Wikipedia:Bots/Requests for approval/Pageview bot


 * The following discussion is an archived debate. Please do not modify it. Subsequent comments should be made in a new section.

Pageview bot
Operator: Verisimilus  T 

Automatic

Programming Language(s): PHP with BasicBot framework

Function Summary: Counts the number of times a page has been viewed over the past month.

Edit period(s) (e.g. Continuous, daily, one time run): Monthly run

Edit rate requested: 10 edits per minute to bot's userspace, once per month.

Function Details: Following from this discussion, the number of times a page has been viewed can be seen but only realistically by a bot. This bot will display the number of times a page has been viewed over the past month if it is tagged with pageviews. It will also be able to rank pages in a WikiProject by number of views.

Discussion

 * Hows it getting the stats/where is it getting them from...? —  Ree dy  22:26, 3 April 2008 (UTC)


 * Copied from above discussion:

Wikipedia uses Squid cache servers, which cache pages when they're requested. For example, if you enter URL http://en.wikipedia.org/wiki/Kitten and get sent the page, the next time someone goes to to the same page, the servers will remember what the parsed wikitext was and greatly reduce the back-end server work. The servers also count how many times a different page is accessed. These stats are provided in raw form at here, and that's where the numbers in stats.grok.se's graphs come from.
 * Grace notes T § 03:33, 15 March 2008 (UTC)
 * Verisimilus  T  22:32, 3 April 2008 (UTC)

In userspace probably doesn't matter. Any objections to a trial? — Werdna talk 13:41, 4 April 2008 (UTC)
 * None from me... SQL Query me!  14:25, 4 April 2008 (UTC)
 * for one full run. Max S em(Han shot first!) 07:17, 5 April 2008 (UTC)
 * I notice this task hasn't run in almost 2 weeks, how's this coming along? SQL Query me!  14:40, 19 April 2008 (UTC)
 * Problems with a) server space; b) free time. Will run soon. Verisimilus  T  12:03, 21 April 2008 (UTC)


 * Given that we have the http://stats.grok.se/ site, I really don't see what the point of this bot is. Mr.  Z- man  17:04, 23 April 2008 (UTC)
 * I agree, this bot seems to be redundant with the above stats site.  Nakon  17:19, 23 April 2008 (UTC)


 * Bot makes data more accessible from within WP (few people know about stats.grok.se) and easily visible to WikiProject maintainers. Also produces ordered lists of pages by views for WikiProjects, allowing workers to allocate resources to heavily trafficed pages.  May also flag pages to editors which are higher traffic than expected, encouraging them to add to watchlists to avoid vandalism, etc.  Has gained warm reception, for example, here. Verisimilus  T  17:34, 23 April 2008 (UTC)
 * It hasn't gained such a great welcome here though. Mr.  Z- man  17:52, 23 April 2008 (UTC)
 * Yeah, I'd agree with Simetrical on this one pretty strongly. I don't like this idea at all.  Voice -of- All  19:55, 23 April 2008 (UTC)

(unindent) I've blocked the bot indefinitely and left a note on the bot's talk page. There are far better options for this task, including a possible stats.grok API. Frankly, I can't think of many ways to do this worse. --MZMcBride (talk) 19:59, 23 April 2008 (UTC)
 * I should also note that if easy access to the stats site is the main reason, I have a 7 line JS script that adds a tab that links to the stats site for the page you are viewing, to see the stats for the current month. Mr.  Z- man  18:16, 23 April 2008 (UTC)

I'd really have to side with the nay's on this one, doesn't really meet a couple of the major (imo) bullet points of bot policy. Q T C 18:11, 23 April 2008 (UTC)
 * Also don't see the point. Ral315 (talk) 19:59, 23 April 2008 (UTC)
 * Using the bot to create comparison tables by project or by genre would be useful analysis; creating a user subpage for (potentially) every page on wikipedia just to hold statistics for that article is completley redundant to stats.grok.se. Happy‑melon 20:07, 23 April 2008 (UTC)
 * Yup, I'll agree with that one. Stifle (talk) 20:21, 23 April 2008 (UTC)

FYI, I just added an API (completely untested so far) to stats.grok.se. If you insert "/json/" into the middle of the URL, you'll get the stats in json format. Like this: http://stats.grok.se/json/en/200804/Main_page. henrik • talk  20:24, 23 April 2008 (UTC)
 * This bot's operation has clearly upset a lot of people, and duplicates the grok site (which I hadn't known about when I wrote it). I'll restrict its function to creating single lists of pages by number of views for participating wikiprojects, which I gather will attract less controversy. Verisimilus  T  20:26, 23 April 2008 (UTC)
 * You could possibly use a large #switch template, to accomplish the same as many subpages. Would also require less edits. SQL Query me!  21:08, 23 April 2008 (UTC)
 * I'd considered it, but the size of page would be too large for WP to handle. I can't even post a bullet list of all ~15k articles to one page.  I guess I could use nested switch templates, but surely that becomes messy and resource intensive? Verisimilus  T  12:38, 24 April 2008 (UTC)
 * It can't be much more resource intensive than creating 15000 pages every month. However, given that the stats site already exists (and I don't buy that you didn't know about the site since it was mentioned in your quote from Gracenotes) I really don't think this meets the bot policy, specifically, "does not consume resources unnecessarily" and "is useful." If it uses switch statements it won't use as much resources as creating thousands of subpages, but if a few big projects use this, its still going to be making lots of edits to massive pages for minimal benefit. Mr.  Z- man  17:46, 24 April 2008 (UTC)
 * I guess I should note here that a very large #switch will fail. The servers can handle processing a certain limit, but if you try to create a two-thousand line switch, it won't work well. If it processes at all, it will take a long time and will be resource expensive. The best option is have the bot hit Domas' stats or Henrik's API and generate an analysis for specific WikiProjects. Also, a user script could be easily written to hit Henrik's API and generate a count on page view or on request (perhaps in the toolbox). --MZMcBride (talk) 19:39, 24 April 2008 (UTC)
 * Thought so. I guess I'll restrict it to creating a single ordered list of articles, using Henrik's API, for each interested wikiproject. Verisimilus  T  07:45, 25 April 2008 (UTC)


 * not approperiate for a bot task. βcommand 2 20:14, 11 May 2008 (UTC)


 * The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made in a new section.