HTML is machine readable it just has problems with the machine understanding it part. I think you are asking more for a semantic web which hasn't been particularly successful so far. People mostly struggle to just put together a web page that people can understand. Getting them to build diverse information to standards machines can reliably understand is hard. XML has mostly proved this.

I'm not necessarily opposed to their being attachments sitting on Twitter's servers that I can access if I'm interested. I just totally dont want them bombing me with it and sending with the message by default. Their messages are already too bloated. And as I said the fatter the data is the more problems you are going to have with copyright, legality, offensiveness and security risks both for the service and end users.