Comment from joost schouppe on 4 September 2014 at 00:28
I agree that it would be great if iD would give warnings if you do weird stuff. It might be a good idea to have a system whereby a first edit goes through instant validation by a real person. In fact, it would be great if a live chat were built into iD so new contributors can have live guiding as they go. I’ve been dreaming about a system where experienced users can “adopt” newbies, giving them personal feedback and pointers. And then of course have some statistics by Pascal Neis about how your adoptees are doing as compared to other people’s adoptees.
But the focus should really be on how to turn new contributors into recurrent contributors, not on how to avoid damage by them. I predict the next three reactions will mention how we most definitely -do not- want to be like Wikipedia. The best way to kill a project like this is to heighten the barrier to becoming a contributor.
BTW, when it comes to destroying relations: when fixing administrative boundaries, I noticed experienced users are at least as good in this as beginners are.
Comment from alistair31415 on 4 September 2014 at 01:23
Rather than putting the burden on experienced users to check the first edit from new users, it may be more useful to introduce functionality similar to that provided by git, which would make it much easier to track and correct the mistakes.
Also, a user’s first edit may not be the one that causes trouble, the first edit may be extremely simple, it could be later on as they’re starting to get a bit of confidence that they make more extensive and damaging changes.
Disclaimer: As a beginner I fit right in to this category - knowing just enough to be dangerous.
Comment from cadsauxk on 4 September 2014 at 01:35
I found your comment disheartening and very disappointing. I have made no suggestion that we should become like wikipedia and it is not me who uses “the wikipedia of maps” as an opening line all the time. What I am suggesting is that some level of protection to the map would do us a lot of good and we would stop overburdening our experienced mappers (at least here in Brazil).
I also have made no suggestion of “heighten the barrier to becoming a contributor”. On the contrary. A new mapper needs to have confidence that his/hers initial edits are not breaking anything. I see that as lowering the barrier, very much like learning how to drive on a quiet residential street instead as on a busy highway.
About your comment “BTW, when it comes to destroying relations: when fixing administrative boundaries, I noticed experienced users are at least as good in this as beginners are.” I would agree, except that in Brazil the damage of new users by far outnumbers that of experienced users. Besides, experienced users are talking to each other all the time and those things end up corrected more quickly. New users don’t participate in mailing lists, they don’t even know such exists. We have to find out about them and their edits. It is a painful and laborious manual process.
When I have to tell a new user that we had to revert his edits, the reaction is often of frustration and dismay. I would prefer telling a new user “go ahead, do your mapping, the system is safe, you are in no danger in breaking anything”. I think this would encourage a lot more people to contribute then the present system.
Comment from naoliv on 4 September 2014 at 03:12
É… é triste essa situação…
Comment from Skippern on 4 September 2014 at 03:44
Gerald, I fully intend your concerns, and I am of those experienced users who “waste” quite a lot of time on others mistakes, wether beginners or more experienced contributors. I would love to see more of the beginners entering on the mailing list, and more of the contributors in general entering on our IRC chat, that way we could help more quickly, but answering questions to beginners and fixing problems where they occur.
I have had quite a high threshold against “messing with other’s mistakes” but had to bite the grass and install plugins for reverting, and actually fix it myself. I monitor a quite large area (the entire state of Espírito Santo), and have to validate approximately 10-15 edits daily, with the quick superficial glance on most of them I still spend about 1 hour looking at other people’s probably healthy work, on order to find a few extremely damaging edits.
It would be appreciated if new users could receive an official invite, so that we can get them onto the mailing list and the IRC chat, avoiding any great damages, and hopefully help newcomers becoming regular contributors instead of “fixing one POI and leave”-type users.
BTW: Gerald, you are hereby invited to participate on irc://irc.oftc.net/#osm-br Hope to see you there and hope we get more newcomers there.
Comment from SOSM on 4 September 2014 at 05:33
It should be pointed out, given that neither the rate of new contributors joining nor their number of edits has noticeably changed over at least the last 5 years, the “damage” that they cause both relative to the number of larger and more experienced contributors and to the amount of pre-existing data has actually gone down and not increased.
