That was tried by the Nabu Network in 1983 on cable TV systems, broadcasting data at 6 megabits per second! https://en.wikipedia.org/wiki/... It could have worked over the air just as easily as any TV channel.

The receiver + computer just had a bootstrap ROM to download the OS from the network, then that would fetch the menu of the day, and then it could download the program you wanted to run. Since it was unidirectional (designed as bi-directional, but cable TV hardware of that time didn't work that way), the server at the cable company would repeatedly broadcast a cycle of all the available files. And of course since it boots from the network, software updates are a matter of just updating the files on the server.

Unfortunately, they sold it as a home computer with software options, not a pure data service (imagine a BBS with files available for downloading a thousand times faster than the 1200 baud of the day). It rapidly became obsolete and failed quickly. By the way, old new stock Nabu PCs are being sold on eBay and a group has gotten the old software working again with a simulated network (see https://nabu.ca/) and is making progress with new software.

Just what is the reliability difference? Small transistor size chips have been pocket tested in mobile phones, with thousands of steps of walking. But not against car vibrations, pothole plunges, and much hotter and much colder temperatures (temperature may be a bigger problem).

Then how long do those chips last before they degrade? Can they go for 20 years? Look at the problems Commodore 64 owners have with a certain PLA failing after a few decades (it runs hot normally thus the shorter lifetime). And that's a chip with relatively large transistors.

There's a nice chart at https://semiengineering.com/ma... which suggests 20 is the upper limit for automotive chips. They also talk about building chips with monitoring for degradation. They also suggest traceability of chips right back to the manufacturing batch, so you can notify people if they've got a short life chip in their car and inform the chip makers that there's something to improve on.

Normalizing their stars is an interesting idea. So is giving value to someone using a variety of stars, hopefully making the voice louder of people who think about reviews, rather than giving 5 stars every time. Then you could get into metamoderation (like Slashdot does) to rate the quality of a person's reviews.

For my less evil reputation system, I'm using a weekly allowance of rating points (say 10 a week). Sure, you can award 5 stars, but it will cost you 5 points. Same thing for a bad review, with 5 negative stars also costing 5 rating points. Thus I'm trying to use psychology to get more meaningful reviews from people rather than using statistics to to extract a signal from noise. Related essay at http://web.ncf.ca/au829/Weeken...

Looking over these comments, I see that meta-moderation is worth stealing for my less dystopian reputation system project (under construction; currently it's mostly an essay (inspired by frustrations with Black Mirror and Orville) and a database design document for a system with forgiveness over time).

My reputation system already rates things (posts, users) sort of like Reddit with up/meh/down votes. I'll have to add rating the opinions too (you liked that post with +2, I think that's wrong, here's -2 to the post and -1 to you), though the user interface may have some work to do to get the point across that your opinion is about an opinion.

Meta-moderation would be done by allowing anonymous opinions on randomly selected opinions. Like Slashdot, you'd get the chance once in a while to meta-moderate a few opinions. The reward would be effectively payment with a few extra rating points to use elsewhere, so there's a bit more of an incentive to meta-moderate. People good at moderating would be noticed, and allowed to join a group of moderators. Moderating the moderators would be done by an ancient technique - performance reviews by the staff.

Time to update that database design with a few new record types inspired by Slashdot...

I agree, Slashdot's moderation system is pretty good. Haven't seen DPReview. I looked at Slashdot, Reddit, and Twitter (and some background papers) to get ideas for my paper on making a less evil Reputation System http://web.ncf.ca/au829/Weeken... Yup, after seeing too many Black Mirror and Orville episodes, I thought I could make a better system. Still evil, but not as much.

Reducing the points you can award (like Slashdot, so you have to think) and some sort of trust or meta-rating of people's rating ability are both good steps.

Thanks for the links. I was wondering what their power source is, and the Wikipedia article has the answer. Apparently they're turning diesel fuel and a bit of wind into plant growth.

I do something similar to the maildir approach, converting e-mail (from mbox files from long ago and from collected Usenet messages too!) to individual text files. Each text file contains the raw message text, encoded attachments and all, so nothing is lost.

However, I'm using BeOS's BFS to store these files so they are a more quickly searchable than plain text files (subject, from, to, dates are all indexed by the OS). Plus I file them away in directories by category. I also wrote a conversion utility [MailboxFileToBeMail] for mbox and Usenet messages that automatically attaches attribute tags to the message files, and even does international character set conversion on the subject and other key text.

- Alex