[1vuio0pswjnm7]

As a "collector of reliable DNS servers"^1 I can report there are DoH servers that will actually take a traditional DNS query that does not support EDNS0 and, perhaps using the client IP from the TCP connection, return a response that includes EDNS0 Client Subnet (ECS). Whether the DoH provider is sending the ECS to authoritative servers I do not know, but to me it is quite sad to see this being returned in the response given I did not request it. Anyway, ECS is supposedly the reason 1.1.1.1 does not include DNS data for archive.is

The site once used a tracking pixel as a poor mans ECS. The client IP address was inserted into the image name. Apparently the operator of the site explained this was used to achieve CDN-like functionality:

https://news.ycombinator.com/item?id=27501867

1. Perhaps we should be clear that "servers" here means open resolvers. These servers are of course not authoritative for any name, and generally recursion is slower than iteration, i.e., use of authoritative servers only (fee free to challenge me on this and I will share a citation, although I know this is true from own experiments). Thus "reliable" is perhaps ambiguous. Not all of them always return the same results. Some will return different answers, and not always for "load balancing" reasons. Some may be missing data entirely. Some will return wrong answers, e.g., pretending to be authoritative. Much DNS funny business on the internet today. I gather results from a variety of resolvers, from authoritative servers as well as other sources of DNS data, e.g., public zone files, scans and crawls, and I compare notes; I personally would not feel comfortable using one open resolver (third party DNS) as the source for all DNS data; I could not rely on it. As such, "reliable" is IMHO a loaded term if used to describe open resolvers.