The real geeks I know would at least dream of building a computer of which they control all pieces of software. Maybe for the achievement only, but they at least dream of it.
I am pretty sure that RMS gets his work done as well, whatever that work is. I wouldn't suspect him to be _lazy_.
What sort of ultra-geek uses that sort of setup? The current generation of that notebook has a processor that runs at less than 1GHz! The only reason he gets away with it is that he mostly does mail on emacs from console.
It's too minimalist for any present day software development work
I know a reasonable number of present-day developers who use a setup not much more complex than that: emacs or vim in a console. Not very practical for webdev, but you can do scientific-computing or embedded dev that way. For scientific computing you'd probably want a beefier CPU to run tests, but those are often run remotely on a computing cluster anyway.
On the contrary! I spend most of my working day using vim in a terminal (with tmux) to do web development. It's just text generation, after all.
I'd go so far as to say that if you're doing test-driven web development, it should be possible to build a system entirely within a terminal, and that's what I aspire to. It's only the front-end styling (CSS etc.) that actually needs anything more than a terminal.
And yes, I find this efficient: far more efficient than hacking and pressing F5. I think there are other benefits to this way of working: it tends to make progressive enhancement and accessibility the path of least resistance.
Yup. For embedded dev I use Vim in a console. It's convenient to have a big screen so I can have multiple terminals and reference materials open at the same time, but using a tiny computer like that is not out of the realm of possibility.
If by "any present day work" you mean "web development," then maybe I buy your point.
I know plenty of climate science and financial programmers who spend 100% of their development time SSHed somewhere else, or who write reporting code using a minimal test case set of data that runs in 1s on their (slow, old) machine. For these people a slow, old machine is often totally acceptable - it doesn't take much horsepower to SSH (or even to run emacs and perl locally).
I could do 99% of my job effectively on a ~200MHz Pentium with maybe 64MB of RAM. About the only thing that would really bite me is the occasional need to help the client team, which often means booting Windows in a VM and replicating their comparatively bloated development environment.
I was using a 633MHz PIII until about 3 y ago, then using it again recently when my main computer died, and its gotten a lot harder (on Linux) over those 3 y.
No, ubuntu specifically on that machine. The only tricky part as I can see it should be the webbrowser. You'd either have to use an old one, or a featureless one.
I was using an old one (Firefox 1.5) and it was extremely painful with Amazon.com. Amazon.com seems to have changed in the 3 years since I used it last with that browser.
I always thought that a "featureless" one (one without modern JS or CSS) would be extremely painful or useless, too.