Community Wishlist Survey 2019/Admins and patrollers/Monitoring of new external links

Random proposal ►◄ Admins and patrollers The survey has concluded. Here are the results!

Monitoring of new external links

Problem: Right now, lots of new website links are being added to Wikipedia every day, lots of them are spam, lots of them are sneaky spam. Keeping track of them is hard.
Who would benefit: Patrollers

Proposed solution: There should be a special page that I can see recent new external links that are being added but it will be flooded with lots of trusted websites (like bbc.co.uk that's been used often). So it should have a whitelist of trusted sources to remove them for review list.

More comments:

Phabricator tickets:

Proposer: Amir (talk) 23:21, 1 November 2018 (UTC)[reply]

Discussion

So this would sort of work like AbuseFilter? Admins would have access to editing the whitelist?

(Would we want a central whitelist (sr.se might be more common to link to on Swedish Wikipedia than on German Wikipedia, but it's equally unlikely to be spam in both cases) or a per-wiki whitelist? But that's a discussion for later, maybe.) /Johan (WMF) (talk) 11:34, 2 November 2018 (UTC)[reply]

This already exists: #wikipedia-en-spam^connect and #wikimedia-external-links^connect. MER-C (talk) 17:35, 2 November 2018 (UTC)[reply]

We're also planning a more thorough/direct solution as part of the Knowledge Integrity program. See T199189. Samwalton9 (WMF) (talk) 10:42, 6 November 2018 (UTC)[reply]

@Ladsgroup: Do the above solutions work for you? MusikAnimal (WMF) (talk) 17:23, 13 November 2018 (UTC)[reply]

I doubt, that would target tool developers but I want something for patrollers. Amir (talk) 18:49, 17 November 2018 (UTC)[reply]

I caution that this, if implemented, could quickly become just as problematic as any other automated "filter" system. It can easily become abused, and eventually do more harm than good. Filters are a very steep and very slippery slope towards censorship. Lostinlodos (talk) 01:40, 19 November 2018 (UTC)[reply]

Actually don't see it related to censorship at all. Would just be like what we have today, but minimizing time wasted on URL that are highly likely to be clean Wk muriithi (talk) 15:00, 24 November 2018 (UTC)[reply]

An extension for this is to have a boot that go checking every link once a week. If a link isn't reachable 3 times (3 weeks), just have it listed here so that admins can remove it Wk muriithi (talk) 15:03, 24 November 2018 (UTC)[reply]

