4 ms·

I'm curious if you don't mind sharing, how many people are working full time on the HN codebase right now? I don't know if that number is even less than or grea

by lacker 12y ago

I'm curious if you don't mind sharing, how many people are working full time on the HN codebase right now? I don't know if that number is even less than or greater than 1. I'm also curious if there's a straightforward way to serialize these closures. I'm guessing no or at least they would not be persistent across restarts. FWIW, if I were working on HN, I would be concerned that at some point, HN will require more hardware resources to scale, and it may require some fundamental architectural changes, like more separation between app logic and DB. I wonder to what degree you have scaling plans, like if we hit capacity X we would need to replace subsystem Y.

dang 12y ago
> how many people are working full time on the HN codebase right now? Less than 1. If you include everything HN-related, that number goes up to 2 or a bit more, depending on how much time kogir has. > I'm also curious if there's a straightforward way to serialize these closures. Do you mean to disk, or to the client and back? My sense is that it's technically possible but the I/O overhead would make it not practical. However, that's not based on much knowledge. Perhaps we could do it now that we don't have so many. > I wonder to what degree you have scaling plans We have a lot of scaling plans, but I wouldn't call them definite.
kevin 12y ago
There are two people dedicated to HN development. Daniel and Scott. Four other people from the YC software team also contribute, but not full time because they have to also work on other projects: Nick, Brett, Trevor and myself. Other people on the software team: Garry and Dalton. We actually just upgraded the hardware for HN very recently. A lot goes on behind the scenes and it's a testament to the team that it's all fairly transparent to the community to the point where most of you think nothing changes at all. The point is to keep the community focused on contributing, voting and commenting the very best content for hackers. It's one of the reasons I was delighted to see Dan write about some of the work they do like reducing expired links. It's a rare look at how much thought goes into what feels relatively simple.
- jacquesm 12y ago
  Making the complicated look simple is the hallmark of any great success. You guys are doing absolutely stellar work, every now and then I get a glimpse of what is being done and I feel that HN has definitely changed for the better over the last year or so in a technical sense and the moderation transparency has also greatly contributed to the change in atmosphere.
  - drostie 12y ago
    If you think about what magicians do and how they do it, "making the complicated look simple" is also a nice definition of magic. In other words, if you're ever thinking, "how could she have known that I'd choose the 7 of diamonds?!" you're probably victim of some implicit thought like, "she couldn't possibly have bought 52 decks, pulled out the 7 of diamonds from each of them, and built a deck consisting only of the 7 of diamonds, so that no matter how I shuffled and cut that deck the top card would be the same. No one in their right mind would spend the time and effort to do that." But, that's exactly what she did. She did a lot of abstracted preparation so that the hard work behind the scenes just vanished at the higher level of performance. That's what a good cook does, it's what a magician does, and it's the entire role of "administration" and "middle-men," theoretically-speaking.
lysium 12y ago
> I'm also curious if there's a straightforward way to serialize these closures. The closures include functions, which makes this a bit difficult.

by lacker 12y ago

dang 12y ago

> how many people are working full time on the HN codebase right now? Less than 1. If you include everything HN-related, that number goes up to 2 or a bit more, depending on how much time kogir has. > I'm also curious if there's a straightforward way to serialize these closures. Do you mean to disk, or to the client and back? My sense is that it's technically possible but the I/O overhead would make it not practical. However, that's not based on much knowledge. Perhaps we could do it now that we don't have so many. > I wonder to what degree you have scaling plans We have a lot of scaling plans, but I wouldn't call them definite.

kevin 12y ago

There are two people dedicated to HN development. Daniel and Scott. Four other people from the YC software team also contribute, but not full time because they have to also work on other projects: Nick, Brett, Trevor and myself. Other people on the software team: Garry and Dalton. We actually just upgraded the hardware for HN very recently. A lot goes on behind the scenes and it's a testament to the team that it's all fairly transparent to the community to the point where most of you think nothing changes at all. The point is to keep the community focused on contributing, voting and commenting the very best content for hackers. It's one of the reasons I was delighted to see Dan write about some of the work they do like reducing expired links. It's a rare look at how much thought goes into what feels relatively simple.

jacquesm 12y ago
Making the complicated look simple is the hallmark of any great success. You guys are doing absolutely stellar work, every now and then I get a glimpse of what is being done and I feel that HN has definitely changed for the better over the last year or so in a technical sense and the moderation transparency has also greatly contributed to the change in atmosphere.
- drostie 12y ago
  If you think about what magicians do and how they do it, "making the complicated look simple" is also a nice definition of magic. In other words, if you're ever thinking, "how could she have known that I'd choose the 7 of diamonds?!" you're probably victim of some implicit thought like, "she couldn't possibly have bought 52 decks, pulled out the 7 of diamonds from each of them, and built a deck consisting only of the 7 of diamonds, so that no matter how I shuffled and cut that deck the top card would be the same. No one in their right mind would spend the time and effort to do that." But, that's exactly what she did. She did a lot of abstracted preparation so that the hard work behind the scenes just vanished at the higher level of performance. That's what a good cook does, it's what a magician does, and it's the entire role of "administration" and "middle-men," theoretically-speaking.

lysium 12y ago

> I'm also curious if there's a straightforward way to serialize these closures. The closures include functions, which makes this a bit difficult.