Wikipedia talk:Good article statistics

Good work on the graph

Very attractive, very cool. Ideogram 23:14, 2 July 2006 (UTC)

Since it's showing change over time, wouldn't it be better represented as a line graph? Schi 04:02, 9 November 2006 (UTC)

Updated

Updated the stats to Nov 06. There is no data that I can find for the exact number of articles on Wikipedia in September and October - the old versions of pages show the current stats script, ie they show the current number of articles even if the old v is from months ago.

Also, I don't really understand what the bottom grpah is getting at, so I left it alone. Chrisfow 21:01, 14 November 2006 (UTC)

I just updated both plots. The bottom graph shows the proportion of GA or FA articles to the total number of articles. It looks like the proportion of GA has peaked as Wikipedia is growing exponentially. --RelHistBuff 10:37, 17 November 2006 (UTC)

December 2006

I just updated the table with GA and FA numbers. Does anyone know where to one can find the total article number? --RelHistBuff 11:26, 1 December 2006 (UTC)

The only place I know is WP:Size. Geometry guy 01:51, 3 February 2008 (UTC)

Not linear

GA growth is no longer linear. The number of GAs per month is steadily increasing. I'm not strong enough in statistics to get much more specific than that, but it's obvious that when this page says the growth is linear, it's wrong. Wrad (talk) 18:45, 10 August 2008 (UTC)

It is approximately linear, with seasonal variations and occasional blips. The monthly growth (from the table) this year has been 201, 143, 200, 215, 200, 170, 205. I don't see steady increase here. Eyeballing the bar charts supports this. I also plotted a graph which suggests linear growth since February 2007. Geometry guy 19:54, 12 August 2008 (UTC)

When transformed to a simple 12 month moving total the growth in number of GAs per month presents as linear. (12 MMT's are a simple way to control for seasonality and the occasional outlying data point without employing more sophisticated statistical techniques. Box-Jenkins, anyone? I thought not.) GA's as percent of total articles also appears to be growing at a linear rate. That is a telling statistic. I long for the day when 10% of articles are GA-class. Majoreditor (talk) 15:03, 18 October 2008 (UTC)

I agree with this analysis. I think we have enough data now to argue that the monthly growth is increasing approximately linearly and I have updated the page accordingly. Geometry guy 20:19, 6 November 2008 (UTC)

Like what you've done with the graphs, much improved. Lampman (talk) 23:49, 2 December 2008 (UTC)

update of data

Hi, I am unable to update the data even though I would like to. My problem is to identify the number of good articles at a certain date, i. e. at the 1 May 2009. I would greatly appreciate it if somebody helps the cause. Thank you and best regards --Marbot (talk) 16:03, 31 May 2009 (UTC)

Can you not extract it from the history of {{GA number}} or from Category:GA-Class_articles?  Skomorokh  16:08, 31 May 2009 (UTC)

Hi Skomorokh, the history of {{GA number}} stopped in late March 2009. I wonder how it is updated now? Category:GA-Class_articles works only if you have a look on the exact day, which is a problem for me. Regards --Marbot (talk) 16:05, 2 June 2009 (UTC)

I've had the same problem. I updated the number for 1 April, but realised later that it was probably inaccurate. I just don't know how to find a historic GA number. Lampman (talk) 23:06, 2 June 2009 (UTC)

Look at the history of WP:GA. You could count the articles on a historical version. However, you might notice the edit summaries have counts. Gimmetrow 03:31, 3 June 2009 (UTC)

I asked the question over at Wikipedia:Village pump (technical), and apparently there is no way around this. The number is "no longer...updated by a bot but suing an expression", whatever that means. I did what Gimmetrow suggested, and I got the following numbers:

1 April - 6621 (+258)
1 May   - 6811 (+190)
1 June  - 6987 (+176)

This was done by transferring the list to a Word-document, and searching for every instance of "[[" (first removing "[[Image:...", there are 42 of these). These numbers are not quite accurate, which can be seen by the fact that the number at the end of May is higher than the current number. Apparently people haven't been good enough at removing delisted articles. If these numbers are to be believed, there seems to be a declining growth rate, but I think much of this can be explained by increased activity over at Sweeps, where a lot of articles were delisted last month. Lampman (talk) 11:02, 5 June 2009 (UTC)

"You might notice the edit summaries have counts". April = 6582 (=6600-18), May = 6775 (-1) , June = 6946 (+2). Gimmetrow 15:56, 5 June 2009 (UTC)

Ah, now I see what you're saying. I didn't notice this at first since very few editors do it. That should give 219, 193 and 171 instead. Also, there appears to be about 40 non-GAs that are listed on Wikipedia:Good articles. Lampman (talk) 19:04, 5 June 2009 (UTC)

