Posted 06 May 2011 - 05:56 PM
Amazing the amount of technology that can be had in a ~$25 dollar chip these days.
Also interesting to note is the relevence of high-volume gaming technology that is applicable to this area of building cheap PCs for the third world -- The chip here, and many like it (that is, graphics and sound hardware + moderately fast, if simpple, CPU) that would be readily applicable to similar efforts often find their volume in handheld LCD games like the Canoo, GPX, etc -- and in even-cheaper chinese knockoff handheld emulators/media players. There was also an effort awhile back to produce a $10 PC based off commercial hardware that was available for not much more than that on the streets of India -- inside the keyboard was a SNES-on-a-chip.
Many great programmers came up through having a Commodore 64 when they were young, so you don't need much. Having a $10-$15 dollar computer of that caliber widely available, and able to plug into the old televisions that are all over the third-world is, I think, generally more important.
I like this device (already had read about it this morning) and I think it could really benefit *institutions* such as libraries and schools in the third world, but it won't go very far into the homes of the third world due to the HDMI/DVI interface.
If you want to get into the homes/hands of the third world, you really need to either use the existing technology they have around, such as old televisions, or integrate the device with everything it needs to function -- LCDs, battery/power/solar, input. I think we can do the latter for around $100 today using commercial technology, but we're a long way off from 10, 20 or even 50 bucks. The former, we can do today at incredibly low cost -- $20 or less.
I think a lot of these efforts, even including OLPC, fail at recognizing what is a need vs. what is a want (in as much as any computer is a *need* for someone who may not have clean, running water). They end up producing what is, essentially, a high-tech toy that may be wonderful for learning, but which is ultimately too expensive to deploy widely -- even at a price target of $100. I believe the best way to get these things out there is to make them something that the people, themselves, could afford, and not to price them where only governments can provide them.
If I were doing this today, I would use an arm or MIPS -based SOC, with inbuilt hardware to drive a TV, or a simple external circuit/encoder to drive it. It would also support analog VGA for higher-resolution, low-color graphics. It would be built into a keyboard shell -- something the size of one of those cut-down, slim keyboards with only the qwerty keys (no 10-key, no inverted T). It would have a 9-pin DSUB connector on each side for gamepads/mice/input devices, headphones/mic on the left, and an SD-card on the right. It would have a built-in MIC, and possibly small-builtin speakers. It could be run from the wall, using Micro-USB (cell-phone standard, widely available) to draw power and serve as an administration/service port, and possibly to network the devices. It would also run on simple AA batteries. I would expose any left-over GPIO pins on the back, possibly using some to impliment a parallel port. It might have only 128K-1MB of memory, so a full-blown OS is out -- but it would have a single-user OS in the firmware, including applications for basic tasks and for programming. Applications would be allowed to take over the entire device if needed, but would primarily run in a simple multi-tasking (possibly co-operative) kernel on a simple desktop environment.