MapRoulette Critique

Note: some proposed edits in that challenge have quick edit disabled.

Quick note: This is not a feature of maproulette itself but something I “hacked together” myself, which is I think a feature maproulette lacks.

Does it? I only see 2 of the tags that will be removed and none of the ones that will be added. If you’re not locking out the OK button until the user has scrolled to the bottom then I think that’s not good enough to count as “user has reviewed the changes”.

1 Like

Okay, let me chime in with my two cents on the topic as a whole.

No, the challenge I created and other TagFix challenges are not “effectively a bot edit”. If they are, everyone who accepts iD Editor suggestions, and even everyone who uses StreetComplete is performing bot edits by that definition.

Theese kinds of tools are incredibly useful to quickly fix broad mistakes or enter data that is effectively already there.

The idea behind maproulette tagfix challenges is to especially not perform a bot edit but to check every proposed solution before committing.

If someone misuses this tool by just blindly accepting every proposed edit, they are misbehaving (and should be e.g. banned), not the creator of the tool or the creator of the challenge.

Yes, an editor or a “sub-thing” of an editor (in this case: A challenge) can have systematic problems that need to be adressed on a per-editor level, not on a per-user level. And here I also would wish for a more coherent process to report challenges. More than once I have seen a challenge and thought to myself “This challenge as a whole is just wrong! How do I stop this IMMEDIATELY!?” I would like to have challenges that have been reported at least a number of times to instantly be placed “on hold” or something, and for the challenge owner to be forced to address the concerns before the challenge goes live again.

I also see a lack of possibility for contacting the challenge owner - heck, even just knowing who they are! Why do challenge owners need to be anonymous? Why is the only way to contact a challenge owner via a public comment? There were several instances where I would just like to send a private message to a challenge owner, e.g. to ask for the source code of the challenge.

Also, I can understand that some challenges may be broken because as someone who uses MapRoulette at work I can tell you… it is really frustrating to be a creator at MR. Debugging challenges is near impossible, errors are cryptic or non-existent…

2 Likes

I am not sure this is @Mateusz_Konieczny 's argument here. Yeah, I agree that one cannot see everything that will change. But the argument proposed sounded to me like he was saying “It’s not possible to see what’s going to happen at all”.
It is certainly possible to see that some edit will be made. And a good user should scroll down and review the tags before commiting.

I like that idea of only revealing the OK button after all tags have been seen. :thinking:

I could add something like this if you think it could help sharpen the user’s attention. I am after all extremely pro not-blindly-editing.

1 Like

So, did @InsertUser describe your criticism correctly? Would your critique be solved if I expaned the tags section?

What StreetComplete quest does blind tagfix edits? All of the ones I’ve seen ask clearly worded questions with clearly presented options and have an additional option for leaving a note if the question doesn’t actually make sense in context. It also expects you to be near the object where you can see the area with your own eyes rather than just showing you a render of a map without so much as satellite imagery. If you answer a question that is even a little far away e.g. because you are in quite an open area and there is only one candidate in view it then pops up to ask you to confirm that you’ve actually checked it and not clicked on the wrong thing.

iD has been accused of performing stealth mechanical edits on many occasions because of their presentation of things as ‘wrong’ rather than suspect to naive users. Mapcomplete is not being singled out in this respect.

If someone is being asked to confirm a change that they can’t even see then it is not functioning as an interface for review, it is a machine for clicking yes. Or converting an automatically compiled edit set into a real edit without real thought.

6 Likes

I edited the challenge so that all tags are always visible. (hopefully)

1 Like

Yes, the question of ‘what is too fast’ is pointless to ask. It’s like asking ‘how should I map disjointed traces of old railroad’, something that will never reach a concensus. Also completely dependent on the specific challenge one is working on using maproulette. I have spent more than an hour on tasks using this application to add entire missing secondary roads in places like Pakistan, other times I have contributed to challenges where it only took 30 seconds to fix mistakenly ‘dragged’ geometry.

yes, but you need to scroll or press extra button to actually see it

note “default view”, just before quote. You need extra interaction to see them

large part of it

another part is that there are few account that in good faith try to improve things, open MR and blindly press button to apply changes.

Similarly, few people blindly pressed “upgrade tags” in iD or added bunch of fake fords, but MR magnifies it as you can easily and remotely go through hundreds or thousands of tasks.

