Agree, but an interesting point being made here is that as we continue to add more layers of separately-attributable infrastructure the attributions get quite unwieldy.
I don't have a conclusion from this, and if you're going to use something that requires attribution you should attribute. Maybe we just need better ways of giving that attribution. It's certainly easier in an interactive medium, with hyperlinks and collapsable sections than on paper.
Well, even in hypotheticals, I'm not sure how many cases are likely to go beyond 3 layers. And that's just for journalism.
For practical functional applications, 3 layers seems high, and potentially a sign of over-engineering.
To get a little more technical about our hypothetical:
- I've used Mapbox in the past, and they included (at the time) a succint dual-attribution overlay for themselves and OSM as part of their APIs. It was well-designed and baked in (nothing for me to do).
- If I were to then take my own application and offer it as a service for others to embed:
1. There's a strong likelihood I'm offering a commercial service so attribution may not be required for my layer (provided service-charges covering white-labelling)
2. If I am requiring attribution, it might make sense for me to invest in direct OSM usage & cut out the Mapbox middleman.
3. If not, it certainly seems like my added layer would be the absolute extreme (and baking in a well-made 3-component attribution would be challenging but not insurmountable).
I don't have a conclusion from this, and if you're going to use something that requires attribution you should attribute. Maybe we just need better ways of giving that attribution. It's certainly easier in an interactive medium, with hyperlinks and collapsable sections than on paper.