On issues with tag voting

General talk
1

Continuing the discussion from (Informal) Proposal: Recommend using `board:title=` over `name=` to tag the title of an `information=board`?:

I pretty much almost fully agree with everything you’ve written above, so I’ll just add a general (thus somewhat offtopic) comment on this:

Yeah, we don’t have a good way to deal with multiple choices, at least not formalized one. Proposal process deals basically with just boolean yes/no of single proposed change which would be either fully accepted on fully rejected.

So best we can do currently is make a Discourse poll (which does allow multiple options) about which solution most people would prefer, and then make a formal proposal for that one. I agree that Condorcet or something might be more useful and streamlined for such cases regarding choosing actually preferred choice, but it would also make it significantly more complex (and not just in drafting phase!), e.g.:

  • in wiki voting in Proposal Process as it is, one also leaves a reason why are they opposed to the proposal (to help iterate and improve it in its next revision) which is relatively simple (yet, even that is too much for many)
  • with multichoice voting however, they’d have to explain their whole system of preference of each of the options and what they find as a lacking in each of them, and/or why they prefer other solution when neither is perfect, etc.
    I don’t think vast majority would bother, and even if they did, it would be a hell to try to interpret and improve proposal.

TL;DR: probably best to stick with Discourse multichoice vote to pick most popular option, and then detail that most popular choice in formal Proposal process, so it can be iteratively improved in Draft phase (both original phase, and those phases it would fall back to Draft due to voting rejecting the existing proposal).

And more generally, there is other issue which might explain “status quo” preference when changing existing tags: not only is changing existing tags dangerous if not done properly[1], but the definition of “majority” is ambiguous – in wiki voting, usually less then few dozen people vote, while the tag may easily be actually used by several orders of magnitude more people.

  • in some voting systems, there is a notion of minimum quorum; so if in some situation only e.g. 1% of voting members would show for a parliament vote, the poll would not happen / would be deemed invalid if it did happen.
  • some other systems require minimum support before be able to considered a valid voting options (e.g. in some countries, before running for president, you must have 10k or 1% or whatever people indicating written support for you before you’d even be considered as a candidate)
  • Alas, in OSM we regularly have significantly less than 1% of interested parties participating in the votes, so neither of those would help (and would in fact prevent any progress/change if followed).
    So, erring on the side of avoiding potential damage to existing data consumers (i.e. keeping “status quo”, instead of going with modifying existing tagging situation under unsuspecting consumers legs because 6 people liked the idea) seems somewhat prudent, given the circumstances. While frustrating, such insistence of Darwinist “survival of the fittest[2]” to many seems better than the alternatives.

  1. and relatively rarely do we see enough effort invested to do it properly ↩︎

  2. well, most persevering, at least ↩︎

1 Like
2

I don’t suppose there is/could be a a notification sent to everyone who’s used a tag more than x times, when a proposal or vote about that tag is about to happen?

3

There isn’t, but technically it could be done - one would need to parse full history dump filtered by that tag (or tag combination, in case of information=board + name=*[1]), and increment count for each user ID.

I was actually toying with the idea of writing similar tool (my use case for more oriented to finding when some tag was first used and by whom and where, as well as how many different users have used it over time, but it could be easily adapted to support this use case too).[2]

As for the auto-notification, I can certainly see the attraction… But if it were to be implemented, at the very least, there would need to be a way for people to easily opt-out of such notifications.

But, while I believe it would help increase the participation somewhat, I wouldn’t hope for huge surge: my guess would be that vast majority of people who map in OSM are simply not interested in low-level politics of specific tagging (in fact, many might not even know what tagging is used exactly, e.g. if they use StreetComplete, or if they mostly use presets in iD, EveryDoor, Vespucci, JOSM…)

At least the first one, yes. Once you know about the proposal itself, it is easy to subscribe to discourse topic and proposal wiki page to get notified about vote (and other related things like changes to proposal etc).

And those interested in following proposals generally, might subscribe (or use RSS to follow) to RecentChanges/tagfilter=proposal-status-changed

But getting that first notification is currently only happening to people who manually starred a certain wiki tag entry as interesting, so when proposal comments in its Talk page, they will be notified. But people usually subscribe only to few tags they have special interest in – not all the tags they actually use. So while that notification mechanism it might work somewhat OK-ish for niche tags like board:title=*, it would fail for more popular ones like name=*[3]

  1. which this discussion was split from ↩︎

  2. But as always, not enough time for all ideas. :sweat_smile: Maybe someone even has already implemented something similar? ↩︎

  3. where it would be even more important! ↩︎

2 Likes
4

Note that you need to account for reverts, way splitting etc. Or be fine with false positives.

( matkoniecz/who-added-this-tag - Codeberg.org is tool that can do this, but due to how it works it is suitable for tags of limited use, say up to around 20 000 objects.

