daly

"Oil" isn't "Oil".

There are many grades of oil. The U.S. generally produces light sweet crude. The Russian imported oil is heavy crude.

Heavy crude is used for asphalt and diesel. Light sweet is used for other oil products.

Refineries can generally only process a specific type of oil.

So if you want roads and trucking you need heavy crude and a specific refinery process.

While you're at it, increase the information density.

Remove or modify the redundant line about flag/hide/fav... etc. Put it on a right click, put it as a mouse-over or as a button. Don't use half the screen.

It is better design to eliminate redundant non-content.

If you didn't know what McIlroy's program did would you be able to figure out what it was supposed to do?

I'd like to see the TeX program written in a shell script.

As a 50 year programmer I disagree.

Management is a skill. In my experience, MOST programmers don't have the required skill, nor the proper focus, once they "retire into management". MOST of those spend a fair bit of time trying to prove "they've still got it".

I greatly admire Elon but I'm pretty sure he has no idea how to write self-referential, self-modifying code at the heart of his new robot effort. Can he write the code to handle under-actuated motion? How good are his ROS2 skills? Does he write Verilog? Does he know he can replace pattern-recognizers with optics?

I think Elon is an amazing manager but I'd be surprised to find that he wrote some Steele-level or Trager-level great software.

I have Digital hardware books from the 1970s. The "programmer's book" starts by explaining circuits.

These are priceless to me. I loved the PDP-11, especially the PDP11-03 which was a great machine.

Open Emacs. Clip the URL from the browser, yank it into Emacs. Close the tab.

Later, clip the URL from Emacs and paste it into the browser.

I worked on computer-driven IBM Selectric typewriters for years so I was very familiar with their typing speeds.

On a visit to a law office I walked past the secretaries office. She was typing MUCH faster than anything I ever heard. I stopped and asked her about her typing speed. She replied WITHOUT STOPPING ... 225 words per minute. I max out at about 40 words per minute and I've typed every day almost all of my adult life.

Easy fix. hit F1 :-)

Humor, for those too young to know about the "boss" key in games.

I would rarely consider betting against Elon Musk.

However, I've worked with robots starting a research group at Joe Engelberger's Unimation in 1976 (https://www.google.com/search?channel=fs&client=ubuntu&q=joe...).

One of my recent efforts at CMU involved human/robot collaboration to change a car tire. I am currently doing research on self-reproducing systems, essentially robots building robots.

If Musk thinks he underestimated how hard self-driving was then I think he is seriously underestimating general purpose robotics.

That said, I really hope he succeeds. I want to be the first one to shake the robots hand.

Have you looked at Axiom?

https://en.wikipedia.org/wiki/Axiom_(computer_algebra_system...

I have a copy of Freitas's book.

My initial work was done at IBM Research in the early 80s. One was the BOXER project which looked at creating robot assembly plans from 3D solid models. The other was the CLADES project which looked at self-replicating systems. I also did research at CMU on a human/robot system, the TIRES project, to change car tires. A lot of these ideas transfer to self-reproducing systems.

Recent developments, such as low-cost robots, 3D printing, and FPGA/SoC systems make it possible to re-consider ideas that were beyond-the-possible in the 80s.

For example, instead of signal wires to each joint it is now possible to put bluetooth and a processor at each joint. Power can be routed through the structure without free-standing wires.

Or, for example, 3D end-point location can be directly measured using a laser and linear diffraction gratings mounted on the end-effector.

A lot of these high-tech items are hard to replicate. At the moment I'm pondering a method of constructing a servo. Low-tech, low-power, printable conductive plastic with conductive materials might make a servo. Using an harmonic drive would make it possible to generate sufficient torque from such a low-power inefficient motor.

There is also work on building a ROS/Gazebo model that can drive physical servos so part of the system is simulated and parts of it are actual hardware.

Use latex.

I've re-created a 1.2 million line program as a literate program. https://en.wikipedia.org/wiki/Axiom_(computer_algebra_system...

