The Petrochemistry of Software

Resourceful as ever mrs. motz pointed us to an interesting presentation a certain Mr. John Day recently held at the "future of networking" conference in Budapest.

John D. Day (born 1947) might not be as well known as Vinton Cerf, Bob Kahn, Larry Roberts or even Jon Postel - really the most important of them all - but like all of them he is a computer scientist, an Internet pioneer and a historian. He has been active in the development of the communication protocols of Internet and its predecessor ARPANET since the 1970ies and has also also been involved in the design of the Open Systems Interconnection Model (aka OSI Reference Modl.

Day is the author of the 2008 book Patterns in Network Architecture: A Return to Fundamentals and the RFC documents RFC 520, RFC 728, RFC 731, and RFC 732.

A critique of the Internet architecture at a historically and technically fundamental level

You will find a pdf of this presentation of Mr. Day's on Future InterNet Design (FIND) at Forum Athena here. That one is for all practical purposes the same as the one he used in Budapest.

Mr. Day is a very classical sort of technology pioneer. Having been there and having co-designed the t-shirt, he knows exactly where the quarrels were and which decisions (in his opinion) went wrong and which problems of today's Internet are connected with which of those decisions. So nowadays he travels the internetworking congress circuit and pokes his fingers into all the wounds, hoping this will result in a real wakeup call.

Everybody knows, we here at tinytalk plc. are buffs for computing and networking history. If you belong to those who know us a bit better, dear reader, you might also know that we are glad about demystification. And though having moved our HQ to Jamaica we also adore any hint at points where we can see that important computing and networking systems still contain technology invented in Europe and not in California or Massachusetts and who the guys were who developed and provided it.

Listen to a short sample of Mr.Day's Budapest talk here. It is an mp3-audio, also provided by Mrs. motz at motz.antville.org.

Moore's law has made us sloppy.

But enough of the preliminaries. Now we'd like to look at just some of Mr. Days concise aphorisms on why software (especially networking code) is still as inmature as it is: "Moore's Law has allowed us to ignore important questions, it has made us sloppy".

After thorough examination and analysis of this stunning proposition the tinytalk tech stuff gave the following variant as a hopefully even more correct alternative: "The phenomenon that Gordon Moore observed (called a law by Carver Mead) in conjunction with the current mode of production and distribution has forced us* to decide for solutions that were mostly not that bad but had no answers for a lot of important questions".

The meaning of these two variants is relevantely different. Whereas the first one implies that sloppy idiots took the wrong bends in the past and did not listen to better informed people (like Mr. Day maybe). The second one points to a set of circumstances that forced non-ideal decisions that got things ahead while paying a price, 'cause, as we used to say, there is no such thing as a free lunch. The same goes for the sometimes mysterious success of programming languages and the strange relapse of software practice into the primitive when microprocessors came along. It is just that sloppiness was forced but not the point.

*Society in general and the software academia and industry in particular

An echo elsewhere

The proposition of "Moore's law driving the computer industry" like fossile fuel driving the economy at large and by its "cheapness" (eq. easy availability) preventing "better solutions" finds a nice echo in Steven Wozniak's hope that software will get better, once the run of Moore's law hits the "atomic barrier". As the Wired Archive tells us:
But his [Wozniak's] hope for the future lies with the end of the curve of Moore's Law. Thinking about when designing personal computers might be fun again, Woz projects himself into a time when advances in chip design will have run up against some hard physical limits. Finally, the box will be defined, and, rather than rushing new hardware out the door every six months, we can all take a breather and ask, "How should the software work with the human being?" To him, the current system is driven solely by new generations of hardware that ship too fast, leaving no time for careful study.

Here's a question from an e-mail to Mr. Wozniak: Woz, I would like to hear your opinions on modern computer directions. Are we where you expected to be by now? What do you see that is exciting for the future?

And here's WOZ's answer: Too vague for me. Too unpredictable too. My latest hopes are for humanistic software but it may take until about 2020, assuming that Moore's law ends for atomic reasons around 2012.

A preliminary observation

Well, somehow the run of Moore's Law (actually an observation that soon became a self fulfilling prophecy) will not end in 2012, as far as we can tell now, but we we will probably see the day in this very lifetime and not only in our next one. However that be, Dennard scaling has ended in 2006 and with it a lot of what was driving the development commonly called Moore's law. Ever more logic gates, multicore and automatic optimization is what's driving progress now.

