Wikipedia talk:Trust network

Archive of original rejected proposal (before June 2006)

What an English version of the German trust network would look like
For those people whose German isn't quite up to it, this is a summary of how English Wikipedia's web of trust might work:


 * Interested users would set up a user subpage, probably /Trust (shorter than Web of Trust or Trust network, other possibilities exist)
 * That page would usually include a standard template explaining the basics of the trust network and linking back to this page for fuller explanating, and add the user subpage to Category:Wikipedia trust network.
 * They would put in a list of users they trusted (no problem with "distrust" any more). To do this they would link to the equivalent user subpage, whether it exists or not. For instance, to trust Jimbo Wales, you would list User:Jimbo Wales/Trust. This could be done using a template e.g. which might produce Jimbo Wales (talk • contribs • trusts • trusted by)
 * The list might also contain details of why somebody is trusted, e.g. by typing:
 * (founded Wikipedia)
 * (lots of experience)
 * (several FAs on Ancient Greece)
 * (known personally)


 * To find out who trusts somebody, use Special:Whatlinkshere: for instance, the "trusted by" link above for Jimbo Wales was given by a link to trusted by . The standard template for trust subpages could include a "who trusts me" link in the same way.

I would also like to propose a slight change from this German practice. Having the target link to indicate trust set to a page which is likely to contain interesting content in itself is a bad idea. It means any link to that user's "trust" page (e.g. in a link intended to be of the form "hey, look at who this user trusts" rather than "I confirm that I trust this user") becomes an endorsement. It would be better to have the inward links going to an intentionally uninteresting, usually redlinked, user subpage e.g. "/Trusted by" so that the place to see who trusts Jimbo would be trusted by and the template would give "Jimbo Wales (talk • contribs • trusts • trusted by" which looks identical but doesn't link to the "Trusted by" page directly, hence can be used without producing an accidental endorsement. To be certain of issuing an endoresement, there would be an alternative  template, so  would give something "Jimbo Wales (talk • contribs • trusts • trusted by" which looks the same but the space after the user link would actually be the vital "endorsement" link. An easily implemented feature would be.

If any interest gets expressed in this, I may convert it into a formal proposal, to replace the experiment currently gathering dust. Some details would need to be finalised and agreed upon (e.g. wording for a standard template) and whether to go with the original German system of links or with my proposal which aims to stop less relevant pages showing up in the whatlinkshere "who trusts me" link. TheGrappler 04:32, 2 June 2006 (UTC)


 * Sounds like a good idea to me :). Thue | talk 20:19, 2 June 2006 (UTC)


 * Certainly interesting. I'd suggest you go ahead and do a writeup, some people will undoubtedly try it. —Simetrical (talk • contribs) 19:55, 4 June 2006 (UTC)


 * As noted above, trust networks are not a particularly good idea; too divisive, promote factions, too easy to game. I oppose this becoming codified as policy or guideline. FeloniousMonk 22:35, 6 June 2006 (UTC)


 * Great idea -- for anyone seeking internecine wars, backstabbing, vendettas, the establishment of cabals and cliques, extreme divisiveness, abuse, and dishonesty in the form of gaming. That the German wiki has it is irrelevant -- different culture.  Like FM, I oppose such a concept becoming either a guideline or a policy.  (Side note: it reminds me of the "Circle of Trust" in Meet the Parents).  &#0149;Jim 62 sch&#0149;  09:03, 7 June 2006 (UTC)

Less binary?
I'm a little concerned that our only choices seem to be "I have no opinion about this person" or "I trust this person". What if we only kinda trust somebody, or only in matters related to certain topic areas? What if we strongly distrust somebody? Deco 22:01, 6 June 2006 (UTC)

Understand mechanics, but what's the point?
I understand the mechanics of this proposal, but without any concrete proposal for a use I don't really see the point. If 10,000 people had more or less populated /Trust pages, then what? Is it used for anything at de.wikipedia? -- Rick Block (talk) 22:53, 6 June 2006 (UTC)
 * I could see it as a more useful metric to consider in adminship proposals than silly things like edit counts. Also helps to tell the trolls from the good editors when trying to figure out whether edits are vandalism. Deco 22:57, 6 June 2006 (UTC)
 * Is it used this way at de? Do RfAs typically include something about a user's trust network rather than edit counts (this user's directly trusted by n users and less than 3 away from 90% of the current admins, or alternatively, oppose - doesn't meet my criteria of trusted by 20 users and no more than 2 away from 10 admins)?  Would you look at a user's trust list when doing RC patrol rather than the specific edit?  Does CDVF have a auto-trust level setting (an edit by anyone within 3 levels of my immediate trust list doesn't even show up)?  I'm not saying it's useless - just curious what immediate uses, if any, are envisioned. -- Rick Block (talk) 23:36, 6 June 2006 (UTC)
 * It is pretty extensively used in de:, but not in the formal way you are suggesting. I mainly use it (and as far as I can tell, most other people do) in the event of coming across an editor for the first time and wanting to find out a little about them. On de: requests for adminship doesn't seem to have the same level of obsession with meeting particular standards or going through detailed tools-derived statistical summaries as the English Wikipedia. There is much more tendency to simply say "I trust the nominator's judgement that this candidate will make a good admin". As for vandal-fighting, I expect tools could be written, but there isn't likely to be demand for it unless an extensive network is set up. There are possibly other uses too, but I suspect further uses are, again, most likely to emerge if the network actually acquires some scale. The most obvious purpose is to derive a trust metric from the network, and there are a myriad of ways to do that. The original idea Jimbo Wales had in mind was for a kind of user rating system, but there are many problems with the idea (allowing negative reviews effectively invites attack puppetry, incivility and factions, for a start) and the idea evolved into a network system that doesn't attempt itself to provide a numerical rating. Many systems exist, however, to produce one, as has been explained a little in the current version of the proposal (which hopefully clarifies things a little better) TheGrappler 23:47, 7 June 2006 (UTC)

