Hacker News: sn41

New comment by sn41 in "FDA warns top U.S. bakery not to claim foods contain allergens when they don't"

sn41 — Fri, 28 Jun 2024 00:16:24 +0000

I agree. If I am allergic to sesame, then I would also rather play it safe, rather than try a risk.

By the way, Molitor 5901 is the phone number used by the Jackal to know if he could move, in "The Day of the Jackal", if I remember right.

New comment by sn41 in "The case for criminalizing scientific misconduct"

sn41 — Mon, 24 Jun 2024 07:34:02 +0000

With my modest experience as an expert witness in courts, I am against this. It is very difficult to make judges understand, for example, how science works - one can only be so sure, and we go ahead with that, for example, in medicine, as well as cryptography. Judges have a hard time appreciating this nuance.

I agree that something stricter should be done, but it should not be about bringing the legal system into play. I see a fundamental issue with bringing science to trial courts, where rhetoric, appeals to emotions, and other different priorities are paramount, not technicalities about overenthusiastic interpretations, data fudging, p-hacking, empirical anomalies and wilful data manipulation.

Science works by different norms of truth (I would call this statistical) than the judicial system does (beyond reasonable doubt/preponderance of evidence). I believe an international peer scientific committee ostracising a person from publication for X number of years, or forever, might be a better measure than a criminal trial and punishment in open court.

New comment by sn41 in "Detecting hallucinations in large language models using semantic entropy"

sn41 — Mon, 24 Jun 2024 01:11:27 +0000

There seems to be an article on confabulations - seems to be a concept from neuroscience. From the abstract of the article:

"Confabulations are inaccurate or false narratives purporting to convey information about world or self. It is the received view that they are uttered by subjects intent on ‘covering up’ for a putative memory deficit."

It seems that there is a clear memory deficit about the incident, so the subject "makes stuff up", knowingly or unknowingly.

cited from:

German E. Berrios, "Confabulations: A Conceptual History", Journal of the History of the Neurosciences, Volume 7, 1998 - Issue 3

https://www.tandfonline.com/doi/abs/10.1076/jhin.7.3.225.185...

DOI: 10.1076/jhin.7.3.225.1855

New comment by sn41 in "Show HN: I made a puzzle game that gently introduces my favorite math mysteries"

sn41 — Thu, 20 Jun 2024 23:42:22 +0000

Just a quick note: Rahul Ilango is a phenomenal theoretical CS researcher who has made great progress towards understanding the "Minimum Circuit Size Problem" [MCSP], long believed to be, but not yet proven, NP hard. Needless to add, "the username checks out".

New comment by sn41 in "Boeing CEO admits company has retaliated against whistleblowers"

sn41 — Wed, 19 Jun 2024 03:42:09 +0000

Assuming that you trust that the regulatory agency does not have leakers back to the respective companies. When there's a revolving door between regulatory agencies and companies, it gets tricky.

New comment by sn41 in "Indian startup 3D prints rocket engine in 72 hours"

sn41 — Thu, 13 Jun 2024 13:47:18 +0000

It's quite understandable, imho. Immigration is being allowed precisely for cheap labor, especially when citizens are not prepared to go through the extra hardship - for example, I remember reading that the fatalities when the bridge fell in Baltimore around 1 am, were all immigrants, all on duty at that hour.

For specialized labor, there is always a question of possible espionage and back-channel tech transfer. This is not so much perhaps for India as opposed to other technological rivals, but it may be one of the considerations in the immigration policy being counterintuitive.

New comment by sn41 in "Quit Being a Cynic at Work. It's Holding You Back"

sn41 — Tue, 11 Jun 2024 12:35:19 +0000

The title sounded condescending. But the overall theme seems to be that we must not lose trust in others since it may be better for us professionally. I think it is also better for us psychologically.

New comment by sn41 in "An intuitive guide to Maxwell's equations (2020)"

sn41 — Mon, 03 Jun 2024 00:26:39 +0000

Just curious: Sussman and Wisdom have written a book called "Structure and Interpretation of Classical Mechanics" following the classic SICP Scheme book. Has anyone attempted a similar approach for electromagnetics?

New comment by sn41 in "Monads are like burritos (2009)"

sn41 — Mon, 03 Jun 2024 00:16:46 +0000

I don't think you need to understand a concept deeply to use it. The "do..." syntax in Haskell is something that comes naturally to many programmers. It is introduced only at the end of a Haskell course since it uses monads, but many iterative-style programmers switch to that syntax exclusively afterwards. And it actually takes a while to construct the desugared monad syntax.

For a common example of this phenomenon- I took a look into the innards of printf to see how printf("%f",...) and printf("%g",...) works. I am still clueless how it actually works. Does not prevent even beginners from using these.

Also for what it is worth, I think the most useful analogy of monads is "monads are pipes with types", even though it does not give a full understanding of bind.

New comment by sn41 in "All the remedial classes in one place"

sn41 — Thu, 23 May 2024 02:44:09 +0000

I think this is from the traditional "Trivium" - grammar, logic, rhetoric - which was considered to be a well-rounded education (presumably for scions of wealthy families).

https://en.wikipedia.org/wiki/Trivium

In contrast, "Quadrivium" - arithmetic, geometry, music, astronomy - was a follow-up.

https://en.wikipedia.org/wiki/Quadrivium

These constituted the "seven arts". But it was perfectly fine just to be educated in the trivia, if math wasn't your cup of tea.

