As a Linux box it is pretty slow and underpowered.
As a desktop PC
, it's slow and underpowered. But there are millions of Linux boxes out there that aren't desktop PCs: routers, WiFi APs, NAS systems, kiosks, industrial automation controllers, set top boxes, and lots of others. At least many hundreds of different products, if not thousands. The RPi, in theory
, is great for many of those (although that's starting to look a little iffy in practice).
As a teaching tool, well what is it teaching? Linux? (could do that on a PC).
But at what cost? The RPi is cheap enough that many schools in industrialized countries could afford to issue them like textbooks. The "computer lab" then becomes just a collection of monitors, mice, KBs, and whatever course-specific hardware the school chooses. Middle- and upper-class kids would be able to take the actual course system home to do homework (poorer kids are less likely to have an HDMI monitor at home. I think this was a gaffe on the RPi designers' part, because a lot
of kids, especially in developing countries, who could benefit from having a cheap learning platform, would find it easier to get access to older, non-HDMI monitors).
In that environment, the teacher will be the one who picks and configures the system software. Which could change from semester to semester. Or even hour to hour: let's say you're running a school in a poor area that can't afford an RPi per student. The first-period kids are learning web design. They get an SD card with a Linux server system that you know
has Apache x.x, PHP y.y, etc. The second period kids are learning embedded: they get an SD card with an RTOS, or the Arduino IDE.
This opens the possibility of schools teaching a variety of skills that are completely out of reach now due to hardware costs.
But, of course, this isn't going to happen unless they get past this phase where each RPi is set up in isolation by its owner. It's like the very early days of the Model T, when they were being sold into rural areas where there was no infrastructure of paved roads, gas stations, and competing Pep Boys/Kragen/Auto Zone stores supplying parts and accessories. The early adopters had to self-teach and improvise to get the benefit of the new technology, but the infrastructure developed quickly as cars became a true mass market.
The big problem isn't that the RPi needs new software: what it needs is the volume that will motivate people to package the software that already exists in a newby-friendly form.