Serve watermarked tiles to known offenders who neglect attribution

We are constantly monitoring map widget embedding and keep adding offenders to this list:

Some of these may eventually be dealt with by legal means, but it would be great help to us if we could add watermarks as soon as such uses are discovered and only remove it after the dispute had been resolved.

Differentiation could technically be implemented in many ways, like by simply checking request headers (probably Referer or Origin) or potentially introducing newly differentiated subdomains or query parameters:

Watermarking could be either done on the fly via smart compression tricks or pregenerated to a separate folder and the request rewritten to be served from there.

The watermarks could be transparent, small, subtle and maybe have a little wiggle in their positioning as long as it complies with the attribution guidelines.

What do you think? Is this the right channel to propose this?

Yes this is the wrong place to propose this.

Not to mention that the list itself is a silly idea to start with.

Indeed we could start with a blank list, and it could be maintained in either the public wiki or at some other place. (It will also need some more columns involving the headers to filter on.)

So, should I CC this question to the tile-serving list perhaps?

I felt to be insufficient for this, and the OSMF contact page recommended that I first post a query in public before bothering you.

It’s an interesting idea that OSM France has tried before with their #AttributionIsNotOptional initiative.

Of course, this only works for sites which use our tile servers. Some quick checks of entries on the wiki page show that quite a few of them do not. Could you offer an estimate how many known offenders we’re talking about here?

Another consideration is how feasible this would be with the operation team’s infrastructure, including Fastly, and how much work it would take (and by whom).

I’m not aware of any better place to publicly discuss such an idea than the osmf-talk mailing list or this forum. What would you recommend?

I haven’t looked into estimating proportions yet, but all the ones that we have found in Hungary were (are) using’s tile servers.

The ones who I know were smart enough to set up their own tile servers were smart enough to include attribution as well.

Surely this would be a partial workaround, but I think if there was a tried and good configuration that was being used by the official tile servers, the other open tile servers could copy the good example as well.

Maybe some of the website operators would even consider this a feature so they didn’t need to hire a coder to customize their map widgets. We see it pretty often that they don’t even realize where the data is being server from - they just enabled a module in their CMS for showing a slippymap and they were done with it. Others copy & paste from Leaflet examples, and leave out attribution because it is either missing in the examples, or it is in English and the website is not, hence they don’t want to cause confusion for their visitors and didn’t know it was mandatory.

The benefit of such enforcement would be twofold: it would both be nice from a copyright perspective and it would inform website operators who is providing the infrastructure for them for serving the tiles. Hence maybe they could customize their code, choose or even host another tile server or perhaps donate even?

Regarding the amount of work, I think we could delegate part of the work to mod_tile so that others could benefit from such customization as well. On a proof of concept level, I think it could be done pretty quickly by just altering Apache’s config and add hacky custom Mapnik cartocss rules or patching renderd for the text (haven’t looked into its implementation either).

osmf-talk would have been a canonical choice, but I am not an OSMF member and that list was specified as for members-only.

As a matter of policy to suggest it to the OSMF board, as a matter of operations to suggest it to the OWG (and I’m not even nitpicking yet).

The major problem with the list is that it isn’t actually maintained, just as essentially everything in the wiki. Leaving the OSMF and the persons who added “offenders” open to civil libel claims. If you operate a pillory it is somewhat your responsibility to check if you should let, alleged, offenders go because the offense is no longer or never existed, at least now and then.

Indeed those are valid points and it’s great to have such legal insight here.

I’m against public pillory in general (especially if it serves no immediate purpose or campaign), because it could undermine possible future relationships and as you highlight the burden of proof is on the reporter and only a judge can claim a verdict in general.

If we did have a table for tracking “unsettled/in-progress cases”, a last check date column and periodic reminders are a must. It may indeed make sense to hide the full list from the public (perhaps make it queriable by domain via a backend or only publish hashes in the open wiki). Viewing or altering the sensitive details should probably be limited to those who are involved in reviewing statuses in general, are contacting these entities and to official OSM personnel.

As one with software experience, I could actually imagine a bot being implemented that could regular check the mentioned pages and/or the map widgets to monitor any changes and notify the reporter and/or the table maintainers if anything improves for the better or worse.

I could also actually imagine a web crawler that is continuously looking for typical patterns of attribution failure on the top ranked websites, but let’s not open the scope too much.