Comment from Tom Chance on 4 September 2014 at 08:39
SOSM, Gerald suggests there is a particular problem in South America. Maybe that’s the case?
My experience in the UK is that we have no more of a problem today than at any time over the 8 or 9 years I’ve been involved. The only change has been that, as Gerald notes, the increasing complexity of the data makes it more likely that a new user will accidentally break something. But I haven’t found it too burdensome to contact every new user in my parts of south London, checking their first edits.
It sounds like the most simple next step, which would raise few heckles, would be to introduce a system of warnings and prompts in iD. For example:
* on clicking save for the first few times, it could ask for a second confirmation noting that it is the live database, features must really exist, and can’t be from copyrighted sources
* if a user deletes any features in their first ten edits, give them a warning prompt asking if they meant to do that
* if a user changes tags on existing features in the first ten edits, give them a prompt asking if they are sure the change is correct, and noting that they shouldn’t change other people’s contributions if they don’t understand them
All of these could be written to be friendly and encouraging.
One of the burdens on regular contributors is also the lack of tools to track and visualise changesets, so we can see when changes are being made, and understand exactly what has happened. The main OSM web site’s changeset tool still has the bounding box problem, and is helpful when looking at features being added or deleted, but it’s really time consuming to work out where features have been slightly moved, tags have been changed, etc. and to visualise all of this. OWL was very promising, but hasn’t been working for a long time now.
Comment from Vincent de Phily on 4 September 2014 at 09:38
Validation is hard. JOSM has the best in-editor validator, but it is far from being flawless. Tools like osmose have very good algorythms too, but they are after the fact and not worldwide.
It’d be great to have something like a changeset review board that newbies (or anybody having doubts) could upload to instead of uploading to the APIs directly. That board would run a standard set of checks, allow other people to look, have some shortcuts to IRC, and have a “send to DB” button. A man can dream, can’t he ?
Comment from Yorvik Prestigitator on 4 September 2014 at 11:56
Was it a newbie who made the UK underwater (well the land colour blue the same colour as the rivers and the sea) yesterday?
Comment from Pieren on 4 September 2014 at 12:30
You did not catch the spirit of the project : if you start to protect data, it’s not anymore a crowd-sourced spatial database. I follow OSM since 7 years and I red similar complains since the beginning. If the model of contribution would fail, the project wouldn’t be so successful as it is today and the map so complet in many areas.
But you point out a real issue : with iD editor, you can remove an element which belongs to a relation without any warning. This is not the case with JOSM. Instead of complaining, you should open a bug ticket for the iD devs : https://github.com/openstreetmap/iD
Comment from cadsauxk on 4 September 2014 at 13:13
you did not catch the spirit of my comment: we don’t have a CROWD of mappers in Brazil.
Comment from Pieren on 4 September 2014 at 13:19
That’s not the question : everywhere in the world, OSM started with a small community ! And everywhere we have positive contributions and negative contributions. It works until now because we have much more positive contributions than negative ones…
Comment from cadsauxk on 4 September 2014 at 13:38
if you start with a small community and an empty map, that should be largely OK even for a large country. If there is nothing, then you can not damage it.
However, the Brazilian map is not empty any longer, but the community is still very small and the country is huge.
We feel overburdened by problems which could be very easily prevented without limiting the freedom of anybody.
Essentially the influx of new users in Brazil exceeds the capacity of the small local community to cope.
Comment from cadsauxk on 4 September 2014 at 15:15
Here a new user just deleted submarine cables
Comment from ika-chan! UK-USA on 4 September 2014 at 16:26
A quick side question: do we have database backups in case of a serious vandalism incident?
Comment from erjiang on 4 September 2014 at 17:32
I would like to see more UI safeguards in iD to help with some of these issues. A large part of interface design is protecting people from themselves, after all.
E.g. it’s frustrating to see possibly bad edits from a new user that just has “(no comment)” -> stronger message / barrier before uploading, like in josm?