Post scriptum This is - as so often here on these pages - mere speculation and prejudice - but somehow we cannot get rid of the feeling that our Mr. Day's rather acrid view of the internet architecture stems from a sort of insultedness that although the 7 layer OSI burger, in the creation of which Mr. Day played a large role, had support from the DoD to the complete communication elite of the EC (now EU) never really got its feet moving on the ground.

It reminds us a bit of how Mr. Nelson always trumpeted how wrong the designers of http and html got hypertext. All these guys while being adamant about science have their blind spots and have not studied complex biological systems enough. Still all the fundamentals and Woz give us hope that Moore's law and its ill effects on software will end its run in the foreseeable future. What then?

People get ready there's a train a-coming,
you don't need no baggage just a-get on board.
All you need is faith to hear the diesel humming,
you don't need no ticket no no just thank the lord.

A Rich Neighbor

When Steve Jobs found out about Windows, he went ballistic.

"Get Gates down here immediately", he fumed to Mike Boich, Mac's original evangelist who was in charge of our relationships with third party developers. "He needs to explain this, and it better be good. I want him in this room by tomorrow afternoon, or else!"

"And, to my surprise, I was invited to a meeting in that conference room the next afternoon, where Bill Gates had somehow manifested, alone, surrounded by ten Apple employees. I think Steve wanted me there because I had evidence of Neil asking about the internals, but that never came up, so I was just a fascinated observer as Steve started yelling at Bill, asking him why he violated their agreement."

Read more at folklore.org

Andy Hertzfeld also remembers the incident in a similar but relevantly different way. We think Hertzfelds description is probably the more faithful of the two:

"... He (Jobs) was trying to get them (MSFT) to forget about the OS business, since the applications business would be much bigger total dollars. He said, "It's not that I don't trust you, but my team doesn't trust you. It's kind of like if your brother was beating up on my brother, people wouldn't say it was just your brother against my brother, they would say the Gates are fighting with the Jobs." Bill responded that "No Steve, I think it's more like we both had this rich neighbor named Xerox, and you went in to steal the TV, and found that somebody else had stolen it. So you say, "hey, that's not fair. I wanted to steal the TV".

All enhancements added by TinyTalk plc.

Now, one should never forget that Steve grew up with a single hippie mother and Bill in a very-well-to-do intact WASP household. What this did with them and how Jobs for a long time had troubles with the corporate world and Gates had not, how Jobs always fought against open systems and Gates profited from them, how Jobs had self-educated impeccable style and Gates had not at all is a topic for another day.

But now, after all these years, what will motivate Jobs when Gates is a pensionnaire and philantrope? Selling ever shinier stuff to ever richer kids can't be that sharp. AAPL's profits have risen well above MSFT's. Are the suits really still the guys you want to impress? Come on Steve, you have earned to relax just a bit. There are enough young guys'n'gals to carry the torch, :-)

PS: The differences between LISA/MacOS and DOS/Windows - despite everything that respective fans suspect - are mainly based in the ISAs (instruction set architectures), implementation and memory management of the original 16- and 32bit architectures of Intel and Motorola respectively. And then the overall cost of those architectures and in turn with Gates' office business oriented preference for the Intel and later IBM world, while Wozniak/Apple had chosen Motorola/Mostek with its cleaner architecture and better path. Watch the following video if you're interested in why and how Intel won the microprocessor wars until heat and power consumption problems heaved a poverty driven European architecture, made possible by American tools, heaved ARM into the saddle of the winning horse. Which is yet another story for another time.

There was also a corresponding large difference in SW philosophy of which MSFT's was more modern and less performant than AAPL's (bytecode, virtual machines, portability, high level languages vs. pragmatic handcrafted "small is beautiful" Motorola ISA assembly and manually optimized Apple Pascal Code in the beginning). That difference like with the one with Multiplan/Excel vs. Lotus 1-2-3 first gave the advantage to Apple and the hand-crafters and then, after some years to Microsoft. None of these optimizations are necessary anymore, one might add.

The very same reasons also caused the ancestors of both, the beautiful Xerox Alto, Dolphin and Dorado with their advanced Smalltalk, Cedar, Gypsy and Star interfaces (the latter even had a now becoming fashionable in mobiles 3-line menu icon in 1981) to plummet to the bottom of the waters called market rules like a cannon ball.

