Tecnology is evolving at a rapidly increasing rate. The old goes away, replaced by the new. Yet some things remain with us for a long times. I wanted to list here the things that I think will not die until there’s a super-fundamental shift in how we do computing (how fundamental? more than the vacuum-tubes-to-transistors transition, that’s how).
- x86 instruction set: used in all PCs nowadays, able to fit into smaller and smaller spaces, and no longer actually implemented except as a translation layer. It’s an arbitrarily defined bytecode which is supported by everyone and everything.
- ethernet: they’ve said it before, and I’ll say it again. as long as we will do packet switching, we will do it with ethernet. From 10MB into 10GB, and from wire to wireless, the ethernet format has not changed. I doubt it will.
- TCP: they just keep on throwing more optimizations on TCP. It’s as reliable and efficient as possible, and every competitor is just subsumed into more TCP stack algorithms.
- IP: I am hoping that it will at least transform into v6 from v4 — but that is the only possible transition. It’s possible that v4 will have more and more optimizations thrown at it to cope for the new world, or v6 will take over. In either case, it will be IP.
- HTTP: It’s ugly, arbitrary and dead simple. And it will remain, because every computer on the face of the earth has support for it. It’s been extended over and over to cover more and more cases, and it’s got smart negotiations built in. It cannot die.
There are more things which I think will not die, but let’s leave it at that, shall we?