It’s too easy to delete stuff that’s part of a relation without knowing it -> iD modal dialog warning whenever something from certain types of relations is deleted?
Comment from kerosin on 4 September 2014 at 19:02
I don’t think all these mentioned problems can be solved on the UI layer. Of course more warning on editor like iD could improve the situation (of course the difficulty of changing relations already attracted attention, see https://github.com/openstreetmap/iD/issues/1461).
But some problems are caused by the design of OpenStreetMap: everyone can register and change all data he/she wants.
Waze is solving this with ranks of users, that are only allowed to edit some things in “their area”. Google is solving this with checking every edit by their “experts”. Another thing that could be done is something like mentors for every new user, that is watching their edits.
Of course the OSM way has a lot of advantages…But it also requires a lot of QA and some “bad” edits will never get notified.
Comment from MarkusHD on 4 September 2014 at 19:07
Exactly, what’s the point in allowing empty comments at all?
Comment from SDavies on 5 September 2014 at 07:34
I totally agree with Tom’s last paragraph, above. I think the biggest issue is the lack of tools for easily visualising what edits have actually been done. This, along with a few well-thought out extra warnings in iD etc, would make the whole problem that much more manageable (and therefore less of a problem).
Comment from Jean-Marc Liotier on 5 September 2014 at 09:02
In Senegal where the mapper density is extremely low, I often stumble on the works of lonely enthusiastic novice mappers who sometimes make a hash of things with simply bad modeling - contacting them is usually the most productive course of action and they soon mend their ways… Their mistakes are part of a learning process that makes them better mappers - a system where mistakes are not possible is a system where there is no learning.
The other category of errors are ‘fat fingers’ from random iD users who inadvertently move a node some distance away from its original position - always spectacular when that node is at a crossroads and moved a kilometer from where it should be. That one might benefit from some tooling : just like JOSM, iD could validate a number of things - for instance a ‘crossing ways’ warning would catch most of those inadvertently moved nodes. Of course, such validation would have to be light enough for the web environment and pedagogical enough to guide the unsuspecting user - but focused on common errors it could be useful.
Comment from SomeoneElse on 5 September 2014 at 15:10
@Yorvik Prestigitator yes, it was a new user the last twice it happened, but on the most recent occasion it was found and fixed within ~6 hours so I don’t think that was more than a temporary annoyance.
Comment from cadsauxk on 5 September 2014 at 15:27
@SomeoneElse: that is the point, it was in the UK. Compare the amount of mappers in the UK and in Brazil. I would guess that in Brazil we have 1% of the number of mappers that UK has, but perhaps even that number is optimistic. Now compare the size of the two countries. Here in Brazil it is not unusual to find this type of problems only months after. In one extreme case, a illegal import was found 5 years late, and it was a huge import.
Comment from keithonearth on 5 September 2014 at 20:37
As a new editor, I read your post with a great deal of interest. I can see that new editors have the potential to cause damage, but am not clear on what you propose as a solution, other than the rather vague idea that “OSM shifts it priority from gathering tons of new users to preserving the integrity of the map”.
I’m not sure what it is that OSM now does to gather tons of new users. Nor am I clear on what could be done to maintain the integrity of the map. You say both that the map needs protection but that we shouldn’t “heighten the barrier to becoming a contributor”. I’m not sure how these work together.
However, what is more important for me to know as a new editor is what can I do to avoid doing damage?
Comment from cadsauxk on 5 September 2014 at 21:33
Obviously, everybody as the potential do cause lot of damage. However, the pattern that is overwhelming the few active users here in Brazil are those new users, who have just signed up, and start messing up the map in some very amazing ways.
You are correct, I am not proposing any particular solution, I am suggesting that the integrity of the map should become a priority, a principle that should be embedded into the design of of the OSM system. I don’t really know how to rely on a map for my everyday use knowing that key elements could be wiped away or messed up in a matter of minutes.
