Wikipedia talk:Awareness statistics

Alexa stats
I have added a data table, and the Google search used to generate the data in it. However, I do not know if this was the same query used by the original author of the graph above. The Anome

I have added a graph of the Alexa statistics. Mike Storm

Would Technorati information be appropriate here? It tracks links from weblogs. Dreamyshade

Frequency of updates
I am not too happy with having one update of the Alexa stats every day. Imagine what the page would look like now if we had one entry for each of the one thousand days since this started! I suggest one of following approaches: either update the stats only once or twice per week, or update them more often but remove some of the entries later so that no more than one entry per week remains. --Mihai 22:30, 28 Sep 2004 (UTC)


 * Mihai makes a good point. I've been generating the graphs for each section whenever new data is posted, but I've stopped doing it so frequently for that section.  For those who are interested in where it stands "today", perhaps the last value in the table could be updated frequently, but only retained and the table  lngthened each month or couple of weeks.  Could a sub-page be made to hold the short-term data if it's that important?  --Kbh3rd 03:05, 29 Sep 2004 (UTC)

--

I have moved the prediction graph here -- we probably need to do this again, as the current fit is so biased by the clearly visible outlier as to be meaningless -- it also seems clear that the fit should be done in the log domain, as this is an exponential growth process. 213.253.40.223



Alexa traffic by month, calculated by taking the average ranking for each month and using polynomial curve of best fit.


 * Predicted average ranking for the month of January 2005: 538
 * Predicted average ranking for the month of June 2005: 441
 * Predicted average ranking for the month of January 2010: 135
 *  Predicted date when Wikipedia will be the #1 site on the web: August 2484

(predictions and graph will be constantly updated when new data becomes available)

There was formerly a flame here (my own). I have removed it, and I apologize. Mike Storm 23:53, 19 Jul 2004 (UTC)

--

Can both of you please take this to modelling Wikipedia's growth, and please only post graphs of actual known information to this article. -- Karada 16:53, 30 May 2004 (UTC)

Time to go logarithmic with Alexa Reach?
What do you think? Should we use a logarithmic scale with bezier smoothing? Or stick with the way it's been, which currently shows only a biweekly-ish sampling of the data on a linear scale? --Kbh3rd 17:46, 21 Oct 2004 (UTC)

Misunderstandings
Obviously both of you have misunderstood the other. Mike, Dreamyshade was only trying to help, and obviously you updated your graph in a different way than he intended. Being rude does not help anyone. Dreamyshade, could you please provide some more information as to how the graph could be improved, possibly generate your own graph? We need to get this resolved quickly. I agree with Karada, it should probably be moved to a different page. supadawg 02:03, 31 May 2004 (UTC)

More misunderstandings! I only posted the Technorati question. I think Mike accidentally attributed the graph discussion to me. Dreamyshade 23:14, 31 May 2004 (UTC)

Consistency
If you could please try to keep the month names consistently unabreviated or abreviated, but not mixed in the same table, it would make it a lot easier for me to generate graphs with gnuplot. "July" should be followed by "August", not "Aug", though "Jul" should be. Yes, I could easily preprocess them, but that adds another unecessary step. Consistency is a good thing, anyway. Kbh3rd 04:18, 14 Aug 2004 (UTC)
 * I've gone and made a script to normalize the month names for processing the graphs. But it is still good form to be consistent within the same table. Kbh3rd 14:53, 1 Sep 2004 (UTC)

Usenet posts about Wikipedia
What happened in June with Google's usenet stats? Did Wikipedia really see a 10-fold increase in mentions in one month?? I've held off on generating a new graph for this table because it seems suspect. Kbh3rd 14:53, 1 Sep 2004 (UTC)
 * Well, I went ahead and generated a new chart anyway, using a logarithmic Y-axis. The old one is here. -  Kbh3rd 16:28, 1 Sep 2004 (UTC)

Just a try: august 1--> august 19(3880 hits) + august 20 --> august 31(2290 hits) gives 6170 hits for august instead of the august 1 --> august 31(38.500 hits).

