If you find PARC interesting, and especially if you're interested in symbolic computation, I can highly recommend digging as deep as you can possibly stomach into the FGCS:
As a public research initiative, pretty much everything was published when the initiative was completed. PIMs are absolute engineering marvels. The ICOT had command of an army of the absolute best talent in the entire country, and unified them towards a goal of pure exploratory research with no market pressure, with all the excesses of 1980s Japan.
> According to a report from The Wall Street Journal, in 1989 US officials feared that TRON could undercut American dominance in computers, but that in the end PC software and chips based on the TRON technology proved no match for Windows and Intel's processors as a global standard.
“Excess” implies irrationality, but I wonder how could something irrational be a threat to ‘American dominance in computers’?
A great tech book on symbolic computers in general and Lisp machines, is Peter Kogge's 1991, "The Architecture of Symbolic Computers". I believe new efforts by people like Yann LeCun will counter the "LLMs or bust" monoculture along with SOC/ASICs, in-memory compute, neuromorphic chips, dataflow, optical/analog hybrids , etc. that will bring a healthy correction or alternatives to the Von Neumann architecture.
"If you really want to great phenomenal items here is the plan:
- enter a market
- become a monopoly
- use those monopoly profits to fund R&D/building items of incredible quality"
A recent example of that is Apple TV. Apple makes so much money that they can fund the creation of incredibly high quality shows with basically minimal advertising.
Because you have the belief that you can generate a better ROI internally invested. Returning cash to shareholders is usually a negative sign for the company as it means they don't think they believe other companies will be better with the investment than themselves.
It's only recently in the share-buyback age that this investment is rarer.
The classic example is Amazon which was technically profitable for a while, but did not return shareholder cash for many many years, choosing to invest instead.
Unless mandated, why would the people controlling a corporation (and its budget) do that? While the corp has money in the bank, it's kinda their money, in the sense that they decide what to spend it on. If distributed back to the shareholders, the money evaporates from their perspective, so there's not much incentive to do it unless it's required, or unless they will benefit by it (e.g., they have a lot of stock themselves and would like the dividend).
Corporations that have gone public are essentially lost causes. CEOs are incentivized to maximize shareholder profits at all costs, and it's much easier and cheaper to enshittify than to research and develop.
I suppose it's possible for privately owned corporations to be awesome. If the guy in charge cares, awesome things will happen. Valve is the only concrete example that comes to mind.
Increasingly, I think that an agent (and I) would work much better in a malleable, notebook-like, inspectable program, than it would with its current file-based “edit and re-run” primitives.
“Marimo pair” (built into their notebook-like primitive) is an attempt at this. And they have program introspection tools built in.
I also think that Glamorous Toolkit (https://gtoolkit.com/)
might be a similar live environment, but I haven’t investigated it too much other than reading about it.
Is anyone else familiar with “modern” attempts at this?
Love the "Stop Writing Dead Programs" talk - and of course "Inventing on Principle" by Bret Victor is required viewing for anyone that hasn't seen it yet.
https://youtu.be/PUv66718DII
"Residential programming" isn't it similar to interepreter!! Difference I can think of is it can completely rewrite something instead of updating or extending it. ex. rewrite the existing function instead of re-defining it.
In Common Lisp, the compiled code and data are both saved as a memory snapshot, kind of like an OS image (think an ISO you can boot into).
This means you can hot-swap the code part, while keeping the data and run-time configuration, on a running system.
A REPL is part of it, but it's not the whole picture.
It turned out in practice that a Lisp Machine was basically a Lisp interpreter implemented in hardware. A better and much more pragmatic solution was to compile Lisp to efficient machine code and then run it on any traditional CPU. Which is what we do today.
If you find PARC interesting, and especially if you're interested in symbolic computation, I can highly recommend digging as deep as you can possibly stomach into the FGCS:
https://www.airc.aist.go.jp/aitec-icot/ICOT/HomePage.html
As a public research initiative, pretty much everything was published when the initiative was completed. PIMs are absolute engineering marvels. The ICOT had command of an army of the absolute best talent in the entire country, and unified them towards a goal of pure exploratory research with no market pressure, with all the excesses of 1980s Japan.
FGCS: the Fifth-Generation Computing System
I was really excited about this initiative at the time, just starting my computer science undergrad degree.
Hardware that ran Prolog as close to bare metal as possible.
Thanks for the reminder. 40 years ago.
More Japanese excesses from the '80s: the Tron computing project.
https://en.wikipedia.org/wiki/TRON_project
> According to a report from The Wall Street Journal, in 1989 US officials feared that TRON could undercut American dominance in computers, but that in the end PC software and chips based on the TRON technology proved no match for Windows and Intel's processors as a global standard.
“Excess” implies irrationality, but I wonder how could something irrational be a threat to ‘American dominance in computers’?
