xaa

I went to a completely unranked school, not even a CS major. But took Programming I anyway. It was for freshmen, and was in C++.

IMO, for an intro language, you need either something close to the metal, so that students can work their way up the abstraction chain, or something very high-level with flexible constructs if it's desired to teach things from the algorithmic level.

I also think that any dynamically typed language is a bad choice for a first language. We're all used to holding the types in our heads, but beginning programmers don't even know what they are, and it's a really core construct.

Biologist here. Biologists don't try to predict whether a drug will have side effects before giving it to humans. They simply give it to mice. If we do have some strong reason to suspect it might be harmful to humans, we might do a primate trial, but usually not as those are expensive.

People do make statements like "Drug X has target Y, so its side effects should be limited to A, B, and C", but no one takes those claims seriously. It's too hard to predict what a drug will do, and in fact drugs are frequently used in practice without any knowledge of mechanism of action.

The point is that biology is fundamentally empirical here, because in complex systems it seems that rationalism/theory is not that useful.

> Economics is hard because you can't do repeated trials.

Yes. The problem is that they pretend they can predict results of interventions when they demonstrably cannot. Economists are simply used to justify policy decisions made on some other basis.

Yeah, I've been out of the Windows game for some years now, but my wife has a few (desktop and laptop), occasionally comes to me and says "it's slow. fix it." And indeed it is slow, even on moderate browser use -- things just hang. Specs are more than adequate.

I have to say (after halfheartedly looking into it a few times): "Um, sorry honey, I have no idea. Guess it's a Windows thing." She nods, and goes back to patiently waiting minutes for the browser to stop hanging! Seems that people using Windows just accept this kind of thing as a given, to the point they don't even notice it. (I've installed Linux for her a few times, but of course, it's too different, doesn't run Photoshop, etc...)

The trouble is that it is not hard to point out flaws in any system, government programs included.

Many Republicans have taken this to the logical extreme: they say government programs CANNOT work. In solidly red areas, they justify this by pointing at how their local school districts or whatever are failing badly. Why are they failing? In large part due to lack of funding.

So, low funding creates failure which is used to justify lower funding. It's circular. In cases where there is clear evidence that a socialized alternative can work, like healthcare, the evidence is simply ignored.

This is not to say that the government cannot waste money. Of course it can: walls, military spending, and so on. But your argument is dangerously close to the hypothetical Republican one. The only solution I can see is for taxpayers to BOTH be willing to pay taxes AND demand value for their money.

Very interesting paper. Two problems that jumped out at me are: 1) income is not COL-adjusted, and 2) more importantly, there is not a term in the regression for urbanization/population density.

It is well-known that urban areas lean Democratic, and richer states are more urban. Thus the phenomenon may be reflecting urbanization rather than state wealth.

Sure, fair enough. I was implicitly talking about 1980s to present (the period when the inequality began to spike), but obviously was not clear about that.

However, in the present day, I don't think it is just the 1% who are against a corporate tax increase. That would hit IRAs, pensions, and investments for those who can afford such things.

Where's the evidence that this solidarity ever existed? On the coasts at least, the people in this $300K range may talk a good talk about marginalized groups, etc, but the solidarity clearly ends at their pocketbooks.

Wealthier people skew Republican, and that's not just the 1%. OP's example of the school districts is another example. Still another is the fact that, if you look in any city in America, the wealthy (but not 1%er) areas have better roads, better parks, and better everything in terms of public infrastructure. I have a friend who works at the local power company who outright says that whenever there is a storm that takes out power, the wealthy areas come first for repairs, right after hospitals and police stations. Not, you know, the most populated areas or anything.

And it gets worse still, when the 20% try to redirect blame at the 1%. Because who are these 20%ers anyway? They're lawyers, managers, mid-level executives, financiers, maybe developers for banks, big ad firms, or VC-owned boondoggles. In short, they are the white-collar "hands" providing the services that allow the 1% to exert the level of control they do over the whole system.

To hear the 20% disclaim all responsibility for the state of inequality in the country is very similar to how the head of HR or IT tells you "well, I really can't do anything about anything, it's just Policy, set by our betters". But they are enabling all of it.

