**** Cyborg Central ****

Is multitasking really that hard?

One of the things I've been thinking about a lot is how we've gone from single-program 8-bit computers to these huge, powerful multiprocessor systems, but the method of working with them has gone full-circle back to the single-program interface.

Phones, tablets, and even the ugly but commonplace Windows UI have all thrown up their hands and given up on windowing and moved to a simpler, one-screen-one-task interface. For someone like me who finds it easy to multitask, the UI is incredibly frustrating, but for the average person, it seems like this is about what they can handle, so they like it.

What happened?

Back in the 90s, the PC and Mac were taking tentative steps out into true multiprogramming, the NeXT and Amiga were already there, but only NeXT had realized that screen real estate was essential to having a good multiprogram GUI. At some point we hit the line where more resolution didn't help.

The same went for OS isolation - we were doing pretty good with multiprogramming, then virtualization came, and we gave up on that. Then even that was too heavy and containers came.

We just can't get away from the one computer per app model. It's just TOO HARD. Apparently.

So I wonder: at what point do we throw away everything but hardware-provided multiprogramming and just go the fuck back to the actual one-computer-per-app setup? Do we strip Xen down to the bare bones and just run DOS? or EFI?

How shit is that?

Categorized as: Linux