5 questions for YCombinator's Paul Graham

PG: No, not at all. I'm used to the popular language being something other than Lisp. In fact, it's encouraging that it's now Ruby. If you look at the trend over time, the successive popular languages have been converging on Lisp.

MJ: When I originally read Hackers and Painters I got to the part where you explained why the Spamassassin scoring approach wasn't as effective as your Baynesian probability method and I was stung by that (I was the guy who came up with the scoring methodology in the filter.plx precursor to spamassassin).

When I got over myself and reread that section I accepted that your method made more sense for the reasons you described. Have you ever had to swallow your pride and admit somebody else's approach or design methodology had advantages over your own?

PG: I've often changed my mind after trying something that didn't work, but usually more because it blew up in my face than because I saw someone else doing something I wanted to do instead. I've learned a lot from other people's examples, of course, but usually more gradually.

MJ: In the "Revenge of the Nerds" chapter in Hackers and Painters you describe the pointy-headed academic and the pointy-haired boss (both archetypes from Dilbert). What do you call the "hacker boss", guys who came out of a coding and hacking background who now run companies and departments? What are their general characteristics, strengths and weaknesses?

PG: I suppose the danger with people like that is that they wouldn't be into the human side of managing people. I've heard of people
who were great hackers but lousy bosses, because they only wanted to hack. To be a good boss you'd have to care about both.

MJ: You are an evangelist of the bottom-up approach and you probably advocated the Agile development framework before the "agile" label took hold. Do you think bottom-up and agile are just plain "better ways to approach things" or are there situations where top-down and rigidity work better?

PG: Top-down works better when it's hard to undo mistakes. For example, in building something like a dam, you probably want to plan everything beforehand. The reason bottom-up is becoming more popular is not just because people are learning about this new method, but because the world is actually changing. The mediums we use are more flexible now, which makes mistakes easier to undo. Which means a lot of long-established traditions are now out of date.

MJ: In this July 2006 interview you stated that you didn't think we were in another bubble, nearly 2-years later with certain pre-revenue companies doing 3rd and 4th funding rounds with post-money valuations north of the hundred million mark, has your opinion on this changed at all?

PG: It's gotten a little bubblier, but it's still nothing like the Bubble of the 90s. There are a lot of fairly fluffy startups, yes, but the difference between now and then is that these fluffy startups have only raised $100k instead of $3 million.

MJ: And now the obligatory easyDNS-centric question: How did you learn about easyDNS? How and why do you use us?

PG: I learned about you guys from Daniel Giffin, who once worked for Viaweb. It was quite a few years ago, when the median registrar
was probably even worse than now. He said you did things right, which was all I needed to hear.