There are some articles listed twice, you mean. Gimmetrow 22:55, 5 June 2009 (UTC)

Could be one or the other... Lampman (talk) 02:41, 6 June 2009 (UTC)

Thank you for dealing with this problem. What about these 40-something articles? --Marbot (talk) 12:00, 7 June 2009 (UTC)

Update

I have taken the liberty of updating the data up until 1 April 2010 and have added a graph showing total number of articles for comparison purposes. –– Jezhotwells (talk) 01:05, 29 April 2010 (UTC)

Great work, I've been missing some updated charts! The recent decline in growth rate is largely due to sweeps, but there will be a big boost this month (I think it went from 8,300 to 9,030.) Lampman (talk) 19:39, 1 May 2010 (UTC)

Update to 1 October 2010. Jezhotwells (talk) 23:17, 1 October 2010 (UTC)

Some optimistic figures

For fun I took a look at some figures and using an optimistic monthly growth of 4.5% for GA's and letting it climb without limits(quite unrealistic but hey!) and it worked out we would reach 100K GA articles in 5 years time and a million in 10 years time. Regards, SunCreator ^(talk) 01:37, 28 November 2010 (UTC)

Any thoughts on why GA is increasing so much faster than FA?

Not surprised there are more GA than FA, but wonder the reason for the rate difference even on a relative basis?TCO (talk) 14:02, 15 December 2010 (UTC)

Maybe ask this question over at WT:FA?. My understanding is that FA has become considerably harder over time. Especially the referencing and the completeness, basically Wikipedia:Featured article criteria 1c is very difficult for many articles. While GA hasn't changed much in recent times. Regards, SunCreator ^(talk) 00:18, 16 December 2010 (UTC)

That's true, FA criteria are rather stringent and require a lot of work. One would expect that the fast growth of GAs would mean more "fodder" for FA creation, since GAs are the most obvious candidates, but apparently this is not happening. My guess is that most people who create GAs go on to create more, rather that take one of these GAs and make it to FA. I don't think there's anything wrong with it - in fact, in the long run, and by sheer volume, GAs will contribute much more towards the quality of Wikipedia content than FAs. GregorB (talk) 08:43, 16 December 2010 (UTC)

There are several questions here. The most basic concerns the growth rates: over the past 4 years, the number of GAs has been growing by an average of 185 articles per month, while the number of FAs has been growing by about 40 per month. Both processes are short of reviewers, and it is not clear which has most reviewer resource, so perhaps the main factor is that listing the average FA involves 4-5 times as much reviewer effort as the average GA. That would be a plausible explanation, as most GAs listings only involve one reviewer, whereas FAs involve several, and the criteria are much more exacting. Some GA listings involve second opinions, reassessments, etc., but this adds only a small overhead to the overall reviewer time. In contrast, FAC aims to get it right the first time, and so FARs are comparatively rare.

Another question (which I find more interesting) is whether the GA growth rate is static. The FA growth rate appears to be, but GA processes have been designed over time to scale: because most listings only involve one nominator and one reviewer, with nominators encouraged to review, more articles means more nominations means more reviewers means more listings. Unfortunately the graph of GA growth is rather uneven, due to quality control concerns expressed at various times in the history of GA: the two declines in the monthly growth are associated with (1) the introduction of GAN and the earliest quality control measures, and (2) GASweeps, which diverted reviewer effort to check early listings. Nevertheless, it is clear that the GA growth rate now is around 200 articles per month, whereas 2-3 years ago it was more like 150.

One final question is how the review processes limit nominations: with 1000s of editors and millions of articles, why are there not more FACs or GANs? How exactly does the limited capacity of the processes encourage or discourage nominators from using them? Geometry guy 22:36, 16 December 2010 (UTC)

Unless there is an expanding (or long too point of discouragement) backlog, then I don't think you can say reviewers are limiting things. Just look at that to see. I don't really hear people worry much about backlog (on the submitting side). The bigger concern is passing.

It would be interesting to see queue length over time broken down by subject area. I suspect that military history, movies, literature, and highways have a much higher throughput and lower wait time than the other areas. It is possible that the wait times have reached counterproductive levels in some subject areas, but not in others. By looking at just the total GA backlog report, these effects would be hidden. It is a chicken and egg question -- do we get a lot of GA nominations in certain areas because there are reviewers keeping up with the queues in those areas, or do we have fewer GA nominations outside those areas because people gave up on the length of the queues. Racepacket (talk) 17:55, 21 January 2011 (UTC)