User:Sj/Newpages
from 2005 but quite relevant to current flagged-revs discussions about loosening restrictions on who can edit what.
This is a very preliminary collection of data and partial results; from a test that does not cover a carefully-selected cross-section of edits or times; with significant periodic variances that have yet to be factored out. It is slightly, but not much, better than hearsay. However, I include the raw data used wherever possible so you can decide for yourself how far to trust the interpretations; and can formulate your own vague-but-true statements based on same. If you'd like to split up a larger data-analysis along these lines, write me about it.
Data scour'd from checks through the last few thousand creations / deletions, as of 09:15 to 09:20, 5 December 2005.
Overview
New pages created by anons
Good or bad?
- Jwales said he heard that 5000 pages were created a day and only 1500 survived, with most created by anons being killed. This is an inflated count due to deletion-logs including all namespaces, while newpages counts only the article-namespace; and many people are on image-deletion binges this season. ~40% more new articles are kept than are deleted each hour.
- Mav suggested that "almost all" new pages by anons are crap. Let's have a look:
Hypothesis : Not counting speedy deletions,
- a) a significant fraction of new articles are/were created by new users / IPs / people with no interest in creating an account.
- b) most of these articles are kept.
Forcing users to create accounts / log in before creating pages has
- c) reduced the rate of creation of good new pages
- d) reduced the rate of speedy deletions
Test : Look at 500 new pages, starting on 01:15, 3 December 2005. The first 100 created by anons are listed below. See how many were anon-created, how many survived / were merged / got cleanup tags of var kinds / were deleted.
Results
Creation/deletion rate
Anon page creation was turned off at 14:00 EST, Monday 5 Dec. Comparing rates for the same time-frames before and after the switch, on Sunday and on Monday, we find the following very approximate estimates (noting that Sun/Mon edit patterns are not generally identical, and waiting on better stats from Monday 12 Dec):
summary of short-term changes:
- Newpage creation: down 25% (after removing speedies)
- Page deletion: down 45% (regular AfD + speedy deletions)
- User creation: up 130%
data
Page Creation[1]
14:00 - 24:00, 4 December 2005 : 1060 pages (106/hr). 45% created by IPs.
14:00 - 24:00, 5 December 2005 : 777 pages (78/hr). 0% created by IPs.
Page Deletion (Article namespace only)
20:00 - 10:00, 4-5 December 2005 : 759 deletions, and 14 hrs, between them; 54/hr
20:00 - 10:00, 5-6 December 2005 : 409 deletions, and 14 hrs, between them; 30/hr
User Creation
15:37, 5 December 2005 - 11:29, 6 December 2005 : 5000 users (20 hours; 250/hr),
142 blue-linked as of 11:30.
04:00, 2 December 2005 - 02:49, 4 December 2005 : 5000 users (46.8 hrs; 110/hr)
[1] 4 days out (discounting anything speedied in the intervening 4 days), around 45% of new articles were created by anons. Of these, around 80% are usable articles (the other 20% end up being deleted or merged). Most are stubs, and most do not know/conform to style guidelines.
100 new pages by anons, before the switch
- 10/23 : Jaew bong s, Mouvement de Liberation National du Quebec s, Effort Values (moved to Effort Values (Pokemon)) cs, Thai basil s, Yard long beans s, Shiga Shigetaka s, Lemon basil s, SelectSoft Publishing, Non trivial m, Jasmine rice s.