Yes... the statistics in this article are annoying (par for the course for BMJ, NEJM, and other medical journals). They use mixed-effects models and a lot of potential covariates (good) but then bin the variable of interest (BAD). I've seen a lot worse, but it's not optimal IMO.

This is basically an epidemiology paper, and the apparent statistician is an epidemiologist. As an interesting aside, I recently sat down for lunch with biostatistics grad students, and apparently they consider the epidemiology students/faculty/field to be incompetent wielders of "statistics-lite". I took it with a big grain of salt, but it does seem epidemiology is harder-hit by the replication crisis than most other biology fields.

> Moderate drinking some days but not others, perhaps your liver/brain do better because they get "total rest days", whereas light/moderate drinking every day may put them in a constant state of irritation or stress. Just like with workouts.

Actually my understanding is that the harm of drinking to the brain mostly does not occur during the drinking itself, but the withdrawal period after the depressant is cleared causes excitotoxicity for neurons. This is why quitting heavy drinking cold-turkey may not be best health-wise (but may be easier to do psychologically). Also, cycles of binging and abstinence cause your neurons to be increasingly vulnerable to damage via some kind of priming mechanism.

Personally, I've been trying to cut back, and I'm much more worried about my brain than my liver. It may be that the optimal pattern may vary by organ.

Sort of. My understanding is that it means "declaring your (virtuous) opposition to <bad thing X>, regardless of whether <current topic Y> is actually an instance of X or whether Y is actually harmful."

On the right, it might be: murder is bad, therefore I oppose abortion.

On the left: gay marriage is good, and opposition to it is homophobia, therefore Brandon Eich should be fired.

When the link between X and Y is too specious, it implies the speaker's primary goal is to show virtue by opposing X, rather than any real concern they have about Y.

> In my view critical thinking without constructive thinking is as big a problem as no critical thinking.

I disagree. I see critical thinking as a prerequisite for constructive thinking. Without the ability to identify problems, you can't offer solutions.

I and many fellow students in grad school went through exactly this evolution. First, you are taught to be critical and skeptical of everything. But at some point, you realize you can't get anything done in your own research if you are constantly skeptical of everything, so you learn how to find "good enough" solutions to tough problems.

So WRT society, I think it could use a healthy dose of critical thinking, because it suffers from the same problem. People can't identify good arguments, so they don't know they even need better ones.

I am aware the law changed, now; admittedly I didn't until reading other replies to my OP. And I am not familiar with the nuance. We certainly have defamation laws in the US as well, so there's that.

I think criminalizing (or even "tortifying") lies is completely unworkable. This kind of edifice has to be full of exceptions to avoid immediately collapsing under its own weight. In the average political campaign (US or UK), how many lies are told that cause harm to a candidate's reputation or future employment prospects? One a day, conservatively.

The real linchpin is whether these laws are used in practice to protect the powerful from criticism, or normal people from vindictive harassment as you suggest in your example.

If your example actually happened, I think that first, people would ask the person claiming "XAA RAPES CHILDREN" for evidence. If the claimant failed to provide it, listeners would rightly view them as crazy. In any case, it is unlikely I could afford to prosecute it, especially if it caused me harm to the extent of causing me to lose my job or something.

Like everything in law, it's complicated and has pros and cons. My personal bias is to lean towards free speech if individual cases are in doubt.

Eh, the simplified version is:

- Engineers (like electrical etc) use Matlab

- Statisticians use R

- Rich people sometimes use Mathematica

- People with some CS background use Python/numpy/sklearn

For "purer" math, like symbolic differentiation and such, I don't really know any of those people, but I think Mathematica has a lot in that area. The others have stuff, but less developed. But pretty much every major language has linear algebra at this point.

If the laws were the same in the US, and I called Donald Trump an idiot, would I have to furnish an IQ test? If I wrote that Ballmer/Nadella was incompetent and Microsoft products are crap?

I guess the problem is that the nature of speech often results in opinion-based insults being interpretable as a literal factual claim, and that the benefits of people being able to express their opinions freely generally outweigh the potential reputational costs to insulted individuals.