I assume the first to be more correct than the latter and adapted the number in the text accordingly. Donar Reiskoffer 07:56, 24 Sep 2004 (UTC)


 * Those numbers certainly look more believable. I've regenerated the chart.  Don't forget that you may need to force a refresh to get the old image out of the cache. --Kbh3rd 16:46, 24 Sep 2004 (UTC)

recent leap in alexa rank
Is it coincidence that other sites like bbc.co.uk or reference.com laso have a large leap in alexa rank? See

Donar Reiskoffer 07:35, 9 Oct 2004 (UTC)


 * Maybe Google have changed their algorithm or weightings in a way that favours factual/reference sites in some way? -- The Anome 10:12, 9 Oct 2004 (UTC)


 * I believe the dramatic rank improvement for Wikipedia, reference.com, britannica.com etc. has to do with a change in the Alexa interpretation of visits/users statistics. I see that for all these sites the number of users ("reach") has increased, the number of visits per users has decreased, and the product, i.e. total visits, has not changed much. The undisclosed Alexa recipe for the total rank seems to results in a better rank when the reach is increased at the expense of "visits per user". My two cents. --Mihai 18:23, 13 Oct 2004 (UTC)


 * I am pretty sure its because its back to school time. Page views go up because of all students, new article creation should go up becasue of lots of smart college students with free time. -Ravedave 06:37, 2 October 2005 (UTC)

Code for generating the graphs
I've posted the shellscript and gnuplot code I use for generating the graphs on this page to the discussion pages of the separate images. If I get hit by a truck or join a monastery someone should be able to pick this right up. I would ask that folks don't just go willy-nilly updating these images, however. I currently intend to keep them updated at least monthly, and possibly more frequently for those that warrant it. And it is possible that different versions of gnuplot on different platforms could have slight differences in their renderings, and I think it important to have all the graphs on the page look visually similar. (I use gnuplot Linux version 3.7.) --Kbh3rd 17:02, 11 Oct 2004 (UTC)
 * For what it's worth, I am not seeing either graph using firefox 0.9.2Pedant 15:15, 2004 Oct 20 (UTC)

PubSub figures
Hi all.

I ran across PubSub Link Ranks doing my evening's reading; it seems to be sampling a limited set of the web (it looks like weblogs only) and constructing a chart of the most commonly linked pages, although it doesn't provide any "historical" data - or, indeed, anything bar a numerical position. Still interesting to see, though. Shimgray 22:28, 13 Dec 2004 (UTC)

Blogpulse
I think we should add a link or hotlink to the new trend graphs at Blogpulse. This would represent how actively people are talking about Wikipedia. Wikipedia Trend Graph

Google PageRank
The page currently states: 16 July 2005,    9/10

I'm curious as to how you people already know that Wikipedia will have a 9/10 PageRank nearly three weeks in the future. :) --Pidgeot (t) (c) (e) 28 June 2005 05:04 (UTC)

Breaking 50?
Do you think Wikipedia can make it past that threshold? Fuzheado | Talk 9 July 2005 02:18 (UTC)

The real question is whether it will break the 10 barrier. I predicted about 6 months ago to some friends that Wikipedia will eventually become the no. 1 site on the internet. Events and stats seem to indicate that no ceiling has been reached yet. --One Salient Oversight 23:28, 11 July 2005 (UTC)


 * I'm not ready to predict a number one slot, but there is certainly a strong case for it being in the top ten. I think as long as there is Gmail, Yahoo Mail and the like, those sites will occupy the top simply because folks search and check email each day more than anything else. But after that, certainly the body of reference works that Wikipedia/Wikimedia has (in so many langauges) will be useful to any and everyone. Fuzheado | Talk 06:49, 12 July 2005 (UTC)


 * The climb is all uphill from here on out. --Alterego 15:09, July 12, 2005 (UTC)


 * One important thing to realise about these awareness statistics is that the "Rank" figures will increase more slowly now, while the "Reach" figures will be the one to watch. While the intenet has millions of websites, it appears as though the top 50 websites consume the time of quite a lot of surfers. This means that as Wikipedia gets more popular, the journey to the top ten (or top five) will be slow and painful. I'm fairly certain that it will get there though. --One Salient Oversight 07:49, 13 July 2005 (UTC)

18 July 2005 Estimate - Alexa English top 100
Since Wikipedia has disappeared from the English Language top 100, there is a way to estimate its popularity by using the international top 100.


