Hacker News Reader

the_decideron Ask HN: Unemployed almost a year after graduating MIT – a rant

One option is to apply to a Phd program that pays a stipend. If you drop out after you pass your quals (approximately 2 years in) you automatically get a Masters, plus the cred associated with a higher degree

the_decideron You Don't Own Your Reputation

Interesting example use-case of sharing encoding Airbnb reputation on the blockchain to be shared across competing services. There remains a question however of how to get buy-in from those competing businesses to integrate their website or app in order to display that decentralized reputation information. Though in theory, one could encode the address of the short-term rental or house to be easily accessible on the blockchain.

the_decideron Will A.I. Trap You in the "Permanent Underclass"?

https://archive.is/9as3r#selection-309.0-309.49

the_decideron Robert Musil Forgotten Plays Inspired His Greatest Work of Fiction

I remember reading it in 2015 and identifying with main character; a 32-year-old seemingly-talented man who pursued the trappings of success in socially-validated directions like the study of math and engineering but ultimately has nothing to show for it besides a comfortable, hedonistic lifestyle in a technocratic, bureaucratic, meeting-prone society that teeters at the edge a coming collapse which will render all these meetings irrelevant.

the_decideron Abundance Starts with Mobility

Interesting to see NYC public transport compared in a less favorable light to the San Francisco where BART can be a nightmare and busses are not a reasonable way to get around. I guess the abundance grass is always greener…

the_decideron Two new books on John Calhoun and his rodent experiments

"Universe 25 was called Universe 25 for a reason. The study had been carried out on 24 other Universes, each one inevitably leading to the complete collapse of rodent society". This is a quote from a very amusing online lecture on the subject, which explores the cultural impact of the experiment, as well as ways that it can't be applied to humans. https://www.youtube.com/watch?v=CG2Xh2JCoMY

the_decideron Rule-based NLP system beats LLM for analysis of psychiatric clinical notes

This work is based on a fine-tuned Google Palm Model from 2022. I'm not sure if this is a fair comparison to the latest groundbreaking series of LLMs

the_decideron [dead]

I have fond memories of crazy, quirky nights at the DNA Lounge prior to my leaving the Bay Area.

the_decideron Using AI to spot edible mushrooms could kill you

There is a simple 100% precision algorithm for detecting non-poisonous mushrooms in the continental United States. Look under the cap; if its spongy it won't kill you. It it's not spongy (aka ruffled), it could be poisonous so toss it out. Yes, your edible mushroom recall might be slightly reduced, but you're guaranteed to not get poisoned.

the_decideron What are the “trophy jobs” in tech?

Mid-level Engineering Manager at FAANG pre-tech downturn.

the_decideron My Meeting with David Bohm

Photons follow wave-particle duality and electromagnetic waves can be linked to probability distributions. https://en.m.wikipedia.org/wiki/Photon#:~:text=A%20photon%20.... However, unlike electrons, photons have no exclusion principle (aka you can’t have multiple electrons in the same position) nor a repulsive potential. So the same probability wave can apply to countless photons. If you “reach into” a high amplitude portion (high probability) of an electromagnetic wave, you will “scoop out” a handful of photons. If you reach into a zero amplitude portion of the wave, you will scoop out none

the_decideron Hexing Putin with a spell written by GPTChat

All text, including the title of the video was also written by the model

the_decideron Kids in the Hall: Comedy Punks

Francesca Fiore!

the_decideron Social Media Info for Military Unit Involved in Massacre Has Been Made Public

I imagine that some of the Skype links could be out of date. Though the social media links and photos appear to be legit.

the_decideron Social Media Info for Military Unit Involved in Massacre Has Been Made Public

Names / ranks / birthdays released by Ukrainian ministry. https://gur.gov.ua/content/voennye-prestupnyky-neposredstven...

the_decideron Starting ML project to identify Russian soldiers currently occupying Ukraine

Please state your argument for why the use is inappropriate. The full names of 120K invaders have been leaked. The names can be queried for on search engines / social media sites. Some of these names yield photos of men in military uniforms. These photos can be identified via classifier. Afterwards, relatives of the occupiers can be contacted and encouraged to promote anti-war messages

the_decideron Multiple Sclerosis Causality

An immunological biomarker aasociated with MS had been discovered a few years back, but sadly nothing came of that research. https://www.nature.com/articles/s41598-017-04439-5

the_decideron I analyzed 20k recommendations made by Jim Cramer during the last 5 years

Yep, there's a reasonable chance this analysis suffers survivorship bias. Outliers whose stock-prices severally crashed are less likely to be represented. Hence, impactful sell recommendations are probably missing from the data, explaining why sell-side recommendations don't appear to to perform as well.

the_decideron Ask HN: Who wants to collaborate?

I’m exploring how reinforcement learning can be applied to safely carry out Solar Geoengineering in order to delay the damage from climate change. https://arxiv.org/abs/1905.07366

the_decideron What the Russians thought of James Bond in the 1960s

I read a few of the earlier Bond books; there the antagonistic organization isn’t Spectre but the actual WW2 era Soviet counter-intelligence org SMERSH, whose name is short-hand for “Death to Spies”. https://en.m.wikipedia.org/wiki/SMERSH

the_decideron The American dream is on life support in the Bay Area

This Fall I moved to NYC after 16 years residing in San Francisco. I’ve been amazed how clean and safe the Manhattan streets seem relative to SF. Over the course of my life in San Francisco I’ve: 1) Had a crazy homeless person try to push me under a moving Bart. 2) Was assulted by two thugs outside of City Lights Books 3) Seen multiple laptop snatches at various coffee shops 4) Had a good friend who twice got punched in the face out on the streets.

I loved SF back in the day, but that city feels gone. Buried under a pile of dangerous filth. Now that I’m in New York, I couldn’t be happier!

the_decideron Highly accurate protein structure prediction with AlphaFold

Could somebody please share a link to the PDF? Sci-Hub is currently not processing new papers. It's a shame.

the_decideron A Concrete Introduction to Probability (2018)

The following data science book also does a great job of balancing problem solving with underlying theory. https://www.manning.com/books/data-science-bookcamp And it starts with sample-space probability problems in Python, much like Peter’s tutorial.

the_decideron Ask HN: What are the best data science bootcamps?

I don't think you need a bootcamp. Foundational DS concepts can be self-tough given the right resources (such as Manning's "Data Science Bookcamp" https://www.manning.com/books/data-science-bookcamp)

the_decideron [dead]

“In the near future, it will seem unfathomably bizarre that one of the world’s most dangerous propagandists was allowed nearly unfettered access to the most powerful communication tools in human history, all out of some misbegotten allegiance to a warped notion of free speech, and that the only thing that could break the spell, in the end, was an attempted coup.”

the_decideron Claude Shannon Invented the Future

I agree with this sentiment. "Dark Hero of the Information Age" https://www.amazon.com/Dark-Hero-Information-Age-Cybernetics... lays out how both Weiner and Shannon created information theory independently but Weiner took it further; incorporating feedback as a way of lowering the entropy and helping lay the path for today's ML revolution. The book goes on to lay out how Weiner's growing popularity in the Soviet Union and socialist views (he feared automation leading to unemployment) led to the blacklisting of his contributions within the academic community. Its no wander that the first trained Deep (more than three layers) Neural network with published in a Soviet Cybernetics journal in 1970.