Some colleagues suggested modifying iD for that. I agree. However iD is just the editor of the moment. Previously we had potlatch with identical problems. As smartphones and tables get more popular other editors such as vespucci will be used and so on. It has to become a guiding principle.
On your comment: You say both that the map needs protection but that we shouldn’t “heighten the barrier to becoming a contributor”. I’m not sure how these work together.
Let me ask you this: if you had the option, would you chose learning to drive in a car without seat belts, air bags or without instructor controls? When I started editing I was very afraid to break things. In fact, I felt so insecure that I almost gave up. If I knew there were safeguards, either at system level or at user interface level, I would have started with a lot more confidence.
Protecting the map is not prohibition. Suppose you have some family pictures on you computer that are very important to you, you could write-protect them to avoid accidental deletions. I honestly don’t see why a similar scheme would not work on OSM. It would give everybody a lot more confidence about the map itself and would be far less workload than constantly finding and undoing random damages.
And your next questions says it all: However, what is more important for me to know as a new editor is what can I do to avoid doing damage? you shouldn’t have to worry, but in the way the system works now you have to. Currently, the best thing you can do is to never delete anything without knowing what you are doing and start to take part in a email list or user forums.
Comment from Skippern on 6 September 2014 at 00:36
@keithonearth, @Gerald & others
There are several features that can help us in this task to prevent new users making damages.
Comparing UK/Germany/Netherlands with Brazil will not help much in this. Brazil have an extremely small active community, with an area to cover similar to the entire of Europe.
There are several points we can look at:
1) Editors should enforce users to write comments
2) Expereinced users could be nominated “area monitorers”, an area monitorer should receive a message when a new user edits in that area.
3) The ability to “write-protect” special objects, and limit the number of users who can change them, this can be useful to avoid breakage of national boundary relations, and can be used to prevent edit-wars
4) A generic welcome message presenting new users to the communities where they does their first edits, these messages should be prepared for the different communities, presenting mailing lists, chat-lines, forums, etc.
There might be a few other things that can be done. Some easier than other, some have more effect than other, and some might require quite some work to implement. The editors are up to developers and maintainers of each individual editor, though I believe the other suggestions are possible to look into on the main API, or interfaces linked with the existing API.
Comment from Aury88 on 6 September 2014 at 07:40
why not send a warning to some “local mentors” when particular types of elements are modified/deleted by an user with low number of edits?
I think for example that relation is the more difficult element to understand by a novice (I made big error twice when I was a 100 edit-mapper ) so if a new user deletes or changes a relation a warning is sent to him and if he saves a warning is sent to the “local mentors” with the critical changed element list (and possibly an easy way to visualize the change and revert element per element).
This list of element can also comprehend boundaries, coastline etc etc
I agree that one of the critical points of the project is the lack of tools to track and easily visualise changesets and the integration of QA tool in online editors.
also I totally agree with Skippern suggestions
Comment from jutezak on 6 September 2014 at 08:20
The freedom OSM allows in tagging and organizing might be the problem here. NOT the freedom to start mapping.
Slowing things down, extra popups, forcing comments: they try to make mappers more careful. But the mapper still thinks he’s doing the right thing!
I think that editing at a higher level of abstraction is the solution. Not nodes, ways, tags, and relations for the average user. But a crossing. Street. Building outline. Country border. River. Each can be given an example to clarify (Highway, for example the A1 motorway).
That would make for an easier start and prevent mistakes. And get more consistency as a bonus.
Comment from keithonearth on 6 September 2014 at 18:05
Thanks for your post @skippern those all look like real concrete suggestions that would be beneficial. The only alteration I’d make, and I have no idea if this would be feasible, would be to to supplement “4) A generic welcome message” with a non generic message sent by a human. I’ve been editing relatively actively since may of this year, and have not gotten any positive or negative feedback. I’m pretty sure I’m doing things right, but I’d be surprised if I’d not made any mistakes on the way. Your second point regarding area monitors would work well with sending a friendly greeting, I think a community project like this does benefit from social niceties that remind people are in a community of people.
