Derek M. Jones

Karma: 44

Derek M. Jones 14 Apr 2020 12:43 UTC
3 points
on: Scraping websites currently free due to coronavirus
The ACM is offering free download of their articles: https://dl.acm.org/

Derek M. Jones 14 Apr 2020 20:46 UTC
5 points
on: Is this viable physics?
Take some interesting ideas that allow larger structures to be built up, run an awful lot of computer simulations, and then have somebody who knows a huge amount about physics look for those outputs that match how parts of the universe have been previously modeled (with some success).
Where are the predictions? There are no predictions about basic stuff we know, like the electron orbitals of a Hydrogen atom, let alone predictions about stuff we don’t know.
This work looks interesting, and Wolfram is a great story teller. I hope something comes of it, but at the moment it is pure arm waving of just-so stories found in the output from a few computer runs.

Derek M. Jones 6 May 2020 13:42 UTC
4 points
on: Maths writer/cowritter needed: how you can’t distinguish early exponential from early sigmoid
More examples of the difficulty of predicting the future using fitted regression models:
https://shape-of-code.coding-guidelines.com/2020/04/19/predicting-the-future-with-datalogistic-regression/

Derek M. Jones 21 Jun 2020 16:18 UTC
2 points
on: My weekly review habit
You are one of the few people with the discipline to record what they do and create todo lists. I could not keep this up for a week. Do you try to estimate the time it will take to complete a task?
Have you done any global analysis of your data? I analyse software engineering data and am always on the lookout for more data. I offer a free analysis, provided the data can be made public (in anonymous form). Here is one I did earlier:
http://shape-of-code.coding-guidelines.com/2019/12/15/the-renzo-pomodoro-dataset/

Derek M. Jones 14 Aug 2020 17:49 UTC
2 points
on: Room and Board for People Self-Learning ML or Doing Independent ML Research
I wish you lots of luck. Don’t go so far north that day trippers from the south cannot drop by :-)
http://shape-of-code.coding-guidelines.com/2017/03/31/gentleman-scientists-in-software-engineering/

Derek M. Jones 27 May 2021 20:40 UTC
1 point
on: One Year of Pomodoros
Here is an analysis of one person’s 10 years of using Pomodoros, including the data:
http://shape-of-code.coding-guidelines.com/2019/12/15/the-renzo-pomodoro-dataset/

Pomodoros worked during a day: an analysis of Alex’s data

Derek M. Jones31 May 2021 13:27 UTC

5 points

0 comments3 min readLW link

(shape-of-code.coding-guidelines.com)

Derek M. Jones 23 Oct 2021 11:55 UTC
−5 points
on: How to estimate my probability of getting a (competitive) job in Academia
Your post shows that you do not love the subject and that you lack the blind certainty of achieving your goal.
I suggest you give up now.

Derek M. Jones 23 Dec 2021 0:25 UTC
2 points
on: A spreadsheet/template for doing an annual review
Fishing for data here.
If anyone estimates what they plan do during the day and records what they actually achieved, then I’m willing to do a free analysis provided an anonymous version of the data can be made public.
Ten years of Pomodoro data: http://shape-of-code.com/2019/12/15/the-renzo-pomodoro-dataset/
From an earlier Lesswrong post: http://shape-of-code.coding-guidelines.com/2021/05/30/pomodoros-worked-during-a-day-an-analysis-of-alexs-data/
A connection between the brain’s number systems and estimating: https://shape-of-code.com/2021/09/26/the-approximate-number-system-and-software-estimating/

Derek M. Jones 9 Jan 2022 17:14 UTC
−1 points
on: Total compute available to evolution
This is a poorly thought out question.
Evolution implies a direction of travel driven by selection pressure, e.g., comparative fitness within an environment.
A sequence of random processes that are not driven by some selection pressure is just, well, random.
What is the metric for computational effort?
Are you actually interested in computational resources consumed, or percentage of possibilities explored?

Derek M. Jones 9 Jan 2022 20:09 UTC
4 points
in reply to: redbird’s comment on: Total compute available to evolution
Chemical space, https://en.wikipedia.org/wiki/Chemical_space, is one candidate for a metric of the possibilities.
The book “Chemical Evolution: Origins of the Elements, Molecules and Living Systems” by Stephen F. Mason might well contain the kinds of calculations you are looking for.

Derek M. Jones 23 Jan 2022 18:35 UTC
−26 points
in reply to: iceman’s comment on: Why rationalists should care (more) about free software
AI alignment is pseudo-science.

