Funny thing, I'm at the Computational NeuroScience (CNS) conference right now, and the previous talk started with a satirical comment on Markram's statement : "a leading scientist of our field just said we could make artificial brains in ten years. let's start now, we have a lot of work to do."
... window managers were actually managing windows.
I mean, the floating windows paradigm is so flawed nowadays that applications have to manage their own set of windows!
And web browsers are not alone: any application that lets you view/edit several files within the same window is managing something it shouldn't have to manage.
Alternatively, WMs could implement the equivalent of "tab circling", which is really just the ability to circle through a subset of windows.
In that sense, tilled window managers are doing a much better job than traditionnal WM, especially those implementing tabs.
There's something I find really sad to read. At the begining they are talking about fighting global warming. Nice.
And then, you have this :
"First Solar is great, as long as we're talking
megawatts or gigawatts," he says. "But as soon
as they have to start rolling out terawatts,
that's where I believe they will reach some
limitations."
Well, another solution could be to promote a lifestyle that doesn't make you count in terawatts.
But no, we won't do that because we are too lazy, and some technology will solve
Of course I'm not saying solar panels or technology sucks, it's more about the way it is used. It reminds me of this prank made by the "Yes Men" to a french politician, tricking him into believing that boeing was planing to drop huge ice cube on the north pole to fight the polar ice melting.
If only I could be modded up to eleven
The first thing that comes to my mind is my field of research : "numerical" artificial intelligence (as opposed to symbolic AI).
When you run artifical neural networks, you usually do it to process noisy, real-life data. In such a case, you just don't give a damn about precision : this is just a bit more noise, and your processing paradigm is studied to tackle with noise.
Right now, we are definitely missing horsepower in this field. With a state of the art computer you can only hope to do very basic stuff in realtime. Unless you have a lot of money, you are stuck.
Such a technology could be a gift for us. ANN need raw computing power, probabilistic chips need noise tolerant computing - it could be a perfect match.
I'm already having wet dreams of PCIe cards with a few GB of fast access RAM and a massive number-crunching probabilistic CPU
Life would be so much easier if we could just look at the source code. -- Dave Olson
