Jonathan Blow

The public version of the Web started taking off in 1995, around the time Netscape Navigator was released. Here's the World Supercomputer List for 1995: https://www.top500.org/lists/top500/1995/06/ … A Coffee Lake GPU in a random laptop is almost double the performance of the top of that list.

So what we are observing is, since the Web started, it has become so much slower that a supercomputer would no longer be able to run it? Does that make sense to anyone?https://twitter.com/Suhail/status/1387063023521595394 …

But the tides are already turning and more and more web developers are starting to understand that the whole Node thing was a big mistake. Look out for Elixir based sites pushing server rendered content over websockets.

Server-side rendering is an *absurd* solution that depends on a severe misunderstanding of the problem in order to seem to make any sense at all.

So SSR is basically what almost all servers do today, but they have a tendency to push a lot of logic over to the client - making the browser do a gazillion asynchronous function calls on the server.

I will agree that it is silly and unnecessary to do a bunch of asynchronous function calls. Communication should be kept to the minimum necessary to do the job. But, this doesn't mean the server should do more work, it just means the client should be architected less terribly.

I believe the server could be a lot more effective if we stopped using JavaScript and did the work in a proper language. JavaScript on the server side is a bad idea for so many reasons.