Talk:Proposed features/no robot
I think this would just promote lazy bot writers, and as such is a bad idea. Bots should be built with many safe guards against breaking people's work. They should just not break things. With this tag bad bots are just going to wreak havoc, and then just ask people to add a no_robots tag to stop them doing it again. It's promoting the wrong emphasis. It also only covers the case where a robot is changing a single tag... there's lots of other fun stuff that bots can do. -- Randomjunk 15:59, 29 September 2008 (UTC)
I generally agree that such a tag might be helpful. I think though, that we should have tags that control renderers, routers, bots and mapping test features marked, e.g. by beginning with a certain character. --Lulu-Ann 09:01, 30 September 2008 (UTC)
- Not a good idea at all. Bots should do things right or not at all. Chriscf 09:14, 30 September 2008 (UTC)
I can see the point of such a tag, but only for specific robots. Similar to the 'noname' proposals: you know that there are exceptions, and you just want them in the database. So I think no_robot should list the names of the robots that are not allowed to modify the object. --Phicoh 13:07, 30 September 2008 (UTC)
I do not like this tag. It either cannot be enforced - then it is useless, because the bad bots will simply do their job without paying attention to any other tag. If it can be enforced, then this tag can be misused to keep incorrect data from being corrected by bots. IMO each bot should, if necessary, implement by itself a mechanism to provide means to disable some or all of its actions on the next bot run, e.g. provide a tag stating which actions or updates should be performed, depending on the type of robot and using a designated tag for just this bot. SvenR 07:39, 1 October 2008 (UTC)
I don't like this feature. If you think you work well, you have to put the no_robots on every node, way,... cause you don't want your work to be chanded / destroyed. This will bloat the database with unnecessary stuff. Like Chriscf I think that bots have to "just work". --Zottel 20:42, 1 October 2008 (UTC)
There is clearly a need to advise automated processes working with the osm data of exceptions in the data. Another example of such a problem is the Noname proposal for streets without names.
I think, the solution of adding special tags to exceptional objects in the database works fine for passive tests like no-name, but not for robots which actually change the data. While tests would only produce false positives for exceptional objects which are not yet tagged with the special tags, robots would change the data from something correct to something incorrect in such a case. Since we do not know every robot that might work on the database in the future yet, it is impossible to say which objects might become exceptional one day and to protect them accordingly with special tags.
While robots are perfect to fix things that are already wrong in the database, I think, in general a better approach would be to ask people for explicit confirmation before adding wrong things to the database. In this case the robot would only be run once to fix all existing wrong entries (which will never be marked as exceptions because when they were created the robot did not exist) and for all later changes the user would be explicitly asked if something is an exception or not. This way it can be ensured that the robot tag is not just forgotten. --Xoff 21:38, 7 October 2008 (UTC)
Bad, bad robots may ignore such a tag, but it's a good idea anyway. It would protect some exceptions from being corrected by otherwise good robots...
Oh. You're the author of Xybot. I see. I'm unconvinced that this tag will actually achieve anything: bad robots will ignore it, and you can't predict easily what tags will be broken be even moderately benign stupid bots. --achadwick 09:12, 31 March 2009 (UTC)
- Is there enough resistance to reject this proposal already at this stage? --Kslotte 22:44, 5 February 2010 (UTC)