Restoring YC's Xerox Alto: how our boot disk was trashed with random data

You can't seriously expect factual accuracy from Malcolm Gladwell.

Chapter 23 of "Dealers of Lightning" points out that everyone describes the Steve Jobs visit(s) in contradictory ways. "No anecdote from PARC's history is burdened by so much contradictory testimony. PARC engineers disagree about who delivered what portions of the demonstration; on how many demos there were and when they took place; whether Jobs and his people saw an Alto or a Dorado..."

I recommend "Dealers of Lightning" for the best description of Steve Jobs' visits and their impact on him. The short summary is there were two visits to PARC in December 1979. Adele Goldberg tried to demonstrate just the publicly-announced technology (WYSIWYG browser, mouse, drawing program) but was forced by management to show Steve Jobs all the secret stuff (e.g. Smalltalk, overlapping windows). Steve Jobs considered the three mind-bending things he saw to be the user interface, object-oriented programming and networking, but he was "saturated" by the UI and missed the significance of the others. The book goes into detail on exactly what Apple was doing before the visit and how it changed their plans. The book has a whole chapter on this, way more than I can cover here, so please read the book if you're interested in this.

What made it possible to track down the RNG function was that the Computer History Museum has a big archive of Alto code [1]. By searching that site with Google, I found the random number source code. Of course figuring out the code is a bit of a challenge since it's in BCPL or Nova assembly.

Hopefully as part of our Alto restoration we'll be able to add more old Alto disks to the online archive. Maybe the disk labeled "Alan Demo" will have an exciting long-lost Alan Kay demo on it :-)

[1] http://xeroxalto.computerhistory.org

I agree that its a great story. Like all good stories it tries to distill the point from a much more complex and nuanced world. And because the world is more complicated than narrative fiction, there are always other aspects that contribute or can be portrayed in different ways.

The reality of things as I came to understand them were these; Jef Raskin was passionate about user interface design like few people I have ever known. The folks at PARC were excited about the possibilities offered by the advanced state of the art in graphics and input devices and they had some really excellent thinkers and engineers who were implementing really different ways of interacting with computers. Steve Jobs was passionate about design with strong opinions on what was "right" and what was "wrong" about any given product feature.

Now you have to put that into an engineering context. Apple is building a new computer system, one of its, if not THE, lead designers has ideas about how the user will interact with it. The company is pushing back on these ideas because they are "radical and new" and questioning how the engineers know that this will be a better way to build computers than the existing examples in the Apple II and CP/M machines of the day. So this conversation within Apple gets more and more energetic. If you are in an engineering company and are leading any sort of project you will have this experience. You propose something different than what the company expects, and the company pushes back on that proposal in a bunch of different ways. In an effect that is oddly reminiscent of how a band-gap forms in a PN junction, groups line up for or against the concept. And it percolates up the management chain until someone in authority "decides" and shuts down all of the side conversations.

Having had conversations with various people both on the Xerox side and the Apple side who were part of that experience, I concluded that Steve Job's visit to PARC where he saw a realized example of what Jef was trying to push through, convinced Steve that it was better design. I based that conclusion on stories about how the conversation within Apple changed once Steve had seen for himself the possibilities presented by a window based system with a pointing device. Jef related in a talk that he gave about interface design that resistance to the ideas essentially evaporated once Steve was on board.

So is that the "real" story? I don't know. It is the most likely explanation of events that followed all of the statements I heard from people who participated in the design of the Macintosh/Lisa and people who build XDE and Star (the Xerox Development Environment and their office suite). Does it matter? I don't know that it does, the origin story for the Macintosh is what it is.

The shift-xor "random" number generator described actually has an interesting property. The algorithm looks like this:

1. If the value is 0, output the magic number.

2. If the value has only its high bit set, output 0.

3. Otherwise, shift the value to the left

4. if there was an overflow (the high bit was 1 before the shift), xor the value by a magic constant.

For certain values of the magic constant, this has the property of generating every integer in the range exactly once.