For the record, I made (and make) no claims about "non-Lisp developers not being smart enough".

My claim is that Lisp is perfect for thinking about a new idea or a new approach.

For example, I implemented a program that merged Expert Systems and Knowledge Representation into a single system (KROPS) that allowed a domain expert to express their knowledge as rules or as facts. Anything the system learned by either method could be expressed in either representation. Thus, the two representations were "unified".

Another effort involved Human-Robot Cooperation to change a car tire (TIRES). The system could interact with the human through pseudo-natural language, learn rules dynamically, and expand its knowledge base of the current situation in real time. So the system self-modifies and learns through human interaction on the task.

Both of these systems required self-modifying code which is rather more difficult to do in other languages. In Lisp this is trivial.

As for commercial sales witness:

Axiom, a 1.2 million line Computer Algebra program written in Common Lisp, was sold commercially by the Numerical Algorithms Group.

YESOPS, an IBM Expert System program implemented in Common Lisp, was sold commercially.

This would be so much easier in Lisp.

Who cares if Lisp is popular?

The "lisp epiphany" is real. Either you "get it" or "you don't".

I've been writing lisp programs for 50 years. I've been paid to program in 60 different languages but nothing compares with lisp.

There is an intellectual "distance" between a problem and its machine solution. I call this the "impedence problem". Lisp lets you think at the most abstract and write to the most specific.

Writing changed the world. But if you give most people a blank piece of paper they don't know what to do with such freedom. Lisp is the "blank piece of paper" of programming languages. Everything, literally everything, comes from you.

I loved my Symbolics machine. It was the closest expression of a "thinking platform" I've ever used. IDEs are horrible for thinking, ever interrupting at every keystroke.

Lisp isn't "popular" because it provides a "thinking platform" you can shape to your thoughts.

Lisp will never be popular. The reason should be obvious.

It seems that your issue is related to funding evidence based medicine, not evidence based medicine itself.

Pasta by Design in Axiom computer algebra system.

https://github.com/daly/PDFS/blob/master/bookvol8.1.pdf

See page 231 onward

You seem to assume that you're in a situation where, by assumption, cash fails but the whole computer and phone network infrastructure "just works". That seems unlikely. "Cash" is on your phone these days. I haven't spent a physical dollar bill in weeks.

SWIFT (Society for Worldwide Interbank Financial Telecommunication) moves a trillion dollars a week. I can't see the distributed blockchain handling that many transactions.

Math is a language. It is the language of science. Even in the life sciences you need to understand sampling statistics and p-values. Chemistry involves reaction rates and exponentials.

Learning any language is hard. But without the ability to read, understand, and use language you limit yourself.

Daly's "Free Coffee Law"... when the coffee stops, the company stops.

I worked at Worldcom. They provided free coffee.

At one point they stopped free coffee because it was too expensive. Worldcom went out of business shortly after that.

When the accounting department starts cutting back "small change" numbers you know the company is in deep financial trouble.

"Instead, the program aims to fund new managing organizations that catalyze community-driven development and growth of OSEs."

I've been writing free software since 1997. At no time was there any need for a "managing organization". The notion that it can "catalyze community-driven development" is utter nonsense.

This NSF effort is obviously being driven by people who have no idea about self-motivated software development.

As a free software developer I need hardware and I need funding for travel to conferences.

The only "managing organization" that makes any sense would be a small group of accountants. Their job would be to act like a University provost. They manage grant funding, dispersing funds, manage receipts, and handle taxes. Most government and business organizations won't give grants to free software developers because they don't know where the money goes.

Beyond Product by Jill Soley and Todd Wilms

I did try it. Rick and I worked on writing network software.

We found that the person typing contributed somewhat less as they had to deal with syntax issues while the other person had time to think. Overall though it was a very effective programming technique. Rick was very strict about his code and wanted to be the person typing. I didn't think it mattered so I agreed.

