[Kick Off] Classic CS Paper Reading Group

Previous thread >>34195 There was some interest expressed in the last thread, so I'm kicking off the Classic CS Paper Reading Group today. Background A group for people interested in finally getting around to reading papers considered "classics" within the field of computer science. We will be using Harvard's CSCI E-191 course as vague inspiration, but we don't have to follow this structure exactly. You can commit as much time as you want though the intention is to choose papers that are considered casually digestible over a two week period. https://canvas.harvard.edu/courses/34992/assignments/syllabus When? Starting from Today (2023-02-06) we will be reading two classic CS papers over a period of two weeks (until 2023-02-20), after which we will begin reading two different papers and so on in this rhythm. The day before the end of the two week period (2023-02-19 in this case) we will try to have a scheduled wrap up discussion on the papers either on IRC or Mumble, depending on what people would prefer. What? The two papers we will be reading are: "ELIZA—A Computer Program for the Study of Natural Language Communication between Man and Machine." by Joseph Weizenbaum. (~10 pages of double column text). "Cramming More Components onto Integrated Circuits" by Gordon Moore. (~4 pages of double column text). Additional reading: https://en.wikipedia.org/wiki/ELIZA https://github.com/norvig/paip-lisp/blob/main/docs/chapter5.md Where? We can discuss the papers either in this thread, or in IRC depending on your preference. The IRC details are: #books on irc.lainchan.org:+6697 (SSL enabled) Thoughts? Nothing is set in stone and I'm happy to hear any and all suggestions.

Cool idea. Happy to see something like this, especially with discussion over IRC.

>>34250 I like the classic cs, But I am probably going to treat this like a general STEM study group and share papers I think other lainanons will like. >>34270 Man your depressing, and a fair bit sad

Sounds Cool, even though I'm late. I'll try to read them when I've some time.

I found some resources that are pretty interesting and related to the papers. Eliza https://sites.google.com/view/elizagen-org/the-original-eliza This gives some historical context to the development of Eliza. It also includes a link to the full source code of Eliza written in MAD-SLIP for the IBM 7094. Here's a link to the source in PDF: https://pomf2.lain.la/f/j1ep8it.pdf And here's a link to the SLIP programmer's manual: https://pomf2.lain.la/f/2msa52ei.pdf A neat video that shows Joseph Weizenbaum briefly talking about Eliza (still, like 15-20 years after the original implementation). https://youtu.be/RMK9AphfLco Cramming More Components onto Integrated Circuits This is famously the paper that planted the seed of what would later become "Moore's Law". Gordon Moore predicted that the number of components on a chip would double yearly until 1975. This would later be tweaked into what's now known as "Moore's Law", the idea that the number of transistors in an IC doubles every two years. https://journals.uic.edu/ojs/index.php/fm/article/view/1000/921 This paper gives a more realistic account of how accurate Moore's Law has actually been, and how accountable Gordon Moore is for actually coming up with it. This quote in particular stuck out to me. >When Moore in 1965 observed that the number of components on integrated circuits seemed to double regularly, this fact was known to many people working in the area [5]. Indeed, it took over a decade, during which Moore had become the co–founder of Intel and its President and CEO, before this observation was called Moore’s Law. During this decade, the content of the original Moore’s Law, however, also started to shift.

>>34270 How old are you? >>34250 I don't recall if I've been in the IRC. Looks like I have to check it out now.

>>34250 Cramming More Components onto Integrated Circuits is a realy good read Thanks

Dropping a suggestion: https://www.cs.cmu.edu/~crary/819-f09/Landin66.pdf It's hard to overstate the influence of this paper on PLT, even this many years later. An absolute banger

Cool idea, count me in. I'm doing a reading challenge anyways so I was actually looking for some reading material. I'm really enjoying the prose of the paper about Eliza. I'm 100% reading this guy's book afterwards.

Reposting from the IRC because it's pretty inactive at the moment. Just read the two papers. Much of the one by Moore went over my head, but I enjoyed the ELIZA paper considerably. Surprisingly relevant today with all the headlines ChatGPT has been garnering. Also gives me the idea for a site in the style of Omegle that randomly pairs users with either a conversational AI or another user and asks them to guess which they are talking to after a certain number of messages... Does anyone know if such a program exists? I suppose the difficulty in that compared to something like ELIZA would be in generating not one conversation base but rather an array of different bases with individual "voices", along with a certain degree of spontaneity inherent to talking with randos on the Internet. >>34292 Agreed, the writing style of the ELIZA paper made it very easy to read. It was also fun going back to the sample interaction at the beginning of the paper of reading and dissecting how the responses were put together.

