The Unix Heritage Society mailing list
 help / color / mirror / Atom feed
From: steve jenkin <>
To: TUHS <>
Cc: Marc Donner <>
Subject: [TUHS] Re: Has this been discussed on-list? How Unix changed Software.
Date: Wed, 7 Sep 2022 15:15:45 +1000	[thread overview]
Message-ID: <> (raw)
In-Reply-To: <>

> On 7 Sep 2022, at 02:09, Marc Donner <> wrote:
> Having spent many formative years at IBM Research throwing (metaphorical) bombs at mainframe systems and infiltrating the place with BSD, I have thought about this question on and off.
> I would like to augment your comments with a couple of extra observations:
> UNIX was built for a particular set of users - document writers and programmers
> Before UNIX the systems were industrial in design and scope.  Sort of like MVS was a locomotive -  designed to be used for hauling heavy freight (acres of data entry clerks answering 800 numbers and entering transactions).  UNIX was more like cars and trucks - designed for use by knowledge workers.
> When I was a grad student I hung out with some remarkable programmers.  We all observed that learning to program was impossible without a body of code to read, study, and learn from.  The best places to learn programming in the 70s and 80s were places like MIT, Berkeley, Bell Labs, and IBM Research ... places with an established culture of sharing code internally and large repositories of code to read.
> By the mid-1980s the Microsoft folks established the notion that software was economically valuable.  People stopped giving away source code (IBM's change in strategy was called OCO - "Object Code Only") and it totally shocked the software developer community by destroying the jobs for programmers at user sites.  Combine that with the mid-1980s recession and the first layoffs that programmers had ever seen and we saw the first horrified realization that the social contract between programmers and employers did not actually exist.
> We, the programmer community, woke up and committed ourselves as much as ever we could to non-proprietary languages and tools, putting our shoulders to the OSS movement and hence to UNIX and the layer of tools built on top of it.
> Of course it helped to have some brilliant engineers like Ken, Dennis, Doug, Rob, Michael, Stu (and and and) and brilliant writers like Brian so that the thing (UNIX) had intellectual integrity and scope.
> It took UNIX twenty to thirty years, but the economic logic of our approach put an end to efforts to totally dominate the tech world.
> =====


Good observations. Thank you.
I’ve never heard anyone mention that “reading large codebases” was the best way to learn programming. Absolutely my experience as well.
If Professional Programmers aren’t doing “Programming in the Large” to provide critical services for others, then what work are they doing?

In its first 10 years (1974-84), the future of Unix was uncertain. The formation of SUN in 1982 and other Unix-only vendors made Unix a commercial alternative, complete with support and a help number.

At UNSW, there was a significant political battle over Unix. The manager of CSU (central Computing Services Unit) resigned over Unix. His 35 staff later supported Unix across the Uni. 
If he’d won the battle, it’s very likely all Unix at UNSW would’ve been expunged, stopping the networking work with Sydney University, shutting down the Unix kernel course & dramatically slowing the spread of Unix in Australia.
Robert Elz at Melbourne Uni was later an important contributor to IP protocols and DNS.

In the 1984 BSTJ issue on Unix, there’s no mention of SUN (1982) & SUNOS, but they do note “100,000 licenses” had been shipped, up from the 300 internal & ~600 total licenses mentioned in the 1978 BSTJ.
While still not “cannot fail” status, Unix’s future was becoming more certain.

Today, there are 2-4 billion active smartphone and tablets - almost all of which are Unix variants - Android and iOS. [ I’m sure other ‘platforms’ exist, but haven’t followed the market closely ]
There’s an estimated 200M-250M “Personal Computers” in active use - 10% of all active devices. Even if all run Microsoft and not Chromebooks, MS-Windows is now a minor player.
I’ve no idea how big the fleet of servers powering “The Cloud” in Datacentres is, but inferring from power consumption, it’s measured in millions. 
Only MS-Azure provide Windows instances and then it runs on top of a hypervisor, not bare metal. Is MS Hyper-V derived from a Unix variant? If not, is certainly influenced by VMware & Xen which were.

To a first approximation, 90%+ of ‘computers' now run a Unix variant. [ disregarding the larger fleet of embedded devices, especially in cars ]
As you say, UNIX & its variants broke the monopoly / lock-in of software by hardware vendors.

The timing of Unix and it displacing hardware enforced “Software Silos" wasn’t accidental. [ A notable beneficiary of breaking Silos is Oracle - their early promise was database “portability”. ]

It falls directly out of “Moore’s Law” and “Bell's Law of Computer Classes”.

The PDP-11 ‘regressed’ to 16-bits compared to the IBM 360’s 32-bits:
	Bell’s Law in action - a new, much cheaper, lower performance “class” appearing each decade.

In 1977, UNSW acquired a PDP-11/70 for teaching that was 1/10th the price of the IBM 360/50 that’d been purchased in 1966. [11/70 was in service in April 1978 - hardware delays]
This 11/70 provided at least the same raw performance as the 360/50, but had ~50 2400baud terminals attached, not cards + printer.
It was much more effective for learning/ teaching and provided much higher “useful throughput” than the IBM batch system. Certainly with VDU’s, much less paper was wasted :)

DEC and others first leveraged the cheaper, faster silicon transistors to build bipolar discrete-part machines:
	e.g. the PDP-7 co-opted for “Space Travel” by Ken in 1969.
DTL became TTL, digital IC’s grew larger, cheaper, faster - with “Mini computer” manufacturers rolling out new models at ever better price-points, more rapidly than ’traditional’ mainframe vendors.

