> The "one laptop per child" program put some thought into the connectivity problem.

OLPC was also ruggedised, had replaceable parts and had things like crank charging for areas with no/little power, it was a well thought out machine all around. Even in developed countries there are many that can't afford the internet, electricity and repairs required so it's a shame to see schools go with regular lapotops or even worse, premium tablets. One of my nephews had an optional programming class that required an Ipad, it's the worst option in just about every way.

> OLPC was also ruggedised, had replaceable parts and had things like crank charging for areas with no/little power, it was a well thought out machine all around

No, it did not. The crank charger was never supplied commercially, it was purely a marketing tactic. Same as the Potenco Pull-cord generator http://wiki.laptop.org/go/Peripherals/Hand_Crank , http://wiki.laptop.org/go/Peripherals/YoYo

Both were ideas which even basic calculations tell you aren't viable.

In addition, your other claims like "ruggedised", "replaceable parts" contradict what I saw with OLPC XO. I would say the product was a one-off that had almost no consideration for long term reuse. Things like the wifi bunny ears prove that the product was driven more by marketing tacticians rather than reality and cost. Honestly, my opinion is that OLPC was a huge scam to take money from unwitting unsophisticated developing country funders who didn't know any better and waste it in swanky offices in One Cambridge. It was one of the worst options about how to expend public funds.

At the moment though this would be way easier with a plethora of ARM boards that are both low cost and powerful. Also with star link there would be a way to solve spotty internet access in remote areas. What is needed I suppose is a way to integrate this with preexisting organisations in the area.

I was working a project where the BOM was to be at most 50 bucks. It is kind of difficult to hit that number. Ironically while the board was cheap it was all the additions to make it a full computer that ended up ruining the idea. That was even before we add in any profit margin. Add in a screen, keyboard, mouse, power supply, and so on. It adds up quickly. What was a 15-20 dollar computer is now 70-80 or more plus paying someone to put it together.

Could it be kept fairly close to $50 if it was simply a "box" with a few relatively high speed USB ports and an HDMI connector?

I know this wouldn't be a "laptop", but with a cheap-ish monitor, keyboard and mouse... it's a small PC. Perhaps with a slightly older CPU which doesn't need super fast RAM, there could be... 2Gb of RAM or more?

Of course the newer RaspberryPi boards are pretty close (and in terms of RAM, even better) to this....

Oh we got 'close'. But the issue was it felt like 'oh yeah we need X' that X would be 2-3 more dollars, and so on. Really at this point the board is the 'meh' part of it. When you are at this price point. Something like an extra USB port becomes a larger portion of the bill. Also at this price point something odd happens. Your support costs also grow much faster. As you will attract more people to the platform at a lower cost. So your margin is not there. Also when you build 1 or 2 you can do that in an afternoon. But when you want 10k of them it becomes a bigger issue of simple things like 'how do I get these into the cardboard boxes', or 'where do I source 10k power cables that are not garbage'.