If the problem with Arc is that it's not different enough from existing Lisp dialects like CL or Scheme, all I can say is that being different is not my goal. I'm trying to make it good, not original. So it will only be as original as it has to be to be the best language for writing programs (as opposed to pleasing managers, or writing papers about, or seeming comfortingly familiar).

Pull quote:

"The reason we built these apps was, in retrospect, incredibly shallow, and I thank my ex-boss for helping me realise this. We built them because we thought they were going to be popular, following the typical web 2.0 path to riches:

   1. Make popular app.
   2. ???
   3. Profit!!1

"Looking back, I think to myself, is trying to make a popular app enough anymore? If you just attempt to make a popular app, how are you any different from the next web startup. The only thing that separates you is ability, and ability can be hired or learned - so the question becomes who has deeper pockets or more free time. Thusly any idiot with time or money can build a web app these days and these were crowded markets we were attempting to break into - productivity and mobile blogging. What then, separates the wheat from the chaff in this case? For me, I think the big one is emotional investment."

One of the weird things about doing something mostly benevolent is that you end up breaking the standard journalism algorithm. We've hit this on several occasions. They feel like they have to find something at least slightly bad to say about us, and they end up choosing something random. It seems like the same thing happens to Google.

I'm not sure if this happens because it's an axiom of journalism that every story has two sides, or because conflict makes stories interesting. Maybe both. But one sees it often on TV news. A city bans people from setting one another on fire, and the reporter diligently tracks down the president of the local chapter of Pyromaniacs of America to offer the other side of the story.

When I was a kid, this was my favorite documentary of all time.

I'd like to highlight number 3, it's harder than you initially thought. But you'll go with it anyway cuz you don't know any better :-)

I'm still considering other syntax, but I want to take my time deciding. Adding syntax is like a constitutional amendment.

I spent a couple days working on getting rid of parentheses, and it got really ugly.

Those bastards actually have parentheses in their urls? Argh. Will fix.

I don't. So far the percentage of companies that go public without at some point taking VC money is still zero.

In effect it takes about a third of all companies, because corporate tax rates are about a third of earnings.

The biggest technology startups inevitably go public, if only because they've given stock to so many people that they have to. So what I'm saying is: no startups seem to get really big without taking VC money.

I think there might be several optimal languages for different domains. Not 100% sure yet whether that's also true for different users. I know I'm not planning to protect users from themselves. If there is a genuine need for that, then different languages might be better for different users even in the same problem domain.

The last Boston meet up was a big success. There was a postmortem discussing the event but I can't seem to find it at the moment. The original thread is here: http://news.ycombinator.com/item?id=17969

I wanted to organize one more event before the end of summer so, if no one else wants to take the ball on this one, I'd be more than happy to get something organized in the next few days.

Internally strings are represented as 32 bit unicode. That seemed like the right plan, and conveniently it's what Mzscheme does.

In news.yc, however, I currently (out of laziness) aggressively convert all input to ascii. Life is too short.

51. Because Courier 12 point went out with IBM's mojo.