It is really analogous to the ongoing debate in US universities about whether mere words can cause such a high degree of harm as to be actionable. I think they can cause harm, but that the chilling effects caused by restricting speech are almost always greater.

I can understand this, but there are plenty of police chiefs, district attorneys, federal prosecutors, and intelligence folks who later go into politics and don't become authoritarian. George H.W. Bush was Director of the CIA before becoming President, a job that would make anyone paranoid if anything would (perhaps a close equivalent to Home Secretary?). Yet he did not become excessively authoritarian.

Since most politicians start out as lawyers, many will inevitably have exposure to crime. So I too appreciate the humanization of May but don't see it as an excuse for authoritarianism. In fact, these people more than any should know about the delicate balance between liberty and security.

I consider the occasional need of the government to tell the media "you can't publish this for national security reasons" to be in a different category. This has legitimate use-cases, and with proper oversight can be done correctly. Surely we would not have wanted the media to publish about the Manhattan Project or Ultra during WWII. The US has NSLs which seem much the same thing.

Specific, targeted actions like this are a far cry from dragnet responses like the snooper's charter or calls for banning encryption (at least any kind that can't be decrypted by the government). The danger of dragnets is that they pick up incidental thought-crime. It's why warrants exist.

Although certainly a case could be made that the media is in a better position than the government to decide what should be made public.

Believe me, I do. I didn't vote for Donald Trump or GWB. But I have to accept that their actions constitute a big part of what the US is, especially as seen from the outside.

To simplify a little, in a democracy, if 51% of voters are "backwards" in some way, I see it as an acceptable shorthand to say that the country as a whole is (in that particuar way). Because that group will be making the decisions for 100% of residents.

Yes, I see the connection, but I have difficulty seeing how politicians think that even with such measures, they can actually control discourse (apart from going full China on the matter).

I have always found it interesting that the great dystopian fiction is always set in the UK: 1984, V for Vendetta, etc. And yet, UK politicians regularly make statements about surveillance that would immediately invoke those associations in the US, but UK residents, seen from the outside at least, don't bat an eye. In the US, we have a great deal of surveillance as well, but the government at least feels the need to hide it rather than flaunt it.

In other words, perhaps those authors also recognized this autocratic tendency (assuming it exists).

Fascinating and useful explanation. I think, then, the difference compared to the US must be that in the US, the majority of both parties have authoritarian tendencies, but there is a vocal minority of "libertarians" in both parties, whereas in the UK, you seem to be saying that libertarian pressure comes mostly from the outside the traditional political structure.

I do wonder, more broadly, whether it is possible for a country to have the benefits of the "European" approach (better social services) and of the "American" approach (more robust civil liberty) all in one, or if they are somehow mutually exclusive.

From the standpoint of security, I don't see much of a correlation between the level of surveillance/repression/etc and the level of terrorist activity. It seems to me the biggest determinants of whether a country will suffer terrorist attacks are: 1) immigration policy, 2) perceived level of "colonialist"/"globalist" behavior, and 3) wealth. It is very interesting to me that Poland has virtually no terrorist attacks whereas their neighbors, Germany, France, and UK, have them quite regularly, but I don't know the explanation. It is surely not that Poland regulates the internet more tightly.

I assumed as much. The question was half-rhetorical, though: such regulations did not and could not prevent what happened. There seems to be a law of diminishing returns on regulation: sure, common-sense things like not allowing weapons to mentally ill people go a long way, but increasing regulation past a certain point doesn't reduce problems and only restricts freedom.

So in that case, the question becomes, has the UK already passed that point for internet regulation? I don't know the answer, but I assume that plotting specific attacks or making specific threats online is punishable in the UK just as it is in the US.

I just don't understand the UK on surveillance and free speech issues. In most ways, the UK seems more sane than the US, but they have these absurd libel laws and take any and every opportunity to cry "regulate the internet" when anything bad happens involving the internet. Notably, there was no call for "we need to increase our regulation of cars or knives" after this incident.