This review wont be a deep dive as others have covered this much more thoroughly, but I will provide a bit background along with my thoughts and some papers worth reading for those who are interested. Moore's Law, as it's called, states that the number of transistors on a microchip doubles approximately every two years. This observation is attributed by name to Gordon E. Moore, who is probably now most known for co-founding Intel in 1968 with physicist Robert Noyce. This observation has its roots in a paper Moore wrote in 1965, which is the topic of this review. It's a short paper at only 4 or so pages of double column text, and in it Moore as director of the semiconductor R&D labs at Fairchild Camera and Instrument Corporation makes various predictions about the long-term and short-term future of the electronics industry given the recent innovations in integrated circuitry. >Integrated circuits will lead to such wonders as home computers - or at least terminals connected to a central computer - automatic controls for automobiles, and personal portable communications equipment. The electronic wristwatch needs only a display to be feasible today. The original "Moore's Law" prediction is however more subtle, and has a vastly different scope than as it's known today. >For simple circuits, the cost per component is nearly inversely proportional to the number of components, the result of the equivalent piece of semiconductor in the equivalent package containing more components. But as components are added, decreased yields more than compensate for the increased complexity, tending to raise the cost per component. Thus there is a minimum cost at any given time in the evolution of the technology. Moore's original prediction concerns this minimum manufacturing cost per component, to which he says has increased at a factor of two per year and should continue for the decade following the paper. The technical feasibility of this, it seems, was already understood[1] by those in the industry at the time. What sets Moore's prediction apart, in my opinion, was his consideration for the role of economics in the production of integrated circuits, and indirectly how this would impact consumers (home computers, wrist watches and so on). Moore, 10 years later in 1975, gave a presentation at the IEEE International Electron Devices Meeting where he assessed his previous prediction and presented modifications. His prediction had held surprisingly well for the last decade, though he couldn't see it lasting much longer. As a result of this, he revised his original prediction and tweaked it to something more along the lines of: >The number of components on the *most complex* chips would continue to double anually to about 1980, and then after double approximately every two years. This is the prediction that according to Moore, Carver Mead coined as "Moore's Law" shortly after. The biggest differences from the prediction in the original paper being that Moore is no longer considering the "minimum manufacturing cost" of integrated circuits, and the rate of change has been adjusted to doubling every two years (instead of doubling every year). From what I understand, if you squint your eyes it could be said that the semi-conductor industry has generally followed Moore's prediction. This however is hotly debated for something that is dubbed a "law". Regardless of whether it's entirely correct, I would recommend the paper to anyone interested in the early days of computing. [1] https://web.archive.org/web/20221130190552/http://www.welchco.com/02/14/01/60/96/09/2601.HTM#AM46 Recommended reading: [PDF] Gordon Moore's paper "Cramming More Components onto Integrated Circuits" - https://pomf2.lain.la/f/2ewgttlz.pdf Wikipedia: Moore's Law - https://web.archive.org/web/20230216135600/https://en.wikipedia.org/wiki/Moore's_law Ilkka Tuomi's paper "The Lives and Death of Moore's Law" - https://journals.uic.edu/ojs/index.php/fm/article/view/1000/921 * Bob Schaller's paper "The Origin, Nature and Implications of "Moore's Law". The Benchmark of Progress in Semiconductor Electronics - https://web.archive.org/web/20221130190552/http://www.welchco.com/02/14/01/60/96/09/2601.HTM#AM46

Hi all, The first two weeks officially ended yesterday. I was supposed to issue the new papers but life took over. If anyone read the previous two papers, please feel free to share your thoughts. The papers for the next two weeks are: - [1968] Edsger Dijkstra. "Go To Statement Considered Harmful" (3~ pages of text, casual read). - [1978] Leslie Lamport. "Time, Clocks, and the Ordering of Events in a Distributed System" (8 pages of double column text, some basic math) Additional resources: - [1974] Donald Knuth. "Structured Programming with go to Statements" - [2015] M Nagappan et al. "An empirical study of goto in C code" https://pomf2.lain.la/f/qff7vxk6.pdf - Dan Rubenstein: Time, Clocks and Ordering of Events in a Dist. System [PapersWeLove] https://youtu.be/hK6m6WBk-d8

>>34433 I'm glad this is still ongoing! I only got around to read Gordon's text and share my thoughts about it in IRC these last two weeks, but I'm glad I got to read one text at least. Here's to another two weeks! Happy reading everyone :)

I'd like to start over the reading group as it seems to have ended. I suggest we start by reading Selinger's lecture notes on lambda calculus. If you are interested or would like to suggest another book, please join #books.

>>38677 I will (maybe) take a look. Can you explain what made you choose this author in particular? (just curious, never heard of them and the names of the two universities he taught at don't tell me much)

Has it really been a year since the last one? Where has the time gone... >>38677 These notes seem quite a lot longer than the original ELIZA paper or the suggested follow ups. Maybe it being a little shorter could help entice people back? It does look interesting though, so i'm game if others are.

>>38764 Yeah it's trippy to me that it has been a year - the group somewhat fizzled out because of the time commitment, and people's diverting interests. I recently made a textboard specifically for going through CS texts. It can be found here: https://csbook.club/ There's basically no community at the moment, so I'm pretty open to any and all suggestions. If you want to talk about it more find me on #books or #laincha on irc.lainchan.org:+6697 (SSL enabled). >>38677 This could be a cool idea, I'll have a glance over it and see what I think.

>>38766 >textboard Oh wow it's been this successful to dedicate time to make one, eh? Thank you for keeping interest and the project running :)