A great tech book on symbolic computers in general and Lisp machines, is Peter Kogge's 1991, "The Architecture of Symbolic Computers". I believe new efforts by people like Yann LeCun will counter the "LLMs or bust" monoculture along with SOC/ASICs, in-memory compute, neuromorphic chips, dataflow, optical/analog hybrids , etc. that will bring a healthy correction or alternatives to the Von Neumann architecture.
Rory Sutherland [0] has a great quote:
"If you really want to great phenomenal items here is the plan:
- enter a market
- become a monopoly
- use those monopoly profits to fund R&D/building items of incredible quality"
A recent example of that is Apple TV. Apple makes so much money that they can fund the creation of incredibly high quality shows with basically minimal advertising.
0 - https://www.tiktok.com/@rorysutherlandclips/video/7314765561...
> Apple makes so much money that they can fund the creation of incredibly high quality shows with basically minimal advertising.
But Apple has not created any high quality shows.
Simply throwing money at something does not automatically make it good.
Crazy take. Severance, Silo, Pluribus just off the top of my head. You don’t think they are high quality shows?
First two seasons of For All Mankind deserve to be on your list.
No they are not. They are bland and unoriginal.
Which service is putting out all the high quality stuff then, and what is it? Any examples?
Foundation?
> use those monopoly profits to fund R&D/building items of incredible quality
But why would a corporation do that when it could simply distribute those profits to shareholders?
Because you have the belief that you can generate a better ROI internally invested. Returning cash to shareholders is usually a negative sign for the company as it means they don't think they believe other companies will be better with the investment than themselves.
It's only recently in the share-buyback age that this investment is rarer.
The classic example is Amazon which was technically profitable for a while, but did not return shareholder cash for many many years, choosing to invest instead.
Unless mandated, why would the people controlling a corporation (and its budget) do that? While the corp has money in the bank, it's kinda their money, in the sense that they decide what to spend it on. If distributed back to the shareholders, the money evaporates from their perspective, so there's not much incentive to do it unless it's required, or unless they will benefit by it (e.g., they have a lot of stock themselves and would like the dividend).
Corporations that have gone public are essentially lost causes. CEOs are incentivized to maximize shareholder profits at all costs, and it's much easier and cheaper to enshittify than to research and develop.
I suppose it's possible for privately owned corporations to be awesome. If the guy in charge cares, awesome things will happen. Valve is the only concrete example that comes to mind.
Shareholders will happily vote to hand all the money over to the CEO if he's charismatic enough: https://www.latimes.com/business/story/2026-05-27/how-elon-m...
In other words:
- enter a market
- become a monopoly
- use those monopoly profits to maintain your monopoly
A playbook that's very much alive in all big tech companies, from Apple to Oracle.
I wasn’t familiar with the term “Residential programming,” but it reminds me of the talk “Stop Writing Dead Programs” (https://jackrusher.com/strange-loop-2022/)
Increasingly, I think that an agent (and I) would work much better in a malleable, notebook-like, inspectable program, than it would with its current file-based “edit and re-run” primitives.
“Marimo pair” (built into their notebook-like primitive) is an attempt at this. And they have program introspection tools built in.
I also think that Glamorous Toolkit (https://gtoolkit.com/) might be a similar live environment, but I haven’t investigated it too much other than reading about it.
Is anyone else familiar with “modern” attempts at this?
> Is anyone else familiar with “modern” attempts at this?
Glamorous Toolkit is the main one I have found, yah.
The TruffleSqueak demos are pretty cool: https://github.com/hpi-swa/trufflesqueak
Enso Analytics also runs on GraalVM: https://ensoanalytics.com/
Love the "Stop Writing Dead Programs" talk - and of course "Inventing on Principle" by Bret Victor is required viewing for anyone that hasn't seen it yet. https://youtu.be/PUv66718DII
You can have the residential programming feel with SBCL; it's a pleasure.
I had a Xerox 1108 Lisp Machine, ran InterLisp-D on it for about two years, then slowed it down by installing Common Lisp on it.
Wonderful for Larry et.,al. to keep it going as open source.
> Is there an Interlisp echo in today’s development culture?
Emacs not mentioned. It better have been.
"Residential programming" isn't it similar to interepreter!! Difference I can think of is it can completely rewrite something instead of updating or extending it. ex. rewrite the existing function instead of re-defining it.
The Interlisp manual defines residential programming like this:
«The "residential" style, where you stay inside the environment throughout the development, is essential for these tools to operate.»
https://interlisp.org/documentation/IRM.pdf#page=17
Does this need anything beyond interactive REPL? If so: what?
An image-based execution model.
In Common Lisp, the compiled code and data are both saved as a memory snapshot, kind of like an OS image (think an ISO you can boot into). This means you can hot-swap the code part, while keeping the data and run-time configuration, on a running system.
A REPL is part of it, but it's not the whole picture.
Getting "too many requests" at the moment.
It turned out in practice that a Lisp Machine was basically a Lisp interpreter implemented in hardware. A better and much more pragmatic solution was to compile Lisp to efficient machine code and then run it on any traditional CPU. Which is what we do today.
[flagged]
[dead]