Rendered at 20:46:03 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
ux266478 2 days ago [-]
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.
watersb 16 hours ago [-]
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.
eggy 19 hours ago [-]
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.
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?
alexpotato 20 hours ago [-]
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.
> 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.
solumunus 7 hours ago [-]
Crazy take. Severance, Silo, Pluribus just off the top of my head. You don’t think they are high quality shows?
dosisking 3 hours ago [-]
No they are not. They are bland and unoriginal.
matheusmoreira 19 hours ago [-]
> 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?
zipy124 10 hours ago [-]
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).
matheusmoreira 17 hours ago [-]
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.
sph 12 hours ago [-]
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.
anonzzzies 14 hours ago [-]
You can have the residential programming feel with SBCL; it's a pleasure.
mark_l_watson 18 hours ago [-]
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.
virajk_31 13 hours ago [-]
"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.
pamoroso 13 hours ago [-]
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.»
Does this need anything beyond interactive REPL? If so: what?
kode-targz 7 hours ago [-]
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.
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.
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.
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?
"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...
But Apple has not created any high quality shows.
Simply throwing money at something does not automatically make it good.
But why would a corporation do that when it could simply distribute those profits to shareholders?
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.
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.
- 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.
Wonderful for Larry et.,al. to keep it going as open source.
«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
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.