- 20/44 : Richard Chandler (American author) d(non-not), Fat Tulips, Massaman curry s, Sirena Irwin s, Sean Dillon s, Eric Buhain , Teakettle principle d (unver), Panang beef s, Joey Hawkins d(non-not), Keota, Colorado s
- 30/64 : Neua pad prik s, At Dawn m, Modern Skirts , Knights of the Round m, Palm sugar s, Blow Me Away s, Kritter s, Vardaan Hospital s, World Net Daily controversy m?, Little Picacho Wilderness s
- 40/98 : Nigger-lover dr, NewsMax misinformation d, Brachylaima , 2,5-dimethoxy-4-ethylamphetamine , Paul DeGeorge m, Bob Holt s, Gostomel Airport , River Bottom Nightmares d, Can seamer , Helen of Troy Faustus d
- 50/124 : Erythorbin acid s, Vanity/Nemesis s, Ayvaini Cave ss, Hip-hugger , David O'Sullivan s, 666 Satan s, Shane Thomson s, David Ross (actor) s, Adelheid Bernstein s, Fen Cresswell s
- 60/145 : Jesse Merz s, Sandsnork m, Cadre party s, Ultimate Wolverine Vs. Hulk us, Burnside High School s, Mohammad Ramzan s, Confederate Motorcycles s, Universal Peace Federation , The Murasaki Shikibu Collection s, My love virus d
- 70/172 : Lynne Miller s, Harun Tekin s, Moonrocks sd, Houston Downtown Tunnel System s, Evil Jim sm, Charles Fritts , Ruotomieli , La Planeta De Agua sm, Number Four s, Live in Hollywood
- 80/195 : Toxic (disambiguation) s, Alkiviadis Diamanti di Samarina s, Cellular Authentication Token s, Golden Rivers Football League s, Henry J. Waternoose , Isopycnic centrifugation s, Nordic combined at the 1994 Winter Olympics, Ad Santel, Ski jumping at the 1994 Winter Olympics , Roadvale s
- 90/213 : Mikey Bustos s, Earnest Allen s, Rob Milanese , Speed skating at the 1994 Winter Olympics , Clay Rush s, Spacedaily, Peter Alfonsi s, Hook Nose Mike s, Nelson Garner s, Freestyle skiing at the 1994 Winter Olympics
- 100/228 : Tim Carter s, Simon Mansfield s, Toniu Fonoti s, Lochschmidt s, S.P.G., Miss South Carolina USA, École Mondiale World School sd (non-not), Green Mountain Railroad s, People in politics in 1789 , Steffdj d(vanity)
Totals: 11 d, 7 merges, 65 other stubs (with ~15 cleanup/wikif/unsourced notices).
- Comparison: 20 blue-user articles from that time; half of them non-stubs.
- Lewes local elections, Flexible s, Hendrick Hudson, Lord of the Rings (film) s, Architecture of Mac OS X, 7 mujeres, 1 homosexual y Carlos, Terry Jenner, Circumstance r, Edward van Sloan s, Australian Property Legislation s
- John Sulman, Hebryzelmis s, Ma Huan , John, I’m Only Dancing , Sunderland station , Jacqui Jackson , Gordon Chater s, Petru Aron , Helioscope s, Vigilance committee (trade union) s
Creation/deletion rate
Anon page creation was turned off at 14:00 EST, 5 Dec. Let's see how that affected other core Wikipedia processes! All dates/times below are EST.
Very approximate estimates:
- Page creation: down 28 pgs/hr
- Page deletion: down 24 pgs/hr
- User creation: up 140/hr
New page creation rate
- 14:00 - 24:00, 5 December 2005 : 777 pages (78/hr)
- 14:00 - 24:00, 4 December 2005 : 1060 pages (106/hr)
Deletion rate
Of the last 5000 deletion-log entries (23:21, 3 December 2005 - 11:52, 6 December 2005),
- 51 were actually "restored" log-entries; generally from a delete/restore pair to fix a history or double-check deleted content.
Removing restores and associated deleted pairs, we have 4908 entries. The non-article deletes break down as follows:
Image: 1695 Category: 181 Template: 208 Talk: 131 User: 84 Wikipedia: 27 Image talk: 24 Category talk: 11 User talk: 10 Wikipedia talk: 1 Template talk: 3
This leaves 2500 log entries for article-namespace deletions. Let's leave two hours lag time after the switch, which both changed community dynamics/expectations, and saw a slight rise in site-visits due to the CNN spot at 15:15.
- 20:00 - 10:00, 4-5 December 2005 : 759 deletions, and 14 hrs, between them; 54/hr
- 20:00 - 10:00, 5-6 December 2005 : 409 deletions, and 14 hrs, between them; 29.8/hr
An extra 24 deletions/hr when anon could make new articles. Makes sense; they are being discouraged from adding both good and bad articles.
Finally: Investigate 100 deletions from before and after the change. Break down their origins into Anon/redshirt/blueshirt/collab.
Before the change : 14:00 back to 13:32 for 50; unusual activity (avg: 54/hr).