It does NOT parse full history dump, that is why it has such limitations. )

2 Likes
5

I’d be amazed if you even got a handful! :cry:

We’ve just smashed our open Notes in Australia, with only ~400 still open. But to take them much further, we need replies on the ~350 that we’ve left questions on. Response so far? … :hear_no_evil_monkey:

For people who are more than likely already on here / Discord etc participating in current discussions, not the vast majority wh apparently don’t ever talk to anybody else at any time!

2 Likes
6

I’m probably going to contradict myself a couple of times in this posting, but that’s just due to the subject matter.

Voting on tagging currently doesn’t serve any reasonable purpose.

Often it is considered a popularity contest that has to be “won” which why many times the votes are brigaded, but for new tagging “winning” can’t be the point, the thing that needs to be done is a sanity check:

  • does this conflict with existing tagging / is this really new
  • does the proposal work from a functional pov and adhere to best practices

Voting doesn’t really do any of that, matter of fact because you can “win” that vote, it is counterproductive and the past shows that this leads to badly thought out proposals “winning” even though they are clearly broken and we have had completely reasonable tagging schemes “fail” because the proposer simply wasn’t connected enough.

The actual popularity test would be if a tagging scheme gets traction and is used.

But in modern OSM this interacts with the default presets in the most used editors and taginfo, to the point where it has become really impossible to determine what is driving use, if it is real interest or gaming of the system (and yes then there is StreetComplete and MapComplete that just do what ever they like, but this feeds in to the system too).

In any case as said, stop the voting, it just doesn’t do anything helpful, provide a checklist of best practices that a person designing new tagging can follow and leave it at that.

The other case is changing existing tagging and “deprecation”, this has a real impact both on contributors to and consumers of OSM data, and there the current voting rules are an unmitigated disaster, with minimal brigading you can change tagging that tens if not 100s of 1000s of people are using. Again, voting doesn’t do anything here except make things worse, giving fake legitimation to destabilise tagging.

Just stop it.

3 Likes
7

Did you read at least TL;DR if not the whole of previous discussion? i.e. specifically:

Given that some people prefer using on information=board to specify its title using tag name, and others board:title, and others title, and yet others inscription:title (I’m not making it up, I swear), etc., would you prefer:

  • that they discuss and agree to define one of them (whichever seems to them to be the best to use for that function) as preferred and standardize on using that, or
  • that they randomly use any and all of those (and dozen others that might sprung up) tags for that same purpose without any preference, or
  • that they should stop micromapping information boards, or
  • something else (what?)

(yes, I get the irony that this is itself kind of a vote, even if directed on only one voter)

If you read previous messages more carefully (like at least that TL;DR), you’d see the purpose was not any kind of “winning”, but instead choosing optimal way to proceed and concentrate efforts on polishing the tag that is most likely to get accepted.

Would you still oppose such course of action?

Nowadays I link to Deprecating is hard - OpenStreetMap Wiki and ask proponent to detail how they would solve each of the bulletpoints there before proceeding.

We have voting rules for that?! Where?

Wait, there is a “system” in OSM, which can be “gamed”[1] (presumably to gain some “advantages”[2])? Where? And what advantages? What you describe I’d call just “regular mapping”.

If I choose to use MapComplete to answer a question to add a title to an information board, is that “gaming the system” because MapComplete chose which tag to use to record that information in? Or is it just me choosing what tool I want to use and what data I want to record?

And if it is “gaming the system”; what if I used Vespucci preset to record description of that information board, would that also be “gaming the system”?

And what if I enter a raw tag in Level0 editor (from my head, no presets involved), is it still “gaming the system”, or is it finally me “playing fairly”?

Stop what? Discussing in Discourse in general, or only on certain subjects?
(Beware: I might agree! :smiling_face_with_sunglasses:)

And would you care to give your reasoning why? Has discontinuing discussion between negotiating parties historically ever led the humanity in better direction?

  1. and what is a difference between “real interest” and “gaming of the system”? Isn’t the “interest” of “gaming the system” (whatever that means) as “real” as any other interest? And if not, what are acceptable interests? Making money out of OSM? Feeding own sense of self-worth and accomplishment? Helping navigating AI suicides drones toward their target? Exclusively pure altruism? ↩︎

  2. point of “gaming something”, as I understand, is to gain unfair advantage in “the game”. As opposed to just “gaming” which means, AFAIUI, just “playing the game”, without implied unfairness of gaining advantage[3] as you progress in the game. ↩︎

  3. I assume one would gain advantages, unless they totally suck at playing that game and never progress in it ↩︎

1 Like
8

I’d assume the most reasonable way to implement this would be by creating and then maintaining per user stats about tag usage, which could then be used for this kind of thing.