The downside came later. My boss slammed me for "bothering Rick" who he claimed was doing excellent work. I tried to explain the idea of pair programming but he ignored me. Rick got an excellent rating and a raise. I got slammed and no raise.

I don't recommend pair programming.

"MIT researcher Tamay Besiroglu joined OpenAI cofounder Ilya Sutskever in warning that some machine learning AI may have achieved a limited form of sentience,"

The definition of "sentience" is missing which means the phrase "limited form of sentience" is meaningless. Provide a definition of "sentience" that is testable and can distinguish sentience from non-sentience.

Oh, your robot dog "knows" when you enter a room? That's amazing! So does my motion-detecting light switch. Neural nets recognize patterns? So does statistical pattern recognition, a field I studied in the 1970s. Reinforcement "deep learning"? Wow. You've discovered Norbert Wiener's work circa 1950.

The students get a failing grade.

I realize there is "big money" in the game. I realize Elon Musk is involved. He is someone I greatly respect and admire. I've used the OpenAI gym. I'm researching using deep neural nets for self-reproducing systems (to terraform Mars). I've looked into the reinforcement learning and the associated math. I have a deep background in robotics. I've built AI-related systems most of my life. It really annoys me to see "market-speak" nonsense pollute the research.

It is this kind of ill-founded market-speak that led to the first AI winter.

Such language is a cancer that will again grow into another AI winter.

Your achievements are amazing. Please stop trying to anthropomorphise them.

As far as I'm aware, the only way to run a game on an iPad is through the Apple store. The company did not want to publish their educational games through the store because they were only to be available to students of the company.

Prior to being acquired by Microsoft, github did not require two-factor authentication. You could either use a password or a public key.

Wow. This "history" misses a LOT of other major efforts, muMath (Rich and Stoutemyer), Derive (Rich and Stoutemyer), Maple (Geddes and Labahn), Axiom (Griesmer and Jenks), SageMath (Stein), etc.

"See? You were able to use the hardware you bought, what's the problem?"

"You bought a dishwasher that only allows soap purchased from General Mills. You figured out how to bypass the restriction. what's the problem?" Ummm... I bought a dishwasher to wash dishes.

"You bought a printer that only allows HP print cartridges. You figured out how to bypass the restriction. what's the problem?" Ummm... I bought a printer to print.

"You bought a laptop that only runs MS Store software. You figured out how to bypass the restriction. what's the problem?" Ummm... I bought a general purpose computer to compute.

Vendor lock-in is an "add-on" that controls hardware you bought, not for your benefit but for the company benefit. In some sense you bought "the right to use", not the right to own.

I develop open source software. I want it to run everywhere so I have various versions of hardware and software. I bought this laptop for Windows testing.

Unfortunately I didn't realize it was Windows 10 S. Initially I couldn't even install X11 which has been around since the late 80s. Testing involves a lot of other non-MS software such as emacs, docker and jenkins.

You don't seem to understand the subtle technique of Embrace, Extend, Extinguish.

"Dual-boot is unsupported by both Windows and Ubuntu"... what? I have dual-booted almost every machine I've ever owned, including my Mac boxes. It is one of the standard options offered during installation. How does that "break lots of assumptions"?

I have the expertise to get around the various roadblocks (features?) but it costs a lot of very unproductive time. I skipped over the saga of trying to get WSL2 working, another source of frustration and loss of time.

I also left out the part that Windows will force updates unless you pay for their upgraded version. When I develop code I have a lot of "state" in open windows, an editor, browser tabs, and test runs. Microsoft just decides that the machine is "idle" and decides to reboot it, losing my work. Is it an "unrealistic assumption" that my work is of some value to me?

> Microsoft motion to lock-in developers, now requiring a Microsoft account.

I had a github account very soon after github was introduced. The github account was protected by public key encryption.

A Microsoft account is much more intrusive. Microsoft wants two-factor authentication which requires personal information. The claim is that this adds additional protection but (a) this is open source and (b) the repo is already protected against modification by public key encryption.