This just comes across as arrogant. You can neither know how smart the median user is, nor does it say much about the merits of the tool. It encourages some weird sense of superiority by putting yourself in a group that's exclusive and think they're smarter than everyone else. There are better ways to pick programming languages than trying to maximise your own pretentiousness.
Even in lieu of that bad taste or elitism, I'd still like to suggest that it takes more patience, self discipline and perhaps a tinge of being clever to pick up something like C, C++, Rust, Haskell, Lisp, APL, Prolog or even regular expressions than it does Java, .NET, Node/JavaScript/TypeScript, Python and others and I say that as a person who relies on the latter for the most part in their daily life.
It's not that you have to be an outright genius to use any of them, but writing code in them well (for example, managing memory and avoiding C++ footguns) takes discipline and definitely could be applauded on some level, as could using APL, Prolog or regular expressions to efficiently solve problems in the few domains that they're actually good at and suited for.
Personally, some of these views of mine formed after listening to a podcast that talked about APL and someone's experience with it: https://corecursive.com/065-competitive-coding-with-conor-ho...
Just because the average developer in a language is bad it doesn't mean that the language itself is bad.
I'd say it depends on what we're defining "first programming languages" as - first PL learnt in intro-level university lecture, or first PL learnt individually? But my impression is that the most popular PL for beginners seems to be Python. And while I'm not really a fan of Python, I understand where that's coming from: Python is relatively simple to learn, has a huge library ecosystem and comes with relatively few footguns.
I feel like this is obviously true when taken at face value, however in practice there are probably loose correlations (not causation) between those two, even if not for the reasons that you'd expect.
I'm not sure whether it was mentioned in the same podcast that i linked, or another one, but there was a very interesting example for Haskell that was offered in one discussion. The argument was that the people who utilize Haskell tend to be more likely than the average developer to have ties to academia (which may incidentally also speak of their cognitive abilities):
- Haskell is popular for numerical processing, moreso than for the somewhat typical CRUD apps
- because of this, Haskell is more likely to have materials describing it which are aimed towards an academian audience
- ergo, those materials won't always be simple step by step tutorials on how to do things, but will sometimes assume familiarity with certain maths concepts and will also focus on solving a particular set of (advanced) problems
- and when these materials attempt to explain these simpler things, invariably you'll still end up with explanations of what monads are somewhere in there, whereas with something like Python you'd have less overhead and a less steep learning curve
- as a consequence of all of this, Haskell won't be as popular as Python, or similar languages
Now, the person might be wrong, since their observations are bound to be subjective and i might be wrong to assume that they hold true in many cases because of how believable all of that sounds, but to answer your question, i'd argue that the average user of Haskell is indeed a bit more intelligent than the average user of Java or many other general purpose languages (even though then we step on the mine of having to define what intelligence even is, where opinions can once again differ).
I'd say that simply happens, because some of those languages are far more approachable and therefore more widely used by a large audience of people, not all of which need to concern themselves with solving the issues that people in academia need to deal with. Of course, there could be the counterpoint that not everyone in academia is brilliant, but i remember seeing some pretty admirable professors and researchers back when i was in my university!
Some languages are difficult to learn initially (like Prolog) but that's not because they are intentionally obscure, but because they need to be different.
- programming language problems like making certain language features, paradigms and kinds of abstractions available (or not available)
or
- solving actual problems when building software - how to build a maintainable and performant server, OS, web UI etc.
For the second you're better off staying away from the 'weird' languages especially if you care about being productive.
Once upon a time (about 1992ish) in a land where the moles
have beaks and swim and the deer jump on their hind legs
(Australia), a certain computer science department (RMIT)
decided that Pascal had reached its use-by date. What shall
we do? What shall we replace it with? Set up a committee!
I was on the committee. We set up a short list.
(1) Scheme.
Tiny language, amazingly capable, REPL so you can try
things out, implementations for all the machines we
cared about. And Rob Hagan at Monash had shown that
you could teach students more COBOL with one semester
of Scheme and one semester of COBOL than you could
with three semesters of COBOL.
(2) Miranda.
The commercial precursor of Haskell.
Melbourne University were also looking at this (or had
already switched to it) which would make it easier to
pick up some of their students.
(3) Ada.
Close enough to Pascal that our staff were comfortable
with it and our material would not need major revision.
A better Pascal than Pascal.
Handled concurrency about as nicely as an
imperative language can.
What was the deciding factor?
Schools.
Our deliberations leaked, and we started getting phone calls from
careers masters saying "if you go all theoretical [read: (1) or (2)]
we'll tell our pupils not to study with you."
Personally, I'd maybe not apply the adjective "smartness" here, but there is some sense in which some languages have better taste than others, and what drives language popularity is not usually good taste, and I interpret pg here to be saying he prefers to recruit programmers who show good taste in the languages they work with.