Comment from Skippern on 6 September 2014 at 21:40
@keithonearth As an area mentor would be notified of new users editing in his area, it would be up to the mentor if he would do a “human greeting”, I would surely give a quick greeting to any new contributors in an area I monitor, presenting myself and what I do in terms of monitoring.
A generic message would be sent at the moment first change set is closed, a human greeting might be sent somewhat later as it requires human interaction.
Giving the new user an idea about where to get in contact with the rest of the community at an early stage I think is a good way for him to be incorporated in the community. I would like to see the 10-20 active users in Brazil grow to 10-20 active for each state, and maybe also the same number for the greater metropolitan areas. If we can manage that during the rest of this year and the next, than I will feel the burden much more shared.
Comment from mcld on 7 September 2014 at 11:01
The ability to “write protect special objects” is an interesting idea - it’s a lot like the mediawiki idea of “protected pages” (which can’t be edited by newcomers). I like the idea of that, as long as it’s restricted to a small number of really special/breakable objects such as coastlines. I’d suggest that the protected objects should be editable only if your user account is more than X months old AND has made more than Y edits. It seems to me that, as with wikipedia, as long as we remember not to apply this lock too widely (which would inhibit contribution), it could help.
Comment from RicoElectrico on 8 September 2014 at 11:52
There should be more power delegated to local communities, it’s annoying and counter-productive that you have to write to the DWG every time to get people messing with the map blocked (Or as I heard there’s “block until message is read” ). People not responding to messages are quite a problem.
There was a HUGE influx of mappers in Poland that was caused by press coverage. Some people added useful data like shops, some data didn’t quite make sense (like an address node with only a name, or area=yes).
But also, newbies deleted things like fire hydrants which had to be reverted.
We have a page listing new users: http://openstreetmap.pl/users/ . At the time of that influx, I tried to check people’s changesets using OSMHV, fix them and message them. With current tools, there’s simply not enough human throughput to review changes.
We also use http://osmapa.pl/w/zmiany/diff.php , but it’s not always up to date (lags a hour or two). It helps spot obvious deletions, though.
Comment from RicoElectrico on 8 September 2014 at 12:13
I blame the iD editor on most of these lousy edits. Its interface makes one seem they are guided, but they aren’t really. For example, people notoriously add everything related to medical care as a hospital. Also, no validation whatsoever, nor info about Bing imagery callibration. There’s better imagery with 100% coverage in Poland (Geoportal), which has no clouds and is perfectly callibrated. (It loads slower, though).
The turorial seems really dumb. People will know how to draw areas, duh. I think that a good idea would be to make goal-oriented minitutorials like “I want to… Draw buildings/ Add one-way to part of a street / Add an address”, which will explain how to draw buildings properly, or how to cut streets to apply different attributes, or that when you add an address in a village without street names you have to use addr:place.
The problem is that presets simply don’t convey all the info about how to map them, which people should know - every country has their own specifics. People should be warned not to tag objects as something other just because they couldn’t find a fitting tag.
Comment from neuhausr on 9 September 2014 at 13:35
Way back up to [Tom Chance’s comment] (http://www.openstreetmap.org/user/Gerald%20Weber/diary/23680#comment27779) about visualizing changesets and finding new users, you can set up a feed for an area of new users and/or changesets via [Pascal Neis’ site] (http://resultmaps.neis-one.org/) (and the changesets are the truly local ones, it doesn’t return global changesets like the osm.org page does)
Comment from Warin61 on 31 October 2014 at 23:56
Not so long ago, as a new contributor, I quickly learnt .. don’t delete stuff.
Even when I’d been past something for years .. and ‘knew’ it was not there .. I’d go and actually look and there it was.
So my principle is to add, maybe move, modify but try not to delete. I think the is a very good guide for everyone.
Comment from tmcw on 2 December 2014 at 05:25
If anyone has a particular validation in mind they want to see in iD, I’d recommend checking out the existing validations and adding one: it’s an open source project and we’re happy to help integrate improvements from the community.