As for the hardware, I agree with you that the one machine per screen setup is inefficient, but while it is not the best technical decision it may be the best business decision. Almost certainly the computer running the screen is built into the screen assembly itself making a simple unit which can be sold as a scalable commodity that which can be shipped and wired up by relatively unskilled labour.

Additionally having all the machines running off a central machine would present a single point of failure – instead of one broken screen, 100 could show a BSOD on a failure or worse…

I would be surprised if there wasn’t a central update facility, but that said the prospect of some nasty cracker putting something on all the TFL screens would be enough to scare most managers (who would have their jobs on the line) and enough to push a fat support contract to TFL and the advertisers which is where a lot of these guys make their money anyway.

Actually, using a mobile OS might not be a foolish idea: in the future I could imagine that these ads might interact with the individual passing them on the escalator, Minority-Report-style, and peering might be more reliable if you’ve got Android at both ends. Ha! :-)

Any IT department can ghost an image onto a computer. Job done.

Clever people know that a small amount of prep time – getting a working linux image and a cron job to fetch video packs – will save them a lot of headache in the long run, but Linux, or NetBSD, or FreeBSD, or OpenBSD, or any of the others are all too niche. Too unknown for non-technical people.

While the entry requirement on knowledge is falling – and the average user (as opposed to developer or enthusiast) intelligence of the open source software world is dropping as a result, the usability hasn’t proceeded at the same rate and the intelligence needed for average companies is still too high.