Voting

Support Liuxinyu970226 (talk) 03:35, 17 November 2018 (UTC)[reply]
Support Dirk Beetstra ^{T C} (en: U, T) 05:35, 17 November 2018 (UTC)[reply]
Support Libcub (talk) 10:16, 17 November 2018 (UTC)[reply]
Support ديفيد عادل وهبة خليل 2 (talk) 11:08, 17 November 2018 (UTC)[reply]
Support —MarcoAurelio (talk) 11:39, 17 November 2018 (UTC)[reply]
Support Martin Urbanec (talk) 13:48, 17 November 2018 (UTC)[reply]
Support JogiAsad (talk) 18:35, 17 November 2018 (UTC)[reply]
Support As the proposer Amir (talk) 18:47, 17 November 2018 (UTC)[reply]
Support Barcelona (talk) 18:53, 17 November 2018 (UTC)[reply]
Support Hanooz (talk) 19:40, 17 November 2018 (UTC)[reply]
Support Mardetanha ^talk 20:08, 17 November 2018 (UTC)[reply]
Support Yamaha5 (talk) 20:34, 17 November 2018 (UTC)[reply]
Support Fatemi ♪ 20:43, 17 November 2018 (UTC)[reply]
Support JOAN ~ ^(Questions?) 20:52, 17 November 2018 (UTC)[reply]
Support Tom (LT) (talk) 23:07, 17 November 2018 (UTC)[reply]
Support Temp3600 (talk) 05:42, 18 November 2018 (UTC)[reply]
Support Poya-P (talk) 06:11, 18 November 2018 (UTC)[reply]
Support Jules78120 (talk) 09:36, 18 November 2018 (UTC)[reply]
Support فرهنگ2016 (talk) 10:39, 18 November 2018 (UTC)[reply]
Support Cool idea Dvorapa (talk) 16:53, 18 November 2018 (UTC)[reply]
Support — Draceane ^talk_contrib. 17:31, 18 November 2018 (UTC)[reply]
Support Shizhao (talk) 02:33, 19 November 2018 (UTC)[reply]
Support Waddie96 (talk) 07:36, 19 November 2018 (UTC)[reply]
Support ~ Nahid ^Talk 09:25, 19 November 2018 (UTC)[reply]
Support Sounds sensible ·addshore· ^{talk to me!} 09:58, 19 November 2018 (UTC)[reply]
Support BugWarp (talk) 23:56, 19 November 2018 (UTC)[reply]
Support If we can censor those trying to spam us I am all for it. Doc James (talk · contribs · email) 03:37, 20 November 2018 (UTC)[reply]
Support Jamesmcmahon0 (talk) 10:15, 20 November 2018 (UTC)[reply]
Support Lofhi (talk) 17:49, 20 November 2018 (UTC)[reply]
Support Magalia (talk) 21:25, 20 November 2018 (UTC)[reply]
Support CAPTAIN RAJU^(T) 22:22, 20 November 2018 (UTC)[reply]
Support Novak Watchmen (talk) 23:02, 20 November 2018 (UTC)[reply]
Support Tisfoon (talk) 06:12, 21 November 2018 (UTC)[reply]
Support Framawiki (talk) 19:33, 21 November 2018 (UTC)[reply]
Support Nihlus 22:12, 21 November 2018 (UTC)[reply]
Support I think it is a good idea to build a way to monitor additions and removals of external links, but without preventing user actions. This is for patrolling purposes only (analysing the wiki, not controlling it). For implementation, I think it might be easiest to implement this as a Toolforge tool with a background process that subscribes to EventStreams (or maybe Schema:ExternalLinksChange). It could be tracked in real-time with e.g. graphs summarising events per hour and per day. Helping to find domains for which the number of existing links is rapidly growing or shrinking. It would also help to track it by TLD, TLD+1 (base domain) and complete domain, e.g. "sub.of.example.org", "example.org" and "org". Krinkle (talk) 01:08, 22 November 2018 (UTC)[reply]
Support Winged Blades of Godric (talk) 13:27, 22 November 2018 (UTC)[reply]
Support Encycloon (talk) 18:30, 22 November 2018 (UTC)[reply]
Support Arian ^Talk 19:33, 22 November 2018 (UTC)[reply]
Support DannyS712 (talk) 19:56, 22 November 2018 (UTC)[reply]
Support SalmanZ (talk) 21:06, 22 November 2018 (UTC)[reply]
Support It is very logic Emgaz (talk) 14:15, 23 November 2018 (UTC)[reply]
Support ~Cybularny ^Speak? 15:44, 23 November 2018 (UTC)[reply]
Support Ghilt (talk) 13:20, 24 November 2018 (UTC)[reply]
Support A great idea. Ya, its better to list only links for new domains. For trusted domain, its too labour intensive to verify every article, better to audit against a domain and even better, just new domains Wk muriithi (talk) 14:13, 24 November 2018 (UTC)[reply]
Support B20180 (talk) 17:44, 24 November 2018 (UTC)[reply]
Support By erdo can • TLK 09:00, 25 November 2018 (UTC)[reply]
Support — AfroThundr ^{(u · t · c)} 01:34, 26 November 2018 (UTC)[reply]
Support Daniel Case (talk) 03:12, 26 November 2018 (UTC)[reply]
Support Dreamy Jazz (talk) 08:47, 26 November 2018 (UTC)[reply]
Support Equoreo (talk) 09:10, 26 November 2018 (UTC)[reply]
Support - FlightTime (open channel) 22:14, 26 November 2018 (UTC)[reply]
Support Izno (talk) 01:01, 27 November 2018 (UTC)[reply]
Support ifny (talk) 03:15, 27 November 2018 (UTC)[reply]
Support YFdyh000 (talk) 15:10, 27 November 2018 (UTC)[reply]
Oppose This could easily go wrong and flag a legitimate site by mistake. I once had a Palmer Report news link end up being removed for no particular reason. I would hate for that kind of false positive to become the norm. PF4Eva (talk) 08:52, 29 November 2018 (UTC)[reply]
Support Denver20 (talk) 15:30, 29 November 2018 (UTC)[reply]
Support Ldorfman (talk) 20:43, 29 November 2018 (UTC)[reply]
Support Schniggendiller (talk) 10:58, 30 November 2018 (UTC)[reply]