Back then, at Bell Labs, a Cherry few know

Cherry: Yes, it is. Very high. And there was this flavor, other people probably talked about this, but it was group dynamics all going on up there. We were all up in the sixth floor. Although I think I worked mostly in my office and visited the sixth floor with questions, [inaudible] the sixth floor. But there was this attitude, there were all these little tools built, and it was the idea of pipes that just kinda, of stringing things together, that was all neat and wonderful. And there was this attitude that he who touched it last owned it. So if you needed PR (?) to do something PR didn’t do, and you went and added it, you now owned PR. And so if some other part of it broke, you owned it.

MSM: You were now PR’s support.

Cherry: You were PR’s support. Yeah, so a lot of the programs floated from person to person because somebody would add a feature to Sort, and they owned Sort. Somebody else would add a feature to Sort and then they owned Sort.

Now that is, how open source was done at Bell Labs in the 70ies. Inside the corp. though, half owned by Western Electric and half by AT&T. Dig the spirit.

... The graphics part of it, I think Tech is still better as far as what EQN does and what Tech does. From a mathematical standpoint I think you’ll find Tech better, but I don’t think Tech stuff is anywhere near as natural to work with. That may be very prejudiced, I don’t have a…

Oral History at Princeton with Lorinda Cherry.

Only in Boston and the Valley: About Venture Capital

From oral history with Don Valentine, VC supreme:

RW: Well, do you - do you think that that as one of the successes of Silicon Valley, the availability of venture funds?

DV: There is no question. It's very difficult. And over the years we've been visited by hundreds of people from every country in the world, almost all of the states in the country. They all want to figure out and clone what causes Silicon Valley to exist and thrive. Many or most of the visitors are interested in the underlying employment creation.

But if you look at venture capital, it only works two places in the world. It doesn't work outside the U.S., and it only works in Boston, or the greater Boston area, and in Silicon Valley. Basically, almost no other major financial center in the world has ever generated companies of consequence and providence that is so large and visibly successful. And you can blame it on the weather. You can give credit to the great universities. You can explain that the venture community here is stronger and more experienced than other parts of the world.

I once tried to explain to the Deputy Prime Minister of Singapore who was here trying to take back the magic, and I said it's a state of mind. You can't take it back. Somehow or other you'd have to move your people here and they would have to have their DNA changed so that when they went back to Singapore, there was a DNA adjustment in the way they thought, the way they were willing to take risks. I mean in a place like Japan, if you start a company and it fails you're disgraced. Some people would commit suicide - if that happened.

So weit wir es beurteilen können, sagt Herr Valentine die Wahrheit. Es ist ein Kreuz aber es ist so. Und das ist noch nicht einmal alles. Und jetzt ist California doch pleite, irgendwie.

Wie sagt General Stillwell in Steven Spielbergs Film "1941"? This is not the state of California, this is the state of insanity.

2009 JVN medal goes to Susan L. Graham

In a continuation of our policy on hinting to pre 2.0 computing professionals, today we have got Susan L. Graham. Mrs. Graham has received the 2009 IEEE John Von Neumann Medal. Past recipients of the Von Neumann medal include computing legends Donald Knuth, Doug Engelbart, and C. Gordon Bell.

The medal was presented on 25 June 2009 at the IEEE Honors Ceremony in Los Angeles. Mrs. Graham receives it for innovations including a sophisticated pattern-matching algorithm and an

elimination-style algorithm for analyzing the flow of values in a program. Susan L. Graham is reachable through graham at cs dot berkeley dot edu.

What Dr.Dobbs had to say:

Known for taking her ideas through the full development cycle from practice to theory then back into practice, the results of Graham's work are widely used in practical compiler systems that convert high-level programming language code into the target language instructions used by a computer. She has promoted the development and distribution of widely used software throughout her career, and her service to the field has been demonstrated through her work on national committees and as founding editor of a leading programming language journal.

What Wikipedia says:

Graham is the Pehong Chen Distinguished Professor in the Computer Science Division of the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley. ... In 1994 she was inducted as a Fellow of the Association for Computing Machinery. She is also a Fellow of the American Association for the Advancement of Science, and the American Academy of Arts and Sciences. She is a member of the National Academy of Engineering.

What her own homepage says:

As a participant in the Berkeley Unix project, she and her students built the Berkeley Pascal system and the widely used program profiling tool gprof. ... She and her students have built several interactive programming environments, yielding a variety of incremental analysis algorithms.