I'm skeptical. People really didn't know that they were sharing physical resources with many other users? or if they knew, they didn't care because it didn't matter?
Maybe this was true with the systems you know about. It wasn't my experience using "timesharing" on IBM-and-compatible mainframes. (I'm putting that in quotation marks because one of the standard jokes had to do with whether TSO (Time Sharing Option -- the support for interactive users in MVS) was misnamed.)
It didn't matter. When it "mattered" as in poor performance, we treated it as a bug and fixed it. IBM was...
In particular, I noticed that response time was very good at hours when other people weren't likely to be around, but during first-shift hours, things sometimes slowed to a crawl. I guess it wouldn't happen if the system were well-designed, and someone had paid attention to how much computing power would actually be needed to support the number of people likely to want it.
Anyway, the point is that if I *really* had my own computer, there would be no such effect, unless I were competing for some other shared resource, such as network bandwidth.
I don't think I was confused. These, distribution, installation, and end, are three separate procedures (although Java-flavors are mixing them up alarmingly...
I was going to argue with you on this point, saying "but what about installing new software? don't you need special privileges ...."
I don't remember how that worked when I was using IBM mainframes; I think it was never an issue in the kind of work I was doing.
Fair enough. Interesting. snip snip I think we're sufficiently in agreement that we don't have to...
But yeah, a good multi-user system will let you install at least some stuff without access to system directories. Maybe to some extent you could say "and the more this is possible, the better the system is"?
In this regard, I think things in the Linux world have gone backward to some extent ....:
Once upon a time, to install something new under Unix you would get a tar file (archive format) somewhere, untar it on your system, and follow the installation directions, which were usually set up to allow some sort of usable installation without write access to system directories. It seems that the preferred packaging for software to install on most Linux distributions is something that (usually) requires administrator privileges to install. (RPMs are the Red Hat version. I think there are some other formats, but I'm drawing a blank on their names right now.) This does greatly simplify installation, but IMO, it's a mistake, and it comes from the "single-user system" view of computers.
But whatever .... I think my point is that even people using a really great time-sharing system must know on some level that the actual computing is happening on some machine that is *NOT* ....
No, maybe it's not really so much about not having total control as about not having total responsibility. One of the benefits of using a good time-sharing setup is that all of the tedious work of managing the system (keeping current with upgrades, e.g.) is done by someone else. In the single-user-system model, the user has to be the administrator too, to some extent. This could be a good thing, but in many cases (maybe most), it's not -- more trouble for the user, worse results, etc.
I have a feeling these points have all been made many times before in this newsgroup, though, and we're probably basically on the same side here, so I'll shut up now.
-- B. L. Mbuttingill ObDisclaimer: I don't speak for my employers; they return the favor.