Minicomputers adopted “chipsets” to implement the CPU, leading in a few years to single-chip “Microprocessors”, often with co-processors for expensive operations, like floating point.

The invention of RISC led to a whole new class of mini-computers with single-chip CPU's and a new class of system:
	the Graphical Workstation [ SUN & SGI ] - not quite Bell’s lowest performance class, one with significant new non-CPU capabilities.

Without UNIX, there couldn’t have been a RISC revolution, because there’d have been no quality software for vendors to pick up: kernel, tools, Graphical UI and 3rd party Software on these platforms.

The “Dot Boom” that ended in 2000 was only possible because of high-performance UNIX servers for web, storage & database.
e-Bay started with Oracle on SUN servers. A solid, dependable system design.

Google didn’t invent Internet Search, but they did come up with the Internet-scale Data Centre, creating highly available systems using low-cost, “less” reliabile commodity hardware.
Is this a new Bell’s Law Class?
	 it’s more a system architecture and operational arrangement, implemented almost alone in software.

Amazon leveraged their expertise and design of Internet-scale DataCentres into a massive “Cloud” business - not bundled into its own products, but ‘rentable’ by customers by the hour.

Netflix, when it changed from mailing DVD’s to streaming, based its business on renting Amazon servers, storage & bandwidth.
We now have cheaper again computing services available, with zero Capital outlay and scalable to unprecedented sizes.
It follows Bells’ Law, while extending it.

In 2007 when Apple redefined the Smartphone - using ARM (Acorn RISC Machine) and a variant of Unix - they created a new class of computing device.

The device was designed to “Just Work” - near zero admin, self-configuring and a highly reliable O/S, UI & Apps.
Critically, Apple never tried to maximise the “utilisation” of the CPU & its resources - they put in fast CPU’s & aggressively managed power consumption to extend battery life.

The Mainframe era economic model was inverted with the desktop computer - minimise wasted User time, not Computer time.
The Smartphone took this “people first” approach to a new level.

For me, Apple’s most important invention - on top of “Just Works” platform - was the App Store.
It builds on “The Cloud” and Internet services, providing an almost direct Software Vendor to Client channel, using a secure & verified distribution system with embedded payments.

Modern smartphone/ tablet system design, based around “Sandboxes” and a stringent control layer, seems to contain “malevolent” Apps well enough (no security is perfect, but “Good Enough” seems attainable).
Without the App Store and Sandboxed Apps, we couldn’t have 2B-4B smartphones.

We know from the MS-Windows PC & Server ecosystem [ and PHP/ Wordpress ] that "Bad Actors” will organise and actively exploit system vulnerabilities, making large fleets of exploitable devices unusable either because resources are co-opted and the device is unresponsive, or it’s compromised and can’t be trusted.

Ironically, Moore’s Law couldn’t have proceeded as long and as quickly as it has since 1965 without the availability of Software to turn raw Silicon + Watts into functional, useful systems.
Intel now owes a lot more business to Unix and its variants than to MS-Windows.

It’s not unreasonable IMHO to say that Unix and its variants “Changed the World” and are now are the most prevalent O/S on the planet.


Sorry for the long piece - I know that TUHS is not the forum for these observations not confined to Early Unix.

I’d have moved this to COFF, but I’ve not been able to get onto that list so far.


Steve Jenkin, IT Systems and Design 
0412 786 915 (+61 412 786 915)
PO Box 38, Kippax ACT 2615, AUSTRALIA

  parent reply	other threads:[~2022-09-07  5:16 UTC|newest]

Thread overview: 31+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-05 23:48 [TUHS] " steve jenkin
2022-09-06 16:09 ` [TUHS] " Marc Donner
2022-09-07  4:00   ` steve jenkin
2022-09-07 14:58     ` John Cowan
2022-09-07 17:13     ` Paul Winalski
2022-09-08 14:12       ` Paul Winalski
2022-09-07  5:15   ` steve jenkin [this message]
2022-09-07 13:20     ` Dan Cross
2022-09-07 13:52       ` Steve Nickolas
2022-09-07 12:53 ` [TUHS] STDIN/OUT vs APIs [was: How Unix changed Software] Brian Zick
2022-09-07 13:19   ` [TUHS] " John Cowan
2022-09-07 15:39     ` Joe
2022-09-07 15:43       ` John Cowan
2022-09-07 16:01         ` Charles H Sauer (he/him)
2022-09-06 15:07 [TUHS] Re: Has this been discussed on-list? How Unix changed Software Douglas McIlroy
2022-09-06 17:13 ` Larry McVoy
2022-09-07  1:40 ` steve jenkin
2022-09-07  2:33   ` segaloco via TUHS
2022-09-07  4:08     ` Steve Jenkin
2022-09-07 13:08   ` Steffen Nurpmeso
2022-09-07 14:56   ` Larry McVoy
2022-09-07 21:27     ` Steve Jenkin
2022-09-07 22:36       ` Larry McVoy
2022-09-08 14:42       ` Paul Winalski
2022-09-08 15:02         ` Larry McVoy
2022-09-08 15:04         ` ron minnich
2022-09-08 15:52           ` Warner Losh
2022-09-08 16:47             ` Paul Winalski
2022-09-08 16:50             ` segaloco via TUHS
2022-09-08 17:58               ` ron minnich
2022-09-06 19:04 Douglas McIlroy

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \ \ \ \ \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).