My mistake, i do see lots of websocket connections to relay nodes in my browser when using iris.to, for instance. I would presume all web clients do this or at least the better performing ones?
So what is the true requirement of a web client - to serve up the static html/css/js and the browser does the work from there? This should be a nothing-burger unless I'm missing something.
Yes that's all a client needs to do. For additional services like text search you need a search server etc
Ah yes, makes sense. I think a web client would be well advised to install a Lookup NPUB button that does a direct GET to a profile URL. I'd imagine 80% of the traffic are lookups from Twitter npub shares.