Tesla is the Lisp Machine of cars?

by philg

Here’s a meme that programmers between age 60 and dead will like: Tesla is the Lisp Machine of cars.

A recent message exchange:

  • friend1: Prediction: Tesla is done. They already closed the Hingham [Massachusetts] store.
  • friend2: Everybody who wanted one bought one

This reminds me of the MIT Lisp Machine, a $100,000 personal computer to which only a handful of programmers could gain access in the 1970s. The computer itself was the size of a refrigerator and lived in a machine room. The programmer worked on a big-by-1970s-standards bitmap display in his or her office, with the display connected via a coax cable (a Tom Knight design). The keyboard was vastly better than today’s pathetic $29.95 examples, though it was rumored to cost $1,000 per sample. The mechanical mouse wasn’t so great, but Knight eventually designed an optical mouse along the lines of today’s devices.

Symbolics commercialized the machine and it sold rapidly for a few years as “Everybody who wanted one bought one”. The company went public. Confident predictions were made. Then sales fell off a cliff after the market of people who had already wanted a Lisp Machine was saturated. Computer buyers who hadn’t been desperate for Lisp Machine would, if reached by Symbolics marketing materials, decide that cheaper personal computers, e.g., “Unix workstations,” from larger and/or more established companies could serve the same function.

Readers: If you’re over 60 and passionate about writing software, what do you think of the meme “Tesla is the Lisp Machine of cars”? There were tens of thousands of (rich, smug) people who wanted a fancy electric car, but couldn’t buy one. They rushed to buy Teslas, which made it seem that Tesla was awesome at selling and taking customers away from legacy car-makers, but actually Tesla was just fulfilling orders from people who had already wanted a fancy electric car. Now they will face the challenge of actually selling and it is unclear if they will be good at it.

[Not related, but from the same friends:

[son] wanted school off tomorrow for Good Friday. I said only if you accept the Lord Jesus as your savior and no longer eat meat on Friday.

]

Related:

 

15 thoughts on “Tesla is the Lisp Machine of cars?”

  1. Did symbolics ever turn a profit? If so, the analogy breaks down because Tesla has only ever been a tax payer subsidized cash furnace.

  2. Tesla is failing right now because they can’t produce Tesla 3’s, not because they can’t sell them, so that analogy doesn’t work.

    Tesla may well eventually fail (or be taken over) because of the experience curve and economies of scale, which do have something to do with Symbolics’ failure.

  3. Bobby: I found a ComputerWorld note from June 10, 1985 that says Symbolics had $18 million in quarterly revenue and $1.2 million in profit, compared to $10.4 million and $132,000 a year earlier. So it does look like profitability is one area where the analogy breaks down!

  4. Stavros: Are you sure that they are selling so many? The Tesla site right now says that if I put down $1,000 today I will get a Tesla 3 in 12-18 months. They previously said that they had 400,000 orders, right? If those are real orders they need to ramp up production to Toyota Camry levels (387,000 in 2017, according to http://www.nydailynews.com/autos/street-smarts/best-selling-family-cars-america-list-article-1.3823037 ) to deliver a car in 12-18 months to a new buyer. Yet Tesla says that their production GOAL for the Model 3 is 2,500 per week (125,000 per year, assuming a couple of weeks of shutdown time). https://www.bloomberg.com/graphics/2018-tesla-tracker/ says that they are currently building at a rate of 1,200 per week. To meet the promise on their current web site, assuming that they do actually ramp up to 2,500 per week (their goal), the majority of the orders they took for the Model 3 have to evaporate. Otherwise the delivery time for a car ordered today would be closer to 4 years.

  5. My Dad used to work on ‘computers’ that took punch cards
    And I’ve worked on computers that used those old floppy disks in the 1980s and now on the latest ones

    I think the analogy is a very interesting one

    It’s certainly one of the smartest ones I’ve seen. A lot of people are mistaking

    ‘demand from people who really want a Tesla’

    for

    ‘that means everyone in the world wants a Tesla’

    And Tesla is Either

    – managing to play supply and demand games perfectly

    OR

    – is incredibly bad at producing enough Teslas to meet demand

  6. And regarding this:

    Tesla is failing right now because they can’t produce Tesla 3’s, not because they can’t sell them, so that analogy doesn’t work.

