I take issue with this. :) I often disagree with what Paul Graham writes. However, to call him a "profoundly unserious public intellectual" is silly. This essay doesn't make the case. Who would dare create a language and announce such a lofty goal? I have never met Paul, but I have been through YC and have interacted with many close to him. IMHO Paul is a rationalist to a fault. However, he has the uncommon ability to inspire others, and he has found tremendous value in a simple formula: 2. Try to understand why you were right or wrong. 3. Improve upon your idea. 4. goto 1. YC is a product that tested this formula, and it worked better than anyone (probably Paul included) could have anticipated. I have talked with hundreds of venture capitalists inside and outside of Sillicon Valley, and no one talks to founders like YC. Like everyone, Paul is far from perfect. However, I admire his positions regarding failure and humiliation. IMHO thousands of brilliant people harbor ideas that are lost to time because of fear of failure and humilation. If you create a new programming language, someone will call it a failure if it is not widely adopted. If you create a social website, someone will call you a failure if only a hundred people use it. I see Arc as a success.and Hacker News (now running on a machine with enough memory to stay afloat) remains the only meaningful deployment of Arc in existence.
1. Test your idea.
This may appear aggressive but it's not meant to be, I'm not a good writer. You don't have to justify it, I'm genuinely curious. What criteria are you using to measure success? And what would an unsuccessful new language be? Is it in the sense of "participation is success" like the non competitive games where just taking part gets you a medal? Or is success just having built something concrete? Do you think you may be biased because Hubski was originally based on HN so you needed to adopt it? I'm not sure what happened to rob05c and I guess he doesn't represent Hubski but I do remember: .. and .. By the way what became of the Hubski conversion from Arc to Racket? I'm not conflating popularity with success but virtually no-one has adopted Arc. The community seems tiny and inactive - I can't even find an official active repo - just a community fork. Did they get it perfect first time so no need to change anything or fix bugs or even discuss changes? I see Arc as a success.
The creation of a language with the desired characteristics. One not created. My comment doesn't regard the qualities of Arc itself. I really enjoy it, but I am not a serious programmmer, and my experience is limited. I just think that it's ridiculous to call an unused language a failure. Paul is far better off for creating Arc. He has implemented his vision, and now he can go from there with all of the lessons learned if he chooses. I definitely wouldn't equate the creation of a programming language with "participation". There is far too much effort and risk involved. It's creation. I suspect that if Zach Tellman had created a programming language, he wouldn't consider calling someone else's a failure. It stalled after rob05c burned out on the project. He was doing all of the heavy lifting, and tbh, it wasn't fair as I was pouring my energy into Forever Labs at the time. He did us an incredible service not only with the partial conversion, but with getting most of our data on SQL, rather than the insanity of files that news.arc used. (Tellman is wrong about that. There's no way that HN is the same news.arc implementation just with more memory thrown at it.) Hubski is partially Racket. I think rob05c was right, there isn't good reason to choose Arc over Racket, and it would make more sense if Hubski was completely converted. I really should commit myself to all new code being Racket. The syntax isn't that different.What criteria are you using to measure success?
And what would an unsuccessful new language be?
By the way what became of the Hubski conversion from Arc to Racket?
Right, you had mentioned you wrote hubski in Arc and so I was interested to see what you thought about this. No judgement about your choice of language, this is your project and you should do what feels right to you. And if you like it, more power to you. About PG though, he is more a guru brand than an engineer, and has been for a long time. For Arc, he set his own goal posts as a "hundred year language", and then failed to meet them by a country mile. He also ignored his own advice about learning from feedback. His M.O. will continue to be: oversimplify a problem and make bold proclamations about it with authority, without regard for accuracy. His proclamations about wealth and economics follow the same vein. Arc is an cautionary tale about success, ego, luck, and bullshit in Silicon Valley. As for the essay author I think he would admit his own failures in his own field, most working engineers will.
It occurs to me that K meets most of these language design goals better than anything, and not many people have heard of it, because it's hard to read. Here is a program to list all the prime numbers between 1 and R: Oh by the way, it's faster (!!) than C. People don't use it except in niche applications because people can't look at it and see what it's doing. To me, it might as well be brainfuck with the amount of line noise it has. Readability and rapid comprehension are even more important, IMO, than brevity and power/expressiveness. That is why Python is so popular. It got all its libraries and ecosystem because it was easy to write AND easy to read, in addition to having a kind of syntactic brevity, so lots of people could rapidly iterate on it. It's not as powerful as Lisp or Haskell or whatever, but maybe it doesn't have to be for most applications. I think Nim is a language to watch for these reasons. 2_&{&/x!/:2_!x}'!R
This is probably the first time I've seen K mentioned outside Project Euler or similar place. Kudos! As far as Python's popularity through readability is concerned, I'd add one extra facet related to readability: it makes writing tutorials, and learning/teaching in general, much easier. It also makes post-intermediate resources more available and treated by multiple people, even as blog or SO posts. Few languages can match it in that regard. Power is important, but for every hacker determined to program something using only folds and flips or what have you, there's a thousand people who are unlikely to need stuff beyond Think Like Computer Scientist and a decent level of reading comprehension (for mentioned blogs or SO posts if anything). I'm saying it as a fan of both Haskell and Python.