I don't have any sort of legitimate control group to compare to, and I'm obviously biased as heck, but I do still hold the impression that I and others who learned that way can learn new languages and programming techniques much more quickly than acquaintances who were fed a steady diet of Java or Python. Many of them, even 20 years on, still don't really have a firm grasp of what single dispatch is in the first place, which hinders their ability to use it well. Which is ironic, considering it's all they've ever known.
I agree; I hate the "smart" perception. My guess is that, if there's anything there, it's just the gift of a stronger grasp of the fundamentals. Scheme brings you very close to the abstract foundation of computer science. C brings you very close to the technical foundation of computer engineering. I'd like to think there's a lot of value in starting out by establishing a firm foundation like that.
Dr. Richard A. O'Keefe is a computer scientist best known for writing the influential book on Prolog programming, The Craft of Prolog.[1] He was a lecturer and researcher at the Department of Computer Science at the University of Otago in Dunedin, New Zealand and concentrates on languages for logic programming and functional programming (including Prolog, Haskell, and Erlang).
As a student, we were required to learn assembly in a microcontrollers course. It was a disaster, because it was a very short semester and they were asking we learn microcontrollers and assembly at the same time when most of us could barely program to begin with. I left a review urging they change to C (still low level, but so much easier) and they actually switched the next year.
Learning ASM is indeed important from a theoretical stance (and practical for some jobs), but you can't expect students to just pick that up in a few weeks worth of classes.
Then when I got an office job, I learned a LOT of engineers do process automation programming. At that point I could see even something as hideous as VBA being used to efficiently save hundreds of hours of work.
What do students really need? It depends on a lot of factors. Are they going into academia to study PL theory? Then Lisp, Prolog, C, and Haskell are probably pretty important. Are 95% of your students going into industry? Then Python, C++, Java, C#, JS, Excel, and SQL are all going to be important.
I think the really theoretical stuff (Haskell, Prolog, Lisp) should only be briefly introduced in undergrad and expanded upon in grad school. Or you have specific academic or industry based programs.
My experience is Haskell actually is highly practical for many tasks. So I think it’s misleading to call it “really theoretical”. It’s the reason you see FP in areas as diverse as spam filtering, distributed computing, smart contracts, various financial services, etc.
While less common, there absolutely are areas in industry that do tend to use FP, and I don’t think you should cut students off from easily entering those specialties (or having to plan to enter them ahead of time).
Additionally, there are concepts used in imperative languages which are not as easy to understand or express in those languages, e.g. promises, are more clearly expressed as monads in a language like Haskell.
It’s really funny you say that, I view HN as being frequently anti-FP. I guess when you split the difference everyone is a bit unhappy. In my case I’ve been programming for over 30 years and only on HN for a few months, so I don’t feel like it’s swayed me much, you never know. I’ve read Graham for years and he certainly has been an influence.
Overall Haskell alone isn’t that big in industry, but it most clearly illustrates many FP concepts and is the originating point (or an early adopter) of new tech that makes it into other languages.
If you look at all the jobs in Scala, Haskell, Clojure, F#, OCaml, Elm, Erlang, and PureScript. Then you throw in all the jobs in multi-paradigm languages Python, JS, etc. with managers or coding standards that dictate functional techniques, then subtract the particularly awful soul crushing dead-end jobs (that regardless of paradigm or language) probably none of us want, I think you may be underestimating the representation of FP.
Ultimately when I hire someone with a 4 year degree from a research university, I’m not expecting them to have 100% vocational training, they should be literate in at least a couple languages, but I’m much more interested in knowledge of concepts and adaptability.
The best Python programmer I’ve worked with described his junior year as being almost entirely Standard ML. I think he was better for it.
> And Rob Hagan at Monash had shown that you could teach students more COBOL with one semester of Scheme and one semester of COBOL than you could with three semesters of COBOL.
Really makes the point about how studying uncommon languages can pay off.
Whik I'm not sure I'd always agree with PG, a recent anecdata I heard was a company selecting for Lisp because picking Python meant wading through mountains of fresh bootcamp grads with (quote) "overblown sense of their own capability" (unquote) - same applied to JS, less so to C#/Java.
(Un)fortunately, the company noticed they didn't have resources to filter that, and decided that by going for "weird but capable" combined with global remote organisation they could filter for people experienced, or willing to experiment with weirder languages. Apparently ithe cost of having to sometimes implement things from scratch was lower than expected.
Paul Graham essays decline in value exponentially with the number of times he mentions Lisp in them. He had one that went below 2^-20.
Doing something outside of your bubble is great for learning some humility and G-d knows PG could do with some of that.
However I am not going to argue that the median javascript user is as smart as the median ocaml user. 99% of people learn javascript because they have to. Most people learn ocaml because they are enthusiasts. They're interested in programming language concepts and becoming more effective.
I would never call it a "good" language. It is an abomination created by the dev teams at 4 different browser companies over the course of 20 years with no coordination between them.