    That’s a bit strange, isn’t it

    How do we know which is true?

    Option 1: Tesla has incredible demand and simply can’t scale up production fast enough

    OR

    Option 2: Tesla somehow manages to create only enough supply to keep a contast illusion of ‘great demand’

    The example PhilG has stated is a very good one. Given their production rate AND supposed half a million preorders (requiring just a deposit, by the way, so not actual sales) it should be 2 to 4 year delays before a new buyer could get a car

  7. Every product depends on constant evolution, once the initial market is saturated. Like so many VR goggle brands & 3D TV brands, there was only 1 Lisp Machine. There have been 5 Teslas, each improving upon the last one. They won’t saturate the market, but they might run out of money. Most everyone expects Tesla to be bought out by a Chinese conglomerate.

  8. Why doesn’t Tesla just outsource the production of the Model 3 to a company that is an expert in mass automotive production? While on a business trip to South Korea, I had a chance to get a tour of a Kia manufacturing plant, one car every 72 seconds off the assembly line and capacity of 600,000 per year, just at this one location. Multi model assembly line, where different models can be produced on the same line, absolutely amazing automation (robots everywhere), quality control, parts tracking and production tracking.

    Why does Tesla not just outsource the production of the Model 3 to Kia in South Korea and concentrate on the design and high end model production?

    Kia could probably set this up very fast and start turning out the Model 3s like bread.

  11. Hi Philip,
    Here’s my model of the demise of Symbolics, for what it’s worth (I worked for a company that was a Symbolics customer).
    Sun came out with workstations that cost around $10,000 each, significantly less than Symbolics. There were Lisp implementations that ran on the Sun, maybe not as well as on Symbolics, but they were useable, so many, if not most, Lisp-based projects were ported to Sun. Also a decent color monitor was available for Sun. Symbolics was late with a color monitor, it was expensive and it wasn’t really that well integrated (if I recall).

    On top of that, Gorbachev took power in the Soviet Union. Many (most?) of the projects throwing money at “AI” at the time were defense-funded, but budgets for that sort of thing were cut once the Soviet Union was perceived as less than a threat.

    So projects using Lisp declined in the 90s and those that persisted moved to cheaper Sun workstations.

    Or something.

  12. Some similarity in that both vendors markets were salted with government funding one way or another.
    However, mostly different – Symbolics did not have a plan to take a lisp machine down to hand held and $500 or be included in every product known to man, like Tesla does with the their roadster->S->X->3->? transition and battery manufacturing. Same with the other 80’s specialized computer vendors SGI, Sun, etc. It is why x86 and arm are the surviving winners from that era.

  13. > both vendors markets were salted with government funding one way or another

    I not infrequently wonder if we’d have space mining colonies by now if all the PhDs weren’t so engrossed in suckling on Uncle Sam’s sweat, creamy teat. The cold war was a disaster for technological progress. The cold war never really ended. They eventually rolled it into the GWOT and China scenarios, and they’re trying to spin up Russia as a threat again. Our trillion dollar defense budget still has legions of hyper smart nerds busy working on useless weapons and sensors. It’s much easier than making an honest living.

  14. I don’t know about the LISP stuff but I do know fundamental engineering and Musk is ignoring it at his peril. The Model 3 is too complex for that price point. He is not shrinking the number of parts needed for the Model 3 car versus the much more expensive Model S and Model X. The battery is just too complex. The battery assembly contains over 6000 parts. No amount of robot fancy footwork will fix this complexity problem thus keeping costs high. So until he stops forcing the issue and redesigns the battery to something simpler he will continue to spend more to build each car than he can sell the cars for. The Wall Street bond people are telling him this right now. His only out is to slow production of the Model 3 and redesign/reprice the car so he can stay in business and make a profit. Unless he does significant changes voluntarily and fast he is doomed. At Space X he has driven test and fail and redesign with great success. Will he adapt this method at Tesla? Who knows?

  15. To this day, the Symbolics was the best development environment I’ve ever used. Of course it was lisp!

Comments are closed.