 * 18 July 2005  Apple Computer is #32 on English list.
 * 18 July 2005  Apple Computer is #62 on International list.
 * 18 July 2005  The Weather Channel is #31 on English list.
 * 18 July 2005  The Weather Channel is #54 on International list.
 * 18 July 2005  Wikipedia is #60 on International list - between Apple and Weather Channel.
 * Therefore we can assume that Wikipedia is not #31 because Weather Channel is more popular on International list.
 * We can also assume that Apple Computer website is worse than #32 because Wikipedia is more popular on International list.
 * Therefore Wikipedia is assumed to be #32 on English language list.

--One Salient Oversight 01:35, 21 July 2005 (UTC)

Google Answers using Wikipedia
In the section in this article named "Google Answers using Wikipedia", what is the unit of measure? Million hits? Billion? &minus;Woodstone 21:22:54, 2005-08-21 (UTC)
 * Given the nature of Google Answers, I'd say it's probably the number of answers that mention/link to Wikipedia. In ones. Optichan 18:20, August 22, 2005 (UTC)

Hurricane Katrina Spike
Dear 82.35.34.11,

I've been updating this page for months, and my research into "activity spikes" shows a correlation between world events and Wikipedia activity. As these awareness statistics get updated, the editor sometimes has to ask "Why is there a spike in activity?" I discovered quite early on that the 2004 US Presidential Election, the Tsunami, the death of the Pope and an LA Times article about Wikipedia all occurred around the same time as spikes in activity.

Your original reason - that "the current increase is more likely to be due to the new term - it happened last year too" - is wrong. If you look at the stats, there was no real increase in September 2004 - in fact the "reach per million" table showed a steady decline in awareness during that period. But notice in October that this picked up quickly - which was occurring in the last month of Presidential campaigning.

The current rise in awareness is considered a "spike", not a "continual rise" as you call it. Look at the graph provided by ALexa here. At that graph you will see periods of rapid growth followed by a plateau, which are then followed by another period of rapid growth. The graph shows that a period of rapid growth - a spike - is occurring right now. Why? Because of Hurrican Katrina. If you look at other spikes in activity you will notice that they coincide with major world events.

The damage wrought by Hurricane Katrina is considered a major news event. People all over the world have been searching the internet about it and New Orleans. During this process many have discovered Wikipedia, which is why there has been an 12% jump in the reach since August 24 2005.

And as far as the rest of the world losing interest in Katrina, go to http://news.bbc.co.uk/ to see the front page of the British Broadcasting Corporation. Here in Australia it is still major news - the second item on the radio news this morning was Michael Brown's resignation from FEMA.

--One Salient Oversight 23:21, 12 September 2005 (UTC)


 * Insofar as traffic does spike during realworld events, it is usually a peak that lasts several hours (sometimes less) during the initial onset of public curiousity, and then load evens back out. --Alterego 01:34, September 13, 2005 (UTC)
 * I've been the main editor of the "More popular than" Alexa ranking page (which seems to have disappeards) all year. Your arguments are incorrect. Wikipedia reached new highest rankings at least a dozen times in September 2005, and the increase in the average reach this month was lower in the days when Katina was most in the news than it is now. I am British so I know more about what is in the British media than you do. Katina is still an important story, but saturation coverage is over. 82.35.34.11 20:36, 15 September 2005 (UTC)

Okay then let's try to work this all out. Can you please respond to the following points: --One Salient Oversight 22:43, 15 September 2005 (UTC)
 * 1) Do you agree that spikes in activity occurred during the Presidential Election in October 2004, the Tsunami in January 2005, the election of a new pope in April and LA Times publicity in June 2005? If you don't believe these then you should remove the comments that I put there as well.
 * 2) It is my theory that during these spikes in activity, more people using the internet are exposed to Wikipedia. As a result, more people "stay", and Wikipedia becomes a regular part of people's internet life.
 * 3) The theory explains why it is that Wikipedia goes through periods of sudden growth, followed by plateaus. It therefore stands to reason that Hurricane Katrina caused a lot of internet activity that led to a lot of people visiting Wikipedia. Many of these people have stayed.
 * 4) What did you mean when you spoke about the new "Term" and how it repeated itself in September 2004?