1-10:
Rosa Parks Peace Prize - ? - nons - 1 anon Genetically_identical - 4m+ - redir - 2 blue The_Jimmy_Timmy_Power_Hour_3 - 8x in 3 weeks - no-content about possible future episode - 4 anons & 2 blue Susannah_Wine - :30 - nons/gay - 1 anon The_drogs - 3:14 - sports/gay - 1 anon What is chromatography? - 15d - redir - 2 blue Egg (zoology) - 40d - redir - 1 anon RCAF_Statio_Mont-Joli - 15d - redir - 1 blue, (self-nom for del) Arun_venkatadri - :45 - nn bio - 1 anon Ass (disgambiguation) - 40d - redir - 1 anon, rfd Grass (disgambiguation) - 40d - redir - 1 anon, rfd
2-20:
Quigo, Inc - :25 - promo/cv - 1 anon Hippolatry - 20d - neologism - 1 blue (self-nom) Steak_Nuggets - :03 - vandalism - 1 anon Angzehornbroals - 1d - wrong name - 1 blue VENTROMEDIAL PREFRONTAL CORTEX - :05 - test - 1 anon The mycenean temple of kadmos - :09 - test - 1 anon Jeremy potter - :01/:22/:05 - vanity; 3x recreated - 1 anon Albert_Ferrer_Florit - 4d - vanity - 3 anons Kelsey is weird - :06 - gay - 1 anon Adeline dozois - 5:00 - vanity - 1 anon, 1x recreated
3-30:
Arnold Renz - :06 - joke - 1 anon List of Halloween costumes - 3m - stub-list - 1 anon Jonathan Yee - :05 - test - 1 anon Pubing - 1:50 - nonsense - 1 red Profidex - 6:30 - ad - 1 anon Nathan stradeski - :04 - nonsense - 1 anon Ijuki - 4:00 - test - 1 anon Ignition Cloth - 4:50 - nn - 1 anon Brent holmes - 3:15 - gay - 1 anon, re-creation Joel Merrick - 1:50 - test - 1 anon
4-40:
Colling, Michigan - :10 - test, blanked by creator - 1 anon
Current events-preview-wikipedia,freeencyclopedia - 3:10 - test - 1 anon
Manoil - :07 - gay - 1 anon
American Religious Identity Survey - :05 - test - 1 anon
WWCL - 3:25 - "this is a test please don't ban me :(" - 1 red
Shoutwire - 0:04 - attack - 1 anon; converted by 1 blue into real art
Emilíana Torrini - ? - move - 2 blue
Va senators and house of representatives - 2:25 - test - 1 anon
Scott matekaitis - 1:45 - vanity - 1 anon
Robert Cooper (Software Engineer) - :01/:24 - vanity - 1 red, 1 anon
5-50:
Sean stevens - :01 - vanity - 1 anon Hindsight(Lad) - 50d - dup - 2 blue Quinlivitarianism - :22/1d - vanity - 2 anons, afd Ador Welding Limited - 5:36 - ad - 1 anon Murray 1 - :00 - nonsenes - 1 anon SCinet - :01/:01 - company spam? - now an art Garden Gnomes Of Doom - :12 - nn band - red Forum 168 - :10/1:20 - nn - 2 anons; now 3 anons + 1 red, on afd Spanish herbs - :12 - zealous stub-deletion - 1 anon Esher college - :04 - overzealous stub-deletion - now a redir to Esher College (stub)
New user creation rate
- 15:37, 5 December 2005 - 11:29, 6 December 2005 : 5000 users (20 hours; 250/hr), 142 blue-linked as of 11:30.
- 04:00, 2 December 2005 - 02:49, 4 December 2005 : 5000 users (46.8 hrs; 110/hr)
General Notes
- Deletion log: 18:10, 2 December 2005 (63 hrs. 50/5000 'restored' rather than deleted; 600/5000 Images (no source drive); 10/100 non-main namespace... more complete breakdown above. About 50% of deletions are not article namespace.
- Upload log : back to 21:41, 1 December 2005 (84 hrs; 60/hr)
- Upload deletions : 600/63 hrs (10/hr; see deletion log)
- Newpages log: 01:15, 3 December 2005 (5000/56 hrs are kept; ~90/hr)
These rates don't quite match the Zachte statistics; likely an indication that the sample size is too small (doesn't cover an entire day, in many cases; certainly not an entire week-cycle).
Current 'new user' pageflow for page creation
This is painful for new users; you lose your place and have to pass through many different pages... much more than a 15-second patch on the previous anon-page-creation process.
- Follow a redlink while reading.
- "Page creation limited
Wikipedia has temporarily restricted the ability for unregistered users to create new pages. You may list the content you wish to have created. You may also log in which enables you to create new pages. - Go to 'login' -- but not directly to account creation page. "Don't have a login? Create an account]."
- Figure out that you can't just log in, you have to follow that "create an account" link; follow link
- Get redirected to the Main Page. Try to figure out where you were before, when you started to edit.
Or, at step 3 above, go to Wikipedia:Articles for creation, which has received around 3-4 requests an hour since it was set up. Compare to