See the "Yes Minister" clip where Sir Humphrey vehemently denies being so low-standard as to be educated in the sciences - he was good enough to study the classics.

https://www.youtube.com/watch?v=ckgt4VWIsf4

(edited for better grammar, logic and rhetoric.)

New comment by sn41 in "McKinsey Under Criminal Investigation over Opioid-Related Consulting"

sn41 — Thu, 25 Apr 2024 00:40:45 +0000

Another related case in Pennysylvania:

https://www.attorneygeneral.gov/taking-action/ag-shapiro-put...

Quote from the website:

"The complaint, filed along with the settlement, details how McKinsey advised Purdue and other opioid manufacturers on how to maximize profits from its opioid products, including targeting high-volume opioid prescribers, using specific messaging to get physicians to prescribe more OxyContin to more patients, and circumventing pharmacy restrictions in order to deliver high-dose prescriptions. When states began to sue Purdue’s directors for their implementation of McKinsey’s marketing schemes, McKinsey partners began emailing about deleting documents and emails related to their work for Purdue."

New comment by sn41 in "The man who killed Google Search?"

sn41 — Wed, 24 Apr 2024 03:57:57 +0000

Great article. But the author can't be serious about no one knowing who Prabhakar Raghavan is. He is, for instance, the co-author of the definitive text on randomized algorithms [Motwani and Raghavan]. He has also been a well-respected database researcher for many years.

In a previous avatar, Raghavan was a pure theoretical computer scientist. As a student, he won the best student paper in FOCS, the Machtey award, which is kind of a big deal. The work was related to randomized rounding, which is a bread-and-butter technique for LP relaxation approaches to integer optimization, similar to knapsack problems.

This is not to defend any bad decisions he may have made at Google and Yahoo, but to make him an anonymous clueless corporate honcho who is good only at scheming and wrecking companies is bizarre. All this information, moreover, is available on Wikipedia and (cough) Google scholar.

https://scholar.google.com/citations?user=FtMADIMAAAAJ&hl=en...

New comment by sn41 in "New Foundations is consistent – a difficult mathematical proof proved using Lean"

sn41 — Wed, 24 Apr 2024 01:11:45 +0000

Congratulations on the verification of your proof! It must be great to have your research life's crowning work being formally confirmed! Also a great victory for the new foundations of Quine.

New comment by sn41 in "US Air Force says AI-controlled F-16 has been dogfighting with humans"

sn41 — Fri, 19 Apr 2024 01:27:59 +0000

Yes, it's very important to get that on record. Roko, get that basilisk off me!

New comment by sn41 in "David Lynch presents 'A Thinking Room'"

sn41 — Wed, 17 Apr 2024 00:32:02 +0000

Very interesting take, thank you!

Of course, my favorite David Lynch movie is "The Straight Story" - what a heartwarming tale!

New comment by sn41 in "'We're the last bastion of rental': video stores resisting rise of streaming"

sn41 — Mon, 15 Apr 2024 10:51:46 +0000

Rental stores were great, except for the late fees. Many local rentals had good collections of older movies and TV series. They are simply non-existent in the streaming world.

New comment by sn41 in "Good news against dengue"

sn41 — Mon, 15 Apr 2024 04:42:00 +0000

Can confirm this. The second attack is potentially fatal.

New comment by sn41 in "2023 ACM Turing Prize awarded to Avi Wigderson"

sn41 — Thu, 11 Apr 2024 01:07:09 +0000

An important idea is to use what are called worst-case-to-average-case reductions. You convert a boolean function f that is worst-case hard to another function g which is average-case hard. In other words, if g is average-case easy, then f is worst-case easy.

These reductions are a major achievement in complexity theory. Constructions use tools like combinatorial designs, list-decoding of error-correcting codes, expander graphs, extractors etc. A good overview of these methods is in Arora and Barak's "Computational complexity", in, I believe Chapters 19 through 21.

New comment by sn41 in "India’s electric rickshaws"

sn41 — Tue, 09 Apr 2024 15:29:32 +0000

I think they are too tall for the wheelbase. The design is narrower than traditional auto-rickshaws. I think one of the issues is I think that the battery is not heavy enough to stabilize the center of gravity once you start tipping over - instead of stabilizing, it probably makes it tip over faster.

I am not talking about crash testing or anything that sophisticated. Something as basic as the tendency to overturn is much higher. I have seen one happen right beside me, when one of the wheels went into a pothole, and the rickshaw tilted and overturned. A traditional auto (the standard Bajaj model ICE, for example) would have "jumped", but not overturned.

On this same note, the Bajaj ICE is slightly narrower at the top than the base, in a tapered design - this will increase stability.

Piaggio Ape is tapered towards the top and has backwheels turned inwards -

https://en.wikipedia.org/wiki/Piaggio_Ape#/media/File:Ape_Ca...

e-rickshaws show none of these features.

New comment by sn41 in "India’s electric rickshaws"

sn41 — Tue, 09 Apr 2024 14:28:16 +0000

E-Rickshaws are very unstable, and they are overloaded on North Indian roads, with people sitting even in the shared driver's seat, increasing the lopsided tilt. A distant relative of mine was killed when the e-rickshaw in which he was riding overturned and crushed him. They are easy to tip over, for example, when going over a pothole filled with water after rains. Traditional rickshaws have a wider base, and lower center of gravity. I think this particular design, popular all over north of India, is deadly. A safer design must be enforced.