Derek M. Jones 24 Jan 2022 10:27 UTC
−1 points
in reply to: Alex Vermillion’s comment on: Why rationalists should care (more) about free software
Would you have downvoted the comment if it had been a simple link to what appeared to be a positive view of AI alignment?
Truth can be negative. Is this forum a cult that refuses to acknowledge alternative ways of approaching reality?

Derek M. Jones 2 Feb 2022 17:57 UTC
6 points
on: Competitive programming with AlphaCode
My reading of Appendix A is that the group did its own judging, i.e., did not submit answers to Codeforces.
They generated lots of human verified test data, but then human implementors would do something similar.
They trained on Github code, plus solutions code on Codeforces. Did they train on Codeforces solutions code that solved any of the problems? Without delving much deeper into the work, I cannot say. They do call out the fact that the solutions did not include chunks of copy-pasted code.
To what extent are the successes presented representative of the problems tried? That is, did they try to solve lots of problems and we are seeing the cases that worked well? The fact that they were able to get solutions to some problems was impressive.
The solved problems had short solutions. How well does the technique scale to problems requiring more code for their solution? I suspect it doesn’t, but then there are applications where the solutions are often short.

Derek M. Jones 13 Mar 2022 13:53 UTC
1 point
on: How to Lumenate (UK Edition)
If you want light, the term you need to know is corn bulb (also available in screw fit).

Derek M. Jones 13 Mar 2022 14:02 UTC
1 point
in reply to: chanamessinger’s comment on: How to Lumenate (UK Edition)
Click on green text, or Amazon UK have a search box, and Google ads displays a 4000 lumen bulb.

[Linkpost] Growth in FLOPS used to train ML models

Derek M. Jones14 Mar 2022 11:28 UTC

3 points

3 comments1 min readLW link

Derek M. Jones 14 Mar 2022 17:24 UTC
1 point
in reply to: gwern’s comment on: [Linkpost] Growth in FLOPS used to train ML models
Thanks, an interesting read until the author peers into the future. Moore’s law is on its last legs, so the historical speed-ups will soon be just that, something that once happened. There are some performance improvements still to come from special purpose cpus, and half-precision floating-point will reduce memory traffic (which can then be traded for cpu perforamnce).

Derek M. Jones 25 Mar 2022 15:55 UTC
5 points
on: Fake Journal Club proposal
Figuring out that a paper contains fake research requires a lot of domain knowledge. For instance, I have read enough software engineering papers to spot fake research, but would have a lot of trouble spotting fake research in related fields, e.g., database systems. What counts as fake research, everybody has their own specific opinions.
My approach, based on experience reading very many software engineering, is to treat all papers as having a low value (fake or otherwise) until proven otherwise.
Emailing the author asking for a copy of their data is always interesting; around a third don’t reply, and a third have lost/not kept the data.
Spotting fake research is a (very important) niche topic. A more generally useful proposal would be to teach people how to read papers. Reading one paper might almost be worse than reading none at all, because of the false feeling of knowing it gives the reader. I always tell people to read the thesis from which the paper was derived (if there is one); a thesis provides a lot more context and is a much easier read than a paper (which is a very condensed summary of the thesis). Researchers much prefer to have their paper cited, because thesis citations don’t ‘count’.
Is a Fake journal club worth the effort? It’s possible to spend more time debunking a paper than was spent doing the original research, and for nothing to happen.

Derek M. Jones 25 Mar 2022 17:03 UTC
1 point
on: Fake Journal Club proposal
This post is about journal papers, not answering real world questions (although many authors would claim this is what they are doing).
With regard to nuclear weapons, Dominic Cummins’ recent post is well worth a read, the book he recommends “The Fallacies of Cold War Deterrence and a New Direction” is even more worth reading.

Is MAD doctrine fake research, or just research that might well be very wrong?

Derek M. Jones

Po­modoros worked dur­ing a day: an anal­y­sis of Alex’s data

Chemical space, https://​​en.wikipedia.org/​​wiki/​​Chemical_space, is one candidate for a metric of the possibilities.

The book “Chemical Evolution: Origins of the Elements, Molecules and Living Systems” by Stephen F. Mason might well contain the kinds of calculations you are looking for.

[Linkpost] Growth in FLOPS used to train ML models

Pomodoros worked during a day: an analysis of Alex’s data

Chemical space, https://en.wikipedia.org/wiki/Chemical_space, is one candidate for a metric of the possibilities.