Comment by Wowfunhappy | Hacker News Reader

Comment by Wowfunhappy | Hacker News Reader

Parent Full thread

> There has been a bug where comment pages can only show a limited number of entries for years, and the wonderful mods like dang continually report it is being worked on.

It's not a bug per se, it's a concession to conserve server resources, which will be removed once some optimization work has been completed.

You might respond "well why is HN so resource-hungry?" Answer: it's not! HN is run off a single core of a single CPU.

https://news.ycombinator.com/item?id=23187455

> The software for both HN and YC was just a single Arc program (and not a large one) for the first 9 years of their existence, during which they went from nothing to massively successful to industry-changing. Written by one person, programming part-time. That is a staggering achievement. The power of using the right language for your project goes far further than most people dream. Our imagination about this is crippled by path dependence, social proof, and the conditioning that comes from only ever doing things the same few ways, like those fish in experiments (which may be urban legends?) who stick to their corner of the aquarium even after a glass barrier has been removed. The solution space of software and programming is so much larger than most of us want to imagine that it is. Sad.

A post from a week or two ago about APL and cellular automata (I cannot find it) perhaps illustrated OP's sentiment better: it's easy for a language to do something easily and elegantly, but quickly become fantastically difficult to accomplish with the introduction of new—and sometimes mundane—constraints. The post showed how it was trivial to implement certain cellular automata in APL, but massively difficult to do when the automata became slightly more complex (while being trivial to update a Java implementation to accommodate the new complexity).

The practical difference between running on one server with one core and five servers with four cores is nothing—to YC it's a rounding error either way. If there's a meaningful problem (e.g., comments being computationally expensive to paginate well) that the language/runtime/system/etc cannot solve easily enough for it to be remedied in O(months), it does call into question whether the choice of language is the right one. Yes, it can be patched over by a human moderator. But it still highlights a significant weakness in the technology's core function.

Perhaps ironically, most YC companies (and by consequence, YC) would fail miserably if they chose purism and principle over pragmatism when it came to technology decisions.

Here's the post: https://news.ycombinator.com/item?id=27231978

> If there's a meaningful problem (e.g., comments being computationally expensive to paginate well) that the language/runtime/system/etc cannot solve easily enough for it to be remedied in O(months), it does call into question whether the choice of language is the right one.

I could be wrong, but I get the sense that HN doesn't have any full-time developers. In fact, I think it might be entirely dang—he's talked before about how moderating HN leaves him with very little HN development time.

Reply on news.ycombinator.com