Any UK resident have insight on why the UK is so backwards on these issues? It is pretty obvious to anyone with a cursory knowledge of computers that terrorist "safe spaces" (assuming that big internet companies knowingly provide them, which I find dubious) are kind of impossible to remove or regulate. The ability of the internet to provide encrypted, point-to-point communications is a pretty core feature.

I am glad to hear Clojure is slowly getting its act together WRT numerics. I reluctantly dropped the language several years ago because the ecosystem for any kind of math/ML was just nonexistent, unless you consider a "call the terribly verbose Java libraries for everything" approach viable.

Especially nice in this Neanderthal lib is the GPU stuff.

But it's not as if China, India, etc haven't benefited from the West's carbon-fueled economic growth. It was that economic growth that led to all the various technologies that they've imported. The West has also borne, and is bearing, the costs of R&D on green energy.

If we are talking about "inequities", I would say these countries took an "unfair" leap in technological development because of us, so it is not entirely unreasonable to say the disadvantage of less carbon use has been more than offset by tech transfer. Especially since China and India are both notorious for not respecting IP rights.

If not for the West, neither of these countries would even be in a position to want higher carbon emissions.

Yes, I know they are different, but I was talking about C++. I heavily use the STL and greatly prefer it to C idioms. I also like namespaces. Templates can be very hairy but void pointers are not great either.

In short, I simply think C++ can be used responsibly if you limit yourself to the subset of features that you (or your team) can understand.

For example, there are people who go crazy with Python metaclasses and such. A statement about interacting language features being dangerous could apply to anyone in any language who is trying to be unnecessarily clever.

Honestly, I never seriously considered it. I thought it was more-or-less an "Apple thing". OFC it is supported on other platforms, but seemingly not widely used (I write scientific code that really only has to work on Linux).

But after looking at its feature set, I wonder how easy it is to get by without templates? It seemingly uses weak typing instead like Go for scenarios that would normally require generics/templates. My attitude towards templates/generics is that they are absolutely necessary for a good standard library and core containers, but implementation of new templates in your own code should be rare.

Also, I do like namespaces in general, although not necessarily C++'s implementation of them.

Sure. I use it for some things and straight C++ for others. Cython is good for simple things, like small tight loops.

One thing that is much more reliable in pure C++ is any kind of threading or OpenMP etc. Theoretically Cython has it, in practice it can cause very weird problems. Also if you want to use C (or C++) libs in Cython, you have to manually declare all the function prototypes before using them.

Also Cython has a tendency to make minor breaking changes to syntax every version, breaking my code and/or making it version-dependent. Since I distribute to others, the best method for linking performance intensive-code to Python I have found is:

1. Write the core in C++

2. extern "C" a simple C API for it

3. Compile both to a SO

4. Access the SO from ctypes

It is more robust than Cython IME. Another advantage of this approach is that you have now have a generic C library you can use anywhere else if you want, not just Python. And you don't have to link -lpython so the C lib can be used in places where Python isn't installed. Finally, it can be nice to have some C++ binaries for startup time, and AFAIK you can't compile binaries with setuptools.

I, on the other hand, am a primarily Python programmer being slowly lured to the Dark Side.

The thing I love about using C++ is that you get direct access to all these rock-solid libraries in C that everything is built on -- zlib, libcurl, etc. So instead of dealing with a tower of wrappers in a high-level language, each of which has quirks and bugs, you get that core lib, where you can safely assume 99% of the time any problem is your own fault. Even when a C++ wrapper exists, I use the C API just to reduce deps and avoid the kind of problems you're talking about, which I think boil down to templates most of the time. I don't use templates unless there is no other way.

This strategy works pretty well for me, but my codebase isn't huge and I'm not doing anything super complex -- just trying to speed up things that are too slow in Python, be it parsing or data structures. I think in practice it is best to use multiple technologies if possible in a project, each one suited to the task.

And WRT the coroutines in OP, I don't know the details, but the thing I miss most from Python in C++ is yield and generators, which hopefully coroutines will eventually help with. Defining your own iterators is very annoying.

