@Tavix: When you say "page views are unreliable" what exactly do you mean? Is there a guideline somewhere that either states that stats are either relevant or irrelevant with regards to this kind of argument? Since my short time back here at RFD I have seen the same arguments between you and @Thryduulf: go back and forth - surely there is a WP policy on this exact matter and if not would it be a good thing to establish convention on? JZCL 19:44, 10 October 2018 (UTC)[reply]
@JZCL: Tavix seems to believe that the page view statistics record something other than what they actually record (human uses of a redirect) and/or that recorded human uses of a redirect mean something other than that a redirect is used, and/or that even if a redirect is used by humans this does not mean that the redirect is useful (or potentially so) if he thinks otherwise. None of these things are backed up by any evidence that he has ever given or that I've ever found elsewhere. It's true that just because n people viewed a redirect, that all of them found what they are looking for, but in contrast to Tavix I believe that (a) helping some people is better than inconveniencing everybody (several hundreds of people on some occasions), and (b) where there is only one plausible target (as with this redirect) it is almost certain that everybody did get to the right target (especially as is the case here where we can see the context of some links). (a) is a value judgement about which we seem to fundamentally disagree (I've never seen a rational explanation of the benefits for Tavix's view though); (b) is not completely objective, but it is when is no disagreement on the plausibility of targets (as in this case) - but Tavix. I do not see a possibility of a guideline emerging when there are such strongly held opposite views. Thryduulf (talk) 20:06, 10 October 2018 (UTC)[reply]
(edit conflict)WP:PAGEVIEWS is a helpful information page which discusses the limitations of pageview statistics in depth. A good thing to keep in mind from that page is: If a page's stats are low, it is not a reason to consider it for deletion, and if high it is not a reason to save it from deletion. --Tavix(talk) 20:12, 10 October 2018 (UTC)[reply]
@Tavix:@Thryduulf: Thank you both very much for your responses. Tavix, I have read the page you sent; while it does say that pageview analysis has its limitations the page seemed more appropriate for articles rather than redirects. Do you believe there is some ulterior motive whereby certain users are deliberately searching for redirects to increase their hits? And assuming to the answer to that last question was no, what exactly in the article would lead you to believe that the stats on these pages were anything other than genuine attempts to look for articles?
With regards to this as a matter of policy, I don't understand why no such guideline should exist. It would save this sort of repetitive and time-wasting discussion being had on every 5 or so redirects. Would it be something we could consider at RFC, or something we could just discuss on the RFD talk page? JZCL 20:25, 10 October 2018 (UTC)[reply]
Yes, that page is primarily written for articles, but that doesn't negate the limitations of pageviews that are discussed therein. You simply can't say that X page views means that X people found a certain redirect useful. There are several ways to register a page view, and the only one we really care about is one where a reader actively uses the redirect to find the page we are looking for, which we have no way to ascertain from simply looking a tool. We don't care about people using the redirect but are not helped by the target, for maintenance purposes, bots (we don't require CAPTCHAs, so you can't tell whether or not a pageview is a bot unless it's from an account marked with a bot flag, which are discounted), misclicks, or even Special:RandomRedirect. And by the way, I agree with Thryduulf that an RfC would be a waste of time. --Tavix(talk) 20:38, 10 October 2018 (UTC)[reply]
(edit conflict) Bots are indeed counted separately, as are spiders (this identification is done, in part, based on user agent and part on request behaviour AIUI). Evidence from unquestionably implausible redirects shows that they receive low single digit numbers of hits per year and are equally likely to be the result of maintenance, misclicks and Special:RandomRedirect. This leaves tens or in some cases hundreds of other hits that are, given the complete absence of evidence to the contrary, somewhere between very probably and certainly human. However, even if 90% are not using it to find content (which is extremely implausible for the reasons just given, it's more likely to be <10%) that leaves 10% who are using it to find content. These are real people who will be inconvenienced (in some cases very significantly) by deletion - which will not benefit anybody (those we've discounted will not see any impact either way - someone who misclicks on this redirect is equally likely to misclick on some other title instead). This is the key pint Tavix repeatedly chooses to ignore: if a redirect helps some people and is neutral to everybody else, then deletion is harmful. Page views are a key tool in determining what is and is not useful. Thryduulf (talk) 20:51, 10 October 2018 (UTC)[reply]
The tool can make an educated guess on the identification based on a number of factors, but that still isn't completely reliable. To give a few examples of the other extreme: WP:TOPRED shows which redlinks received the highest number of pageviews in the preceding 7 days the last time it was updated. If you took the stance that X number of pageviews means a page is useful, then all of those redlinks should be useful for something, right? Hopefully we can all agree that Rtwpsom2 (trillionaire) how he made his fortune would not be useful as a redirect (perhaps to Trillionaire?). Another example, these Arabic-language pornographic terms were all created as redirects because they received a lot of pageviews. However, they are clear violations of WP:RFOREIGN and someone searching that would be looking for actual pornography, which Wikipedia does not provide. In these examples, I've used other means to determine that redirects with high page views aren't useful. Similarly, in RfD discussions editors should use other means to draw their conclusions on whether or not a redirect should be kept instead of using arbitrary quantities. --Tavix(talk) 21:06, 10 October 2018 (UTC)[reply]
