When I taught my son programming when he was around 11 or so, the project I picked was to have him write a stack calculator, including a simple GUI (like a simple version of the GNOME or KDE calculators). It taught most of the basic elements of a program, including a simple GUI, but unlike a really simple game who's novelty would wear off quickly, it was something he was able to use, improve, and be proud of for some time after finishing it.

As to what language to use, that's a hot topic and you'll hear lots of opinions. At the time, which was a long time ago, we used Tcl/Tk because it was simple, had a very easy-to-use fully integrated GUI toolkit (Tk), and was high level in that it avoided having to worry about complex issues like memory management and pointers, which could overwhelm some 11 year olds. OTOH, it taught programming concepts beyond what you could do with something really basic like LOGO. I'm not necessarily suggesting it now, however -- it might not be the right thing today, when there are other potentially better choices like Python, Java, etc.

Although many patents (both software and hardware) are bogus, the basic concept of the patent system has some validity and there are conditions where patents serve the public interest by encouraging innovation and at the same time making knowledge available to the public which would otherwise be kept as tight trade secrets by companies. In the case of universities, they have been loosing other sources of public funding and so earning some money from patent licensing may not inherently be a bad thing, but there should be requirements for patents obtained based on publicly funded research that although licensing fees could be charged for use by private companies, other universities and other publicly funded research institutions should be allowed to use the technology royalty free.

It shouldn't be hard, as long as you have any kernel/motherboard combination where suspend or hibernate work reliably. Just create a shutdown level that actually does a reboot, then modify the startup scripts to immediately do a suspend or hibernate as soon as the machine is booted if that shutdown level was used. The effective shutdown time will be longer (because it's actually a reboot), but the effective "boot" time will be very quick.

While probably do-able, this actually seems like overkill. Why not just use normal suspend/resume?

I have many gigs of digital photos and I have also more-or-less moved away from optical media for backup and switched to HDD. As the original poster mentioned, most of the "information" you find on the net about archival longevity of optical media is personal anecdotes or pet theories, and good hard data on archival longevity of CD-R or DVD+-R is hard to find. My own personal experience is that name brand discs do have fewer problems than cheap "house brands", but it's hard to quantify or say much beyond that.

Backing up to hard drives has a number of advantages:

1. It's a heck of a lot easier - in most cases of personal data backup, a few 1TB HDDs will hold all the data you need to back up, so there's no need to manage boxes of 100's of discs. I usually back up the same data onto two HDDs, and store one of them in a firesafe. If you're really worried, you can store one of them offsite.

2. Since no media will last forever, you will *always* need to roll your data over to new media every so many years. With HDDs, its *much* easier to roll your data over to new media every 5 or 6 years. Think of transferring two or three HDD's to a new HDD (by the time you roll over the data, the new HDD will probably hold all the data from those two or three older HDDs), compared to re-organizing and re-burning hundreds (or more) of CDs or DVDs.

The bottom line is that if a few HDDs don't hold enough data for your needs, then backing up to optical media will be totally out of the question anyway, and you will probably need to use tape.