Yep, Taleb's Medium account is a great place to go if you value any kind of expertise and want to raise your blood pressure.

However, this is one of the very few pieces by Taleb I sort of like. He makes an important point: if you work for an organization, and you aren't financially independent, you are not a slave iff that organization needs you more than you need them. In some ways, he independently discovered the concept of leverage here in 2016.

But there is a danger in this line of thinking that he gets close to when he talks about Putin. It reminds me of the logic of "pick-up artists": the idea is that simply acting independent substitutes for the reality of being independent.

That really doesn't work in most cases, although there are always many people who can't see through the facade, which may make it a useful tactic in politics. If this kind of bluff is used in employment, it relies on the assumption your employer is gullible, which is not a safe assumption to make.

And if you actually do have the reality, I see no reason to boast about it. "Any man who must say, 'I am the king' is no true king", and all that.

The thing that distinguishes my situation from the type of "unicorn job" you are talking about is that I have some realism, and I compromise.

Clearly it is ridiculous to expect I can do what I want 100% of the time without any of the responsibility that normally goes with being a PI. Instead, I accept a middle ground of autonomy and responsibility between postdoc/student and PI.

Your article largely seems to be about why employers won't sponsor pure unicorn jobs. Quite true. That's why I have to continually make it clear what value I provide to the bottom line. But I have still done better than the 10-30% you talk about. It's probably 50-70%.

I have found that science and life in general can be quite flexible if you know what you want and take concrete, realistic steps to make it happen.

I'm at Oklahoma Medical Research Foundation. My focus has been on human, mainly because that's where the impact is, and because of minor technical issues making it easier to do human data. But we do other species as well, the system is just rougher around the edges for them.

The way we get around the bias for (supposedly) hypothesis-driven work is to get a lot of our funding on collaborative and center grants. We can augment our local aging research work by providing them with context and pretty figures for their papers and grants, and they are happy to send a few $100K our way in exchange. We got a Shock Center award a year or two ago which has accelerated things a lot. A lot of it has been catalyzed by Arlan Richardson and Holly van Remmen, who came here from UT San Antonio.

When it comes to "bridging the gap", yes, it is hard. The key I guess is to be flexible and understand what they want. A lot of them are not looking for grand, informatics-driven hypotheses, but they are ecstatic about extra figures or data they can put into their papers to spice them up. Sometimes I slum it a little and do routine statistical analyses on their experiments just to keep the connections live.

In general, I think it is sensible to try to design systems in such a way they bridge multiple species from C elegans to rodents to humans. If you do one, it's not that hard to do them all, and they all have their place, as C elegans is good for quick testing of lifespan stuff but may or may not generalize, so it is considered helpful if you can use data to give tips about what may or may not translate.

The best brief advice I can give is to figure out what the wet lab people want, and give it to them in a way that maximizes the overlap between what you are doing and what they are doing.

> they want to be left alone in a lab to answer their questions while someone else keeps the lights on. And well...yeah...you just described the perfect job.

Well, that IS what I want, probably what everyone wants. I don't like writing grants, as I said, but I am in touch with reality and consider it an important part of my value proposition. Plus, I like my PI and consider him my friend. It's not nice to leave all the shit work for your friend.

EDIT: In many ways, I have tried to alter my behavior to prevent exactly the kinds of problem you are talking about. I absolutely want my PI to feel like my presence is a net benefit. If a staff scientist didn't understand these realities, they could easily be both naive and unhelpful.

> And this hasn't backfired by having one of his "postdocs" not progressing very fast?

Interesting point, one I hadn't considered. We are a pretty small lab, 4 grad students, 1 postdoc, occasional summer/rotation/temp postdocs or students, and me. In the past we had 2 postdocs.

The 1 postdoc who is in the lab now is a statistician who has zero interest in his career or ever being a PI. He is extremely valuable but I don't get the impression he feels held back. He is a "give me a job and I'll do it" type.

For informatics, I believe the productivity multiplier for skill is higher than in wet lab. Thus, it is not only easier but potentially better to have a small lab with high skill than the converse. So we have little interest in rapid growth.