.

However: The mathematics content of 90%+ of mathematics degrees awarded is fully irrelevant to 95%+ of software development tasks. And when that 5% task needs that some kind special mathematical insight, the people who want that task done are going to get the top professional they can find for it. Maybe the prospective math student is going to be that professional, but I don't recommend planning a career for it.

I am not saying there isn't work where some math is useful but the most commonly used applied stuff ... say, linear algebra ... is typically covered in a respectable engineering program; degree in mathematics would be superfluous. Proving theoretical properties of Hilbert spaces or measurable sets or bifurcations of dynamic systems or advances in differentiable topology or fascinating behavior of cellular automata or whatever is going to be gigantic waste of your time if you won't use it later in your career or you don't find it intrinsic motivation in itself.