There is a fair amount of misleading information above, so I will attempt to give a more technically complete answer than I did before.

The HTTP protocol is based on a definition technically called RFC2616. This is specifically designed so that operating systems (iOS, Android, Windows, etc) and word size (64, 32, 16, 8 ..) do not matter in the least. In fact, the whole purpose of the internet protocols is to allow dissimilar computers to communicate effectively. I have used it with 24 bit architectures with 6 bit bits, for example, and defunct operating systems like VAX/VMS.   Moreover, it is a standard: the age of the implementation does not inherently matter, though it can do in some respects. For example, if I try to comment to 'The Independent' web site from my iPad mini, they have made the whole website so complex it frequently times out because the iPad cannot cope with how long things are taking, so assumes the server has gone away. But it is 'old consumers' of the data where this is likely to be a problem, not 'old providers'.

My experience is that when I try to connect to http://mudcat.org something switches it to https://mudcat.org. Whether this switch happens in my browser or the server I have not experimented enough to determine. Internal errors, for me, occur when I try to connect to 'https://mudcat.org' but not to 'http://awe.mudcat.org'. That suggests one of two possible sources. Either the security management is having trouble, or the load sharing mechanism that is trying to allocate 'mudcat.org' to a specific server is having a problem.

https establishes a session. It is possible - I say no more than that - that it partially established a session and then failed, and subsequent requests are attempting to use that same faultly partial session and so also failing. But it could be something else entirely.

So as I said originally, it is something that Max may be able to fix, or may have more information in server logs that tell him more precisely what is going on. It is not for us, as consumers of the data, can resolve. All we can do is use a workaround. By avoiding https and using http, and explicitly selecting the server (awe) things seem to work most of the time.