It is the swiss army knife of programming. Even top-tier hackers use it as a primary tool. Check out Frida[1], it uses the Chrome v8 engine to inject parasitic tracing code written in JavaScript into any OS (incl. android & ios). There's no language that would have fit the bill like JS+v8.
- Relatively fast
- Enormous ecosystem - the "good parts" of which still have practically everything you need
- Top notch typechecker in the form of typescript
Every single programmer interested in any language in the ML family or Haskell has considerably above average[1] intellectual horsepower. The syntax and type system can both fairly be called "weird."
The Lisp programmers I've encountered are along a wider range. They've ranged from average to ML tier, the latter being the ones that, as PG observes, are more interested in "weird" features like macros, call/cc, and so on.
And then you have extremely popular languages like C, C++, Java, JavaScript, Python, and so on. Virtually every programmer has used one or more of them to a significant degree, so the mean programmer using one of those languages is virtually the same as the mean programmer in the general population. This is expected since in cases like this the mean of a subset approaches the mean of its superset as the size of the latter approaches that of the former. Programmers who master C++'s template system are an exception to the above. They're also well above average.
[1] In this comment everywhere I say average or mean I'm referring to programmers and not the general population. Average isn't bad, the average person here is quite bright for example.
Just to point that there's nothing strange or unexpected about this. You will get people that are more intellectual than average it you filter by any niche intellectual activity with no immediate non-intellectual reward.
I mean, if you search for programmers that understand plant physiology, you will get the same effect. It basically means that smart people tend to be curious.
I'd say that since there's lack of Haskell jobs, then non passionate people do not touch it since it's useless to them
Passionate people check it out cuz other passionate people recommended it
And circle repeats.
I interpreted it at second effort as:
"Pick a language that is famous for being an oddball but that has good credentials, since you have seen that people that to you are worth credit use it".
I’m reminded of his argument about how “hackers” choose languages that let them type the fewest characters possible, an argument that seems empirically unfounded to me.
Ah yes, just like real writers write in Ithkuil, a language constructed to maximize the meaning encoded per syllable.
It’s fine if that’s his preference, but the argument that the best programmers (“hackers” in his parlance) prefer what he prefers to be both self serving and contrary to actual real world evidence.
I sort of get the sense that he doesn't realize how many crazy talented (but not famous) programmers there are out there. He's got a microphone and is determined to keep using it to tell us how great his ideas are, but the audience has gotten a lot more discerning since the 90s when producing a CRUD app in Lisp could still be considered a feat.
I believe I learned a lot, and most of the people I encountered on those slacks and forums seemed a lot more experienced and capable. That's also because the language is kind of niche but powerful, so not everyone is starting out their career trying to print Hello World in Erlang. It is a language one would discover when trying to solve a specific problem, in my opinion at least. Therefore, there are less noobs like me and the people I encountered were super helpful and thorough compared to my experience on other languages.
If I see just one smarter than me person using a particular tool for some task that I also do, it might be because they are using it as an intellectual challenge or for variety, but if I see a lot of smarter than me people using it for that task I've almost always found it is worth looking into it seriously to see if I should use it too.
Niche languages, by definition, are not good investment if you're after a well-paying job. A niche language that requires understanding advanced programming concepts is, by definition, filtering for people who a) can understand those concepts, and b) have a non-monetary reason for doing so, which implies c) they're likely not faking it.
Whether or not such people would be good software engineers is another topic. Software engineering encompasses more skills than just raw programming ability. But a proficiency in advanced niche language plus proficiency in a mainstream language plus some work experience? It's hard to imagine how this would not be a good signal for smarts and skills.
But, many years later, I’ve learned I gain nothing from this kind of defensiveness, and simply keep a running list of things I’m “supposed” to know, but don’t yet. Of course, now that is a small list of very uncommon things, which itself shows how much I’ve progressed.
Point being, digging in your heels only hurts yourself.
The way I think about programming changed a lot from having used different programming paradigms. Many programming paradigms have little real-world value, but they do make you think about computation, abstraction, and lots of other concepts differently.
That grows your brain in ways which are useful when you go back to your 9-5 Python/JavaScript/Java/C#/etc.
I no longer have time to do these sorts of deep dives, but I did this a lot when I was young, and I'm a much better programmer for it.
I'd look at what he's writing over how he communicates it. Perhaps the language might be arrogant or self-congratulatory, but it benefits you more to look at what he's saying than how he says it. (responding to thread here; you obviously didn't make this mistake)
I also see no pretentiousness. With smart people hundreds of useless discussions are avoided in the first place. If you work in a popular language like Python that attracts all sorts of people, you have 10 preposterous technical and social discussions per day.
I think this essay isn't supposed to be anything comprehensive about Lisp, it's just encouraging his target audience to consider other options.
“Pick a language that most programmers consider weird but whose median user is curious and shows the initiative to explore strange new worlds of how to get things done.”
I consider myself a very capable hack. I am not really very “smart.” But I am motivated to learn more. And to improve the craft of software solutions.