User talk:Tim Starling/Feature poll

Performance vs. features related to quality

The current poll has majority support for quality. I am sure if the poll had been in a week when the servers were acting like comatose snails, we would get a different result with more emphasis on performance :)

Further Angela makes a good point that we can't use those features with a dead site.

Even so, I would like to extend my plea for quality.

Experience shows that strides in performance, be it at the software and hardware level, only provide temporary respite, traffic tends to grow to consume the newly available resources. This has been proven in the case of motorways (freeways) and seems true here too!

Problems with performance tend to manifest themselves in very direct ways - the web site becomes unavailable! However the "recent changes problem" is never likely to have such a "big crunch". It will slowly, incrementally, become less and less usable as the number of edits increases. It becomes slowly more difficult to meet, greet, help out new users. It becomes harder to keep on pushing up the average quality of article. Thus any "colloboration tools" that can be implemented will be a big, permanent win. Little anti-vandal features like soft-protection of pages (to prevent moving of popular pages and to prevent them from being turned into redirects) will also be permanent wins, just as the template and category features are.

Please don't get me wrong, I love performance enhancements. I bow my head in the direction of Brion and Tim (no mean feat I can tell you, given where they live) and the others whenever they appear. However I see performance enhancements as a means of upkeeping and improving the quantity of Wikipedia. Features are a means of upkeeping and improving the quality of Wikipedia. As we passed 300,000 articles today, we seem to be doing ok with the former :).

As this rantlet is written by someone who has never run a substantial site, or written db or web code, I hereby provide you with a virtual pinch of salt to take with you! Pcb21| Pete 09:20, 7 Jul 2004 (UTC)

The people who voted on this poll may not necessarily understand how close we are to severe performance problems, at any given time. I think it would be irresponsible of me to let service degrade to the point where no-one can use the site, if I could easily do something about it. However there is more than one way for a project to die. It's a reasonable question to ask: what would you rather have, a fast site or a good encyclopedia? An interesting way to look at it is to consider growth to be caused by an increase in capacity. We're constantly focused on performance, to meet the growth we observe, but in part the growth is due to our own efforts working on performance. Perhaps we are reaching the point where the site is so large and so busy that the wiki process breaks down. The traditional RC is no longer sufficient. Would it be wrong to mark time briefly, to accept slower growth while we recover the robustness of the original wiki process? Or should we push on despite the warnings of impending chaos? -- Tim Starling 15:19, Jul 7, 2004 (UTC)

Why is there a need for such a tradeoff or prioritizing in the first place? The foremost problem seems to be a shortage of developers. Their number isn't growing proportionally to the number of users. Have their been any attempts to recruit more developers, so that all requested features can be quickly implemented? Gzornenplatz 13:23, Jul 8, 2004 (UTC)

First, I'm writing this after considerable frustration as WP has intermittently been borderline unusable for me for the past couple of days. Lately, I have frequently had to wait upwards of a minute for a diff or a page to display. Very frustrating. Typically, I keep two or sometimes three windows open and while an edit is saving (which almost always takes a little while), I can usually look at RC and check some edits while I wait. That is impossible now as all windows are so slow I'm left twiddling my thumbs or checking email. I probably won't even bother to edit much longer if this keeps up, and checking RC is out of the question. So this is my rant for performance over new quality-related features. Maybe it's not an either/or situation and new features might actually improve performance (e.g. perhaps by not having to refresh my watchlist or RC as frequently), but when performance is like this, doing any work on WP is almost masochistic, whether adding new articles or doing janitorial tasks. older≠wiser 16:35, 8 Jul 2004 (UTC)