This is a semi-satirical / joking-not-joking post (spun out of a private mastodon post) that I will be taking absolutely no questions or comments about. If it does not amuse you please do not tell me in gruesome well-actually detail why it is a bad idea.
My proposal: Computers should have stopped in 1993.
One might argue that I was an impressionable teenager in 1993 and so probably this is "just nostalgia speaking" but I think it is not true: the technologies I had access to at the time were not, mainly, those I will be discussing here. Instead, I claim that as an adult with more fluency in computers and computing history, I can make the recommendation here on the basis of that broader and more-objective view.
1. CPUs and Systems
The
MIPS R4000 existed in 1993 and at 1.2 million transistors, this is about as complex as chips should ever have got. It's got an MMU and FPU, is RISC, is 64 bit, in-order scalar superpipelined. It is predictable and simple and just right. There was a consortium (
ACE) that shipped a spec (
ARC) for open systems built on MIPS and several vendors were using it as their vision of the future. They should have been right!
(If you needed a portable computer you could have the R4000-based
IBM WorkPad Z50 or, if you are a sicko, a
Newton MessagePad which was not R4000 but we can allow 1993's pleasantly small ARM6 chips as well, or 1992's charming
SH-2. Also you can even have some videogames: the PlayStation was R3000-based and the Nintendo 64 R4000-based, and the Sega Saturn was SH-2. If you really really hate MIPS, ARM
and SuperH you can throw in the
Alpha 21064 -- the first Alpha, when it was still in-order and 1.6 million transistors -- and I will allow that it doesn't break the mold too much. The Pentium was also in-order but at 3 million I think it's too big. The 68040 at 1.2 million is fine, but of course still just 32 bit like the ARM6 and SH-2. You really want an R4000 or Alpha.)
2. Distributed Operating Systems
In 1993 we had
OSF/1 with
DCE. This was not the best OS one can imagine, but it had qualities and capabilities that have in retrospect not been meaningfully eclipsed in the years since. A DCE installation had a real distributed filesystem, RPC, locking and time services, single-sign-on (Kerberos) and directory service. Stuff you still can't get reliably in our modern cloud/k8s nightmare. One might argue that
Windows NT also got there, but .. sure, fine, you can have that too! Windows NT also came out in 1993, running on R4000. And
Plan 9 was released in 1992. So we really were firmly in the "stuff better than we were ever going to get" future.
3. Languages
In 1993 we had
Modula 3,
Sather and
Dylan; but we had not yet been subjected to Java, PHP or JavaScript. The former are all safe, native-compiled and expressive. The latter are .. not. We should have stopped here, or taken a different path at least, but the web came along.
4. Databases and 4GLs
There was also a suite of higher-level languages -- those classified as
"4th Generation Languages" (4GLs) as far back as the 60s -- still trucking along making it easy to write database-integrated applications. Indeed what was marketed back then as "a database" was typically a full-featured application programming environment, including UI tools, transactional DB-integrated high level language, report generation system, compiler and re-distributable runtime. Products like the
xBases (dBase, FoxPro, Clipper),
Paradox,
PowerBuilder,
4th Dimension and
Access/VB were the standard for zillions of independent developers writing small custom in-house line-of-business / industry-specific applications. This was a much simpler and tidier version of what turned into web development (including "intranet" applications). Again, the web killed most of this with its WAN support and universal client, but at enormous cost and complexity.
5. The Web Was Still Niche
1994 was the year of the first WWW conference, the founding of the W3C, the year Netscape was released .. it was the year "everyone got the web". I believe this was a mistake, and we all would have been better off doing something else instead. So 1993 it is.
Gopher existed then too, along with IRC, FTP, NNTP and
WAIS; things were fine.
The web did bring an enormous flourishing of creativity, expression, universal access and connectivity. But it also brought with it a model of
computing imported wholesale from the magazine industry: software as flashy and visual "content" supported by ads, rather than functionality provided for pay. I argue this has been a net negative to society, despite the subsidy to non-paying users that ad-supported software provides. Nobody was passing laws trying to protect teenagers from the psychological effects of 4GL applications.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
solarbird