There could be full block list (or “follow even if I don’t use this tag”) controls for notifications in user preferences. Every notification would contain a link to the preferences so the user can easily opt-out if this isn’t something they’re interested in.

To be honest, I don’t expect this would lead to significantly larger voting participation. But at least then it’d be people actively choosing not to participate, instead of simply not being aware of a vote/proposal.

I realize implementing this would be a lot more involved, but I feel the whole process would benefit from a bit more structure and visible infrastructure.

1 Like
9

It’s actually far fewer than that. As an example, board:title is used 24k times; the last linked wiki vote I can find got exactly 22.

In democratic political systems we have rules about who’s allowed to vote, but generally we believe that everyone (subject to an age and some sort of nationality qualification) has an equal right to choose who represents them. We don’t require that people take a test (e.g. basic maths, to tell that what some political party is suggesting simply doesn’t add up, or basic history, to tell that what some other political party is suggesting has no basis in fact). We let everyone decide, and whoever the majority chooses wins - and everyone funds their schemes for the next however many years since everyone pays tax.

OSM isn’t really like that. There’s no “one government”, only a rough consensus. Each data consumer is free to do whatever they want with OSM data. Everyone entering data is also free to use whatever tags they like, but the community as a whole will tidy up a mess if one results.

People designing OSM editors think very carefully about how what they put in front of users determines how those users tag things. They have far more knowledge and experience of that then the average wiki voter, and should not be subject to the results of one. I might complain about (say) how iD presents route relations to users, but I can understand how and why the decision was made to have it do what it does.

Sometimes there genuinely isn’t a right answer and each of two options have some merit (I’d suggest that board:title vs name is one of those). Whilst it’s useful to have a discussion about why people think what they think, having a “vote” to choose between two somewhat equally valid opinions will achieve nothing - people will have the same opinions after the vote, and will likely use the same disparate tagging afterwards too.

3 Likes
10

Even if a vote can’t actually change anyone’s individual opinion, I would hope the community as a whole understands the benefits of a cohesive tagging scheme well enough to set aside their personal opinions once a collective decision has been made about which way to go (whether that be through a vote, or some other process).

Having two (or three or five or ten) different ways of tagging the same thing isn’t good for the health of the project. While it’s true it’s not really practical to force a large contingent of people to change how they tag things if they decide to be really stubborn about it, I think we should try to foster a culture that respects the process enough to avoid such pettiness.

4 Likes
11

And what’s the alternative here? If a tag or set of tags need to be deprecated/ replaced, voting is an essential step to legitimise the new tags and allow for a mass edit to move over to the new scheme. Without a vote, people would have nothing to point to as permission to make these changes, and you would have complaints about tags suddenly being changed without a consensus. If you don’t have to put your proposal to a vote, there is no incentive to make it as good as possible or to cooperate with others to get their opinions. Voting is about promoting quality and consensus, the alternative is anarchy.

4 Likes
12

Do you have any evidence or examples to support these claims? You seem very convinced that there are inherent issues with the voting system on the wiki, but I have never heard such complaints from other members of the community. What changes do you think should be made to the voting rules to make votes fairer?

13

note that you are making a certain assumption here that such actions are (1) needed (2) useful (3) must be somehow done.

It is not necessarily shared by others.

(My own view probably is that we would benefit a bit from more well done, discussed bot edits approved by community. And would clearly benefit from having less tag replacements. Except cases where we already have two tags meaning exactly the same thing then getting rid of one is useful. And creating tag duplicates is a terrible idea and should be avoided as it predictably creates repeated problems for nearly no gain. So most tag migrations are terrible ideas. )

(and for establishing consensus that tag X can be replaced by its duplicate Y wiki vote is not needed, see sidewalk=none | Tags | OpenStreetMap Taginfo and Sidewalk=none to no - proposed automated edit )

I think that at least first part is not really accurate (I have no idea what MapComplete tags, I have not get around to playing with it properly so I cannot say much about it). But I am obviously I am very very very biased here. In the “we investigated ourselves and found no evidence of any wrongdoing” fashion. But I definitely tried hard to avoid exactly this mistake.

2 Likes
14

“fairness” is a category error in this context.

The last time we discussed this on this platform I suggested at least an order of magnitude more required votes and a mandatory transition plan with terms appropriate for the change (obviously not all changes have the same impact).

1 Like
15

(offtopic: Discourse really needs a reaction that says “I really agree with 95% of your post, but this one thing I disagree with, so I can’t really add a :+1: without compromising myself”. No idea how it should look like though - maybe like a :low_battery: but yellow and charged to ~65%?)

In any case, I would disagree with this part. At least for myself, popular vote would affect my opinion in majority of cases (i.e. in almost all cases, except those that I have an very strong opinion about[1])

For example, if a vote about name, board:title, title, inscription:title on information=board to indicate its title resulted in vast majority preferring using one of those (even if it is not the one I personally would’ve liked the best), I’d still be happy with proponent making a Proposal Process Draft based on that key (instead of a tag I liked more).