BTW, if someone makes such bad edits in StreetComplete, guessing remotely and adding bad data and ignores changeset comments… Then please write to me or escalate it straight to DWG (ignoring changeset comments already qualifies)

thanks! Not entirely sure how it works, would be nice if MR would do it be default for all tag fix things

there is no review, challenges are applied as soon as creator marks them as active

they can be also edited at any time by their creator

A challenge that realistically requires a field survey every time would obviously be a poor use of MapRoulette. However, if only some of the tasks unexpectedly end up requiring that level of investigation, then that’s what the prominent “Too Hard” button is for. MapRoulette tracks the average time spent on each task; a high duration communicates that it isn’t going to give you instant gratification, so it naturally dissuades people from contributing without the necessary level of commitment.

As part of an import I coauthored, I’ve created some challenges in the past that require detailed armchair analysis. The key to doing it right is setting expectations in the challenge description and in the custom per-task instructions. I wouldn’t have dreamed of using the quick fix feature for my import, because that could’ve led an overzealous mapper to run rampant over a part of the map that I take responsibility for cleaning up.

There’s also a mechanism for others to manually review and validate contributions to a challenge and educate contributors if necessary. Unfortunately, lots of MapRoulette tasks never get validated. Maybe as a community we need to do more to promote validation of MapRoulette challenges?

Like OSM itself, much of this currently depends on the honor code. I think the challenges that stoke controversy are generally broader-scale ones, where the project owner has effectively externalized any risks associated with their challenge to the community. How can we make sure the project owner has a stake in the map they’re helping to shape?

Do you mean MapRoulette? You’d have to use a third-party site, but it just uses your OSM account through OAuth, no separate signup required.

There’s also the possibility of making a challenge public without making it discoverable through the site’s search tools. This is the default behavior for older challenges, under the assumption that the source data has grown stale, but maybe there could be other criteria for increasing or decreasing a challenge’s visibility.

Er no, clearly not. The idea of StreetComplete is that the user is literally sat in front of the thing that they are mapping. They don’t know the final OSM tags used but they have answered specific enough questions about the object.

That sounds like the “guns don’t kill people” argument again…

Someone who is new to OSM and is doing a MR challenge described as “easy” should not be “banned” because the MR creator didn’t describe the challenge properly and the platform didn’t vet it.

2 Likes

depends on whether tool is making clear that it is not OK and is it encouraging given behaviour

My comparison comes from the fact that also StreetComplete users can simply disregard the rule that a on-site survey should be done. Just as a Maproulette-User can disregard the rule that they should check every proposed edit for plausibilty and not simply click yes.

My counter arguments as a whole are directed at the pretense that simply because a tool makes editing very easy and could potentially be misused, all edits made by it are automatically bot edits. If “editing is really easy” + “rules can simply be ignored” = “bot edits”, then yes, StreetComplete would fall under this category. Which it clearly does not, which is the whole point I am trying to make.

That sounds like the “guns don’t kill people” argument again…
Someone who is new to OSM and is doing a MR challenge described as “easy” should not be “banned” because the MR creator didn’t describe the challenge properly and the platform didn’t vet it.

In some way, it is. There is a shared responsibility between the creator of the challenge and the user who solves the tasks. Yes, a challenge can be insuffiently designed so that no matter the intentions of the user, the challenge will produce bad data. In this case, the creator of the challenge should be held responsible. And a challenge can be excellently designed but if a user decides to ignore all this and simply press yes despite every thing at MR telling them not to, then the user is at fault, not the challenge creator or MR as a whole.

The “guns don’t kill people” argument proports that a tool is just a tool and misuse of it is completely on the user, which to some extent, is true. But of course, if a tool has a fatal flaw or it is really easy to misuse, it should be considered who has what kind of access to it.

Someone who is new to OSM and

This sounds to me like you are criticising that there is not enough MR is doing to encourage/inform/(force?) people to understand what they are doing and how they should be doing it. Am I understanding that correctly?

Does MR fail in that regard in your opinion?

StreetComplete goes out of its way to make it hard to break that rule; Maproulette goes out of its way to make it easy. From a DWG point of view I think we’ve had exactly one complaint (out of about 300 a month) that involved “non-local StreetComplete”.

yes

2 Likes

Hi. I haven’t followed this discussion. What’s going on here and what does it have to do with me?

Read the two paragraphs immediate before the bit you quoted and you’ll find out :slight_smile:

@SomeoneElse I wrote detailed instructions for mappers, saying they should review it and read the documentation. If people interpret that as “delete all these layer tags”, that’s not on me.

I will grant that you wrote instructions, but at less than 50 words they do not come across as detailed to me :smile:.

1 Like