Page split
To make the page more manageable, I've moved all the Google stuff to Search engine statistics. --Michael Snow 06:03, 11 November 2005 (UTC)

Graph updates
I'll be producing an update of all these graphs at the end of this month. I'm not sure what would be the best interval for updating them, but it seems that since we're so close to the end of the year that it would be good to wait until then. If they sit for a while before being updated again, they'll look copacetic ending on a year boundary. -- Kbh3rd 17:19, 7 December 2005 (UTC)

January 2006 boom
Wikipedia's added more points to its record weekly reach in the first week of 2006 than it did in four years from 2001 to 2004.

Alexa page views per million measurement
Measuring the values on the 3-month graph is not accurate. Points in the graph are subject to smoothing and time shifts. 10 or 20 days graph display correct results : Search used. - Marc Macé 20:30, 7 December 2005 (UTC)

Alexa is broken???
As of today (January 27th), the Alexa ranking page for Wikipedia claims 0 page views per million, zero reach per million and no ranking. Makes quite a mess of the graph. It also says it was last updated on January 22nd. I'm certain this is just a problem with Alexa; anyway, it looks as though the statistics may not be updated for a week or so -- Gurch 16:23, 27 January 2006 (UTC)
 * Never mind, it seems to be fixed. (And we reached #16, woohoo!) -- 217.35.96.167 09:08, 29 January 2006 (UTC)
 * That last comment was me, curse this computer for logging me out all the time! -- Gurch 09:09, 29 January 2006 (UTC)

Out of date graphs
It's February now and the graphs made up to 31 December 2005 are yet to appear. If they don't arrive soon I think the old ones should be removed anyway. 62.31.55.223 09:29, 1 February 2006 (UTC)
 * I've updated them now. Woohoo!  I'll probably get around to adding some new graphs for the uncharted data sets, but it might take me a week or three before I have enough time.  But maybe I'll surprise you (and me). --Kbh3rd talk  05:23, 8 February 2006 (UTC)
 * Except that the main data without a graph here is Alexa page views per million, which is actually derived from Alexa's graph, and it hardly seems necessary or even reasonable to rechart data derived from a chart. --Kbh3rd talk 05:30, 8 February 2006 (UTC)
 * Seems they're out of date, again. Time flies like an arrow.  But fruit flies like a banana. --Kbh3rd talk  07:35, 21 February 2009 (UTC)

Continuing boom
Reach increased more on 30 Jan than it did in Wikipedia's first two years and nine months. I'm surprised we are not hitting server capacity problems at the moment. 62.31.55.223 06:47, 2 February 2006 (UTC)

so long
Must this page be 1000 lines long? --Jidanni 2006-04-16

Google Trends
Here's a page showing how the number of searches for "Wikipedia" on google.com has risen. Maybe this will be an interesting graph to use. (RCX 19:21, 12 May 2006 (UTC))
 * Definitely, but Image:Google trends wikipedia.gif has a web-screenshot (fair use) copyright tag, which allows us to use it only in the main (article) namespace to identify the website in question. I've added an external link to the graph, I think that's the most we can do. --Zoz (t) 23:37, 28 July 2006 (UTC)

Wikipedia Self-Awareness statistics
How much longer until Wikipedia becomes self-aware? (I thought this page was a joke in that vein when I first saw the title). -Ich (talk) 08:45, 30 July 2007 (UTC)

Is it me or is this page dormant and now historical
Statistics up to 2007... not really what I was expecting,! Suggest a rename to 'historical ....' or maybe some tag about being out of date Lee&there4;V (talk  •  contribs) 23:03, 20 October 2009 (UTC)