And my vote in that proposal process would’ve been only a little influenced by my own preference of a specific name proposed, and much more influenced about how well the proposal process was researched and argued as a whole.

“A rose by any other name would smell as sweet”, eh?

That being said, I do acknowledge that there are some[2] members of community who’d be much more hardheaded[3] then me[4] on the subject, and would not care one iota about popular community opinion… I just don’t believe that they are the only ones[5].

  1. e.g. if the vote vas about “let’s take money Facebook is offering us in exchange of giving them have full control over OSM data”, there is no amount of voting that would convince me that is a good idea. But then, I guess pretty much no amount of arguments would be likely to convince me either. ↩︎

  2. maybe even many, especially among vocal parts of the community? ↩︎

  3. e.g. insistent of never changing their initial opinion – no matter what; or somewhere else closer to that extreme of the spectrum then to the mellow extreme ↩︎

  4. and I’ll also acknowledge that I am (also by far) not a most softheaded person either :grinning_cat_with_smiling_eyes: ↩︎

  5. and would in fact prefer to believe that many more are more mellow and agreeable then that ↩︎

1 Like
16

Yes. Although, I don’t think order of magnitude more would work all that well.

Perhaps better, in such cases of modifying existing tags (i.e. proposals with high breakage potential), make each “no” vote progressively harder to counteract.

For example, single no vote might be overcome by just 2 yes votes to bring the needle back to zero; but 5 no votes would require 15 yes votes to neuter them. And 15 no votes would require 100 yes votes to neuter them. (or some other numbers if they might work better, this was just a quick example of the idea)

This I fully agree with. Would you care to make a proposal about chaning proposal process in that way?

17

As I mentioned earlier though:

How are you going to ask all these Users what they want / think?

The figure that is often thrown around (& I know that it’s disputed!) is that there are 10 million people on OSM. The latest new User (as of 23:05:27UTC) is # 24397346, so 24.3 million.

I can’t find any stats for users here on the main, official, OSM Forum, but looking at Badges that have been awarded: OpenStreetMap Community Forum, it would appear that there are 39.5k Basic users, 1.6k “Members”, 50 “Regulars” & 3 “Leaders”, which would suggest that only ~41k of that “10 million” have ever actually signed up to the Forum?

Going down further, 10.2k have received an Anniversary badge, for having “been a member for a year with at least one post in that year”, dropping to 1.3k “Enthusiasts” “for visiting 10 consecutive days”, & 224 “Aficionado” “for visiting 100 consecutive days”.

Jumping across & looking at Discord, it shows 7054 Mappers, but quite a few of them are also on here!

Sure, there’s Telegram, Mastodon & a host of other channels (which doesn’t actually help matters!), but It would probably be fair to say that out of those “10 million” Users, only ~5000 ever actually participate in any form of discussion with any other mapper!

1 Like
18

That number includes, I think: one-use accounts, no-use[1] accounts, bot accounts, stopped accounts[2], ban avoidance accounts, lost access accounts, etc. etc. Leaving an unknown, but possibly considerably smaller, number of accounts representing actual active people.

I believe you are a bit too optimistic here. “Lower” badges like Basic are not withdrawn when getting a “higher” like Members, so ‘just’ about 39.5k and not ~41k.

  1. people who created an account but then decided that it was not their ‘cup of tea’ ↩︎

  2. people who have edited but lost interest or sadly, are deceased or no longer physically or mentally able to edit ↩︎

19

You are being a bit too pessimistic IMHO.

There are 2.5 million accounts that have contributed to the data, but only ~50’000 that didn’t immediately stop, and around 25’000 that have edited more than a trivial amount (from a volume pov). Add a couple of 1’000 data consumers / service / app providers and I would suggest that you have perhaps 30’000 people that have enough interest in OSM to turn up on one of the platforms.

I’m not suggesting that these 30’000 are the same as your 41k though, but the order of magnitude works.

PS: naturally there are numerous funnels people will signup to OSM through but probably remain at arms length to OSM itself. For example as not reported anywhere: deflock Simon Poole: "@aaron@social.tromdienste.de found the #OpenStree…" - OSM Town | Mapstodon for OpenStreetMap , but naturally maps.me/organicmaps/comaps is similar in that respect.

2 Likes
20

Looking at a local ChangesetMD database, I reckon that ~3000 unique user names** have edited the map today, 74k in the last month and 320k in the last year. That’s not 24 million, but it’s a lot more than 22 (the last wiki vote response I found above).

** select count(distinct user_name) FROM osm_changeset WHERE created_at > '2026-08-08';

1 Like