Eh?

At the time Netware was popular, half the offices in the world still had serial ports on the wall for a terminal. Networking was for big iron. Novell had some massive installations (and others too, 3Com comes to mind with it's XNS based stuff that eventuall became Lan Manager). We're talking tens of thousands of employees in a single setup, not just lawyers offices with two guys and a secretary.

Netware could have been a contender but it smelled too proprietary in an era when "Open" was the catch cry. Worse than that, when TCP/IP re-emerged with the *nixes as the proprietary mini computers died off, you had to load TWO networking protocols on your PC which resulted in all kinds of shenanigans over those 10,000 people. Novell got the idea far too late that it was easier just to run TCP/IP, ported their higher level protocols off IPX/SPX too late and gradually got white-anted by Microsoft bundling Lan Manager with Windows NT. It wasn't fast either, there were still massive installations of Netware up to five years ago because Microsoft took forever to get their directory services working.

Heh. OK, I didn't think my cunning plan all the way through.

It's unproven as the take-up is low, especially outside of the US. Problem one is that because we have no effective micro-payment systems, these things are subscriber based or require the purchase of expensive dedicated hardware like a TiVo (a complete flop in Australia as it happens). Hulu, not available outside the US at all. Ditto for Netflix (I assume because rights to the shows are sold on a regional basis). The only companies at the moment who can afford to buy the rights here in Oz offer some of the most expensive, subscription driven services and their subscriber numbers are either steady or falling.

Second problem is partly related to that: Minimum "on demand" here is approximately $5 an item, which is a lot for something that you will have restricted, time based access to, but the economics from a provider point of view don't add up unless they charge that kind of dough (again, micro payments would help).

Lastly, it's one thing to want to watch "charlie bit my finger" on demand, but most content in the 20m/40m/movie length traditional broadcast TV formats aren't consumed like that. Generally, some consideration is required if you have decided to park yourself on the couch - wasting 20 minutes trying to decide what to watch just doesn't work for most TV oriented viewing. I still think the two different modes of watching content (either pre-planned and saved up, or "what's on the telly") won't merge.

Probably not, but "on demand" is what's unproven, not the technology itself. If somebody came up with a hybrid service that basically shuffled a TV channel of stuff you like (you tell it how long you want to watch, it knows your preferences or you seed it), the Google people could queue up shows for you (with a few wildcards of things you haven't seen but might like, pilots for shows that you might be interested in etc).

That way you could still use your TV as a passive entertainment device without necessarily being tied down to what some broadcast programmer decided you wanted to watch last year when they were doing their media buying.

TV is a passive medium for the vast majority of people and that's how they like it. Plop on the couch, select your channel and let somebody else make the decisions about what you'd like to watch. Most people don't have a home media server and don't understand why you'd want one (because nobody has explained that your DVD cupboard is basically a sneakernet server, and having everything you want to watch a button click away like music just hasn't happened yet).

Actually that's a good analogy - we have a home media server and various cobbled together clients around the house, and it's interesting to watch the usage patterns: Music videos get shuffled like a giant video ipod on the main TV, it's like the best MTV that never was with the bonus you can skip stuff you don't want to listen to. Kids want to watch 3 or 4 episodes of iCarly in a row. Parents want to be able to consume a serial like Dexter without the annoying "wait a week for the next episode" that broadcast TV forces on you.

Most people will want this stuff, they just haven't seen it - so do Google a favour and invite your non-tech-savvy friends to a demonstration of your media serving rigs (assuming you've gone to the trouble of making it demo friendly and can resist the urge to fiddle with technical stuff while you're showing them). What is going to be a killer is pricing - if Google could negoatiate to broadcast a channel of cheap stuff so the "plop on the couch and watch" crew could enjoy another TV channel without having to think too much, they may be tempted to purchase premium content like first run serials without the hassle of torrents.

If the service benefits most US taxpayers. Besides which, if I were going to spend a fortune on sensors, I'd put them on garbage trucks instead. You *know* the garbage is getting picked up, but the mail truck doesn't necessarily go everywhere, all the time

Alternative timeline: Sun should have seen the writing on the wall when NetBSD and Linux started to get popular in the early nineties. Why? Because a lot of us ex-Sun jockeys really, really wanted a Sun at home but just couldn't afford to run even a second hand IPX workstation when a PC was so cheap. Sure the PC was a piece of junk, but loaded up with X windows and all the Gnu tools, you could get most of your support scripts working from home and started not worrying so much about having a Sun.

If they'd have looked downwards to their primary users instead of trying to capture the enterprise market and seen what was happening, they'd have shipped a free Solaris x86 that worked with commodity hardware rather than a narrow (expensive, hard to find) subset and Linux wouldn't have gone anywhere.

The name is a blast from the past for anyone who worked in IT in the 1980s.

They sold a line of IBM 3270 terminal emulation software and some IBM PC compatible communication cards so you could work at your new fangled PC while still looking at the corporate software on the IBM mainframe. I thought they died when 3270 comms protocols went TCP/IP but apparently a shell of the company has struggled on for years sitting on a bunch of acquired patents from subsumed competitors.

Their SDLC cards were a total bitch to work with too - especially if you were a gumby like me and had never seen a 3270 terminal or mainframe but the sales guy wanted to ship a bunch of PCs into some government department "ready to wear". For this reason the company name gives me the shudders.

They do. I've been struggling with commercial and free ones since C++ was the hottest environment around and they all basically do the same thing: spit out useless code that is impossible to optimise on a large database. You can't tune it reliably, you end up with indexes out the kazoo all over your database trying to catch all the special cases that the damn OR mapper makes up and they all have truly, ugly, broken interfaces to anything a normal database programmer uses like stored procedures.

I wrote one myself for the last project I worked on and, yep, it sucked just as hard as the commercial offerings for all the same reason.

They are, however, infinitely superior to object databases which truly are the spawn of the devil.