This is not a proposed guideline or policy
This isn't suitable for being enacted as either guideline or policy. It is, however, a proposed process – if enough people start using it for it to be useful, it will have become an "enacted process". There is no need for it to be be accepted as policy/guideline by consensus; those who disagree with it would be best to take it WP:MFD. Why does the current proposal exist? Essentially (1) because it is a widely used on de: and (2) because the original idea was strongly pushed by Jimbo, which suggests that it is effectively as "official" as it is likely to get. TheGrappler 08:49, 7 June 2006 (UTC)
 * I repeat - widely used on de: for what? -- Rick Block (talk) 13:19, 7 June 2006 (UTC)
 * I ought to clarify that Jimbo didn't actually specifically support any particular implementation, though he certainly liked the idea of a "web" rather than a "network". As for what it is widely used for, see reply above. TheGrappler 18:47, 7 June 2006 (UTC)
 * Which reply? I only see stuff about the mechanics, which addresses how (and I understand the how), but not the why.  Is it used for RfA, or for RC patrol, or what?  -- Rick Block (talk) 19:55, 7 June 2006 (UTC)
 * The one I haven't written yet :-) Sorry, got bogged down doing something else! TheGrappler 22:21, 7 June 2006 (UTC)

Strong oppose
Wikipedia editors are human and status and social approval is important to many of them. This is likely to be used mainly by the status hungry. A great deal of time will be wasted back-scratching and creating cliques. Power will be concentrated in the hands of long-time users who concentrate on gaming the system. Wikipedia is not a social club or support network. Piccadilly 18:39, 8 June 2006 (UTC)


 * "Wikipedia is not a social club or support network" - I agree. This isn't a social club or support network though. It certainly doesn't allow socializing (contrast this with [[WP:ESP|Esperanza, for instance). It doesn't concentrate any power (you get no new rights). I'm not sure about time spent back-scratching or creating cliques: the reason I'd trust somebody is precisely because of the edits I have seen them make; anybody who was really desperate to be trusted by the people worth getting trusted by would concentrate on making a lot of high quality edits. As for generating cliques, they would also be immediately exposed (a bunch of people only trusted by each other would stand out immediately), which may in fact be useful (if you've ever encountered a pack of nationalist POV-pushers, perhaps the most obviously cliquey type of user, I'm sure you can see why it might be useful to identify the size and extent of that clique). What you need to remember is that it isn't a social network (there's no socializing element) and it isn't a rating system. Just because somebody is trusted by a whole bunch of users doesn't necessarily mean anything - it may just be a sign that they are concentrating on back-scratching and clique-creation. The important thing is they gain nothing at all by doing this least of all power! (Please compare the current version to the archived version: most of your concerns have actually been dealt with in the transition) [[User:TheGrappler|TheGrappler]] 18:49, 15 June 2006 (UTC)

Ditto. Edits and comments should not be juedged by who the editor/commentor is, but by their inherent quality. An anon IP can make edits that are just as worth y as Jimbo's - isn't that the whole point of WP?Juneappal 20:32, 14 June 2006 (UTC)


 * It's definitely true that edits should be judged primarily on their quality. Nevertheless, I'd be surprised if you found that an editor had been working on an article you'd spent time working on previously and didn't go and have a browse through that editor's user page, talk page, and maybe their contributions in an effort to "get to know them" a little. The trust network is just another tool to help with that process. It's not compulsory to use it, it's just an extra tool which some people might find useful. If you discover that User X who you have never come across before but has started showing up in your watchlist is actually a longstanding contributor who is trusted by User Y, a user who you have a lot of respect for editing in your topic areas, you might rest a little easier. If you see that X is trusted by User Z, a notorious bad faith POV-pusher, you might want to check out X's contributions that little bit more thoroughly to see if any unreferenced POV claims have snuck in. You yourself may not be bothered with creating any trust subpages, you may in fact choose not to use them at all, but I hope you can see that it has uses which some users may find valuable and wish to take advantage of? TheGrappler 18:49, 15 June 2006 (UTC)

Why I would find this useful

 * When someone I don't know modifies a date without giving a summary, it would be very useful for me to know that several people whom I respect trust this editor. It would save me the 15-20 minutes of research this often entails. - Jmabel | Talk 22:14, 25 June 2006 (UTC)
 * That sounds in the right ballpark, 15-20 minutes is probably about as long as I take when I'm getting to know who I'm dealing with. Sometimes all you actually want to know is "is this user on the level?" and this is a good way of quickly checking that. You wouldn't want to rely on it in a RfA or RfC when you want to know a user in a little more depth, and if you're really interested in keeping a particular article top notch you might want to check out an edit in more depth yourself but for editors with watchlist inflation, you can't verify every edit to every page. TheGrappler 22:27, 27 June 2006 (UTC)