physicsgraph

Find something that other people would see as infeasible or inaccessible and work on that. You are in a position of privilege that can be used to advance well-being and knowledge.

Deciding whether to explore challenges solo versus as part of a team is a crucial differentiator.

To summarize the main two take-aways, measure ROI (return on investment) and be Agile.

Have you searched through https://hn.algolia.com/?q=ask+hn+blogs ?

Sounds interesting, but the github repo doesn't seem to have enough README content to explain who should use the project or how to use the project.

The original blob post [0] referenced hypergraphs and other more complicated structures like property graphs.

[0] https://news.ycombinator.com/item?id=39592444

The back-and-forth exchange between blogs, each with comment threads on HN, is a great use of the Internet.

Thanks for the suggestion. I'm new to running LLMs so I'll take a look at your suggestion [0]. My ~10 year old MacBook Air has 4GB of RAM, so I'm primarily interested in smaller LLMs.

[0] https://github.com/ggerganov/llama.cpp

Your assessment is exactly correct -- the blog post is my note-to-self about getting the repo to work. My "added value" in the post is a Dockerfile for ease of installation.

The premise towards 1-on-1s that "regular 1-on-1 meetings often feel like a burden and overhead" indicates the value of 1-on-1s isn't clear to the author. The 1-on-1s are an opportunity for supervisors to learn about the long-term goals of their employees.

The author's content presents short-term oriented assessment more appropriate for team syncs.

I've collected questions to ask in 1-on-1s here [0].

[0] https://graphthinking.blogspot.com/2022/10/one-on-one-questi...

Previous post: https://news.ycombinator.com/item?id=25607809

The training data used [0] is written using "leaps of logic" which provide context for how the solution was derived. There isn't sufficient detail in the training data to formally check each step using a Computer Algebra System [1]. Therefore, whether or not any output from the OpenAI software is actually correct is left as an exercise for the user. Answering "Was the hallucination correct?" involves manually checking each step.

This advancement is good, but it's limited by the precision of the training data.

[0] https://artofproblemsolving.com/wiki/index.php/2015_AIME_II_...

[1] https://en.wikipedia.org/wiki/Computer_algebra_system

I agree that Approval voting is easier to understand and implement. I got to experience this a few years ago when I tried to implement RCV for a comparison with Approval voting.

I've been active in reforming voting systems since I think that makes a difference to how politics works.

Is there a path or actions that address class issues other than the standard "run for office" and "vote for people who will address class-related issues"?

A useful counter-balance perspective is provided by [0] Tyranny of Structurelessness.

[0] https://www.jofreeman.com/joreen/tyranny.htm

This post is well-timed.

I'm aware of many previous posts on HNews and the main documentation [0]. Does anyone have suggestions for getting started from a non-CompSci perspective? I am comfortable with Python but I'm not confident that suffices.

0: http://lamport.azurewebsites.net/tla/hyperbook.html

Previous post with 7 comments: https://news.ycombinator.com/item?id=19562718

This seems to be a preview of an article available as PDF after you sign in. Without the main content the purpose isn't clear to me. Am I missing something?

The best part of this was getting to the tab I had opened long ago and wondering, "how did I get to this page?" and then realizing (after reading it) that I was rick-rolled by HN.

I'm using Chrome on Android and am getting the intended page (no error).

If "startup" is defined as a for-profit company, then I'm not aware of any. There are non-profit organizations that work on alternative voting (Approval voting [0] and STAR voting [1] and Instant Run-off/Rank Choice), non-partisan primaries, non-partisan districting, proportional representation (no districts), electoral college reform.

[0] https://electionscience.org/

[1] https://www.starvoting.us/

Experienced customers training new-to-the-process participants is a surprisingly common aspect of bureaucratic processes. The peer participants or management are the other possible sources of training, but those stakeholders are not as incentivized to train the new member.

(Author here.) Yes, I ignored that concept in the model. I referred to that as "transfer learning" in the post.

I'm unable to find the specific company and website in my old emails, but I have seen an example on the web of what you're describing. The reader could select whether to expand or collapse sections (e.g., I want more detail on this subsection).

The web interface was similar to a medium blog post, but the amount of work for both the writer and the reader was more than I think most folks are willing to invest.

The claim that "You are line if your role makes up the largest fraction of the org chart" has a counter example: the number of pilots in the Air Force is relatively small compared to the number of non-pilot positions, yet the pilot is the only line role in the Air Force.

I've a good experience with uBlacklist, an extension that blocks results by domain in the search page results.

This is a survey of potential industrial use cases that spans a variety of domains. The article feels like a submarine [0].

It's hard for me to imagine getting from where most businesses are now (unable to share data internally among teams, limited investment in web dashboards that show live data) to a fancier "live data overlaid with physical systems." Is there that much value differentiation? I get the tech appeal, and it looks cool. I'm not clear the cost/benefit is there in general.

I do suspect AR will be adopted in two scenarios. In niche domains where cost/benefit is acceptable (where?). And the other scenario is "shiny buttons and blinky lights" pet projects.

[0] http://www.paulgraham.com/submarine.html

The advantage to smaller teams is nimbleness; the advantage to larger teams is robustness to perturbations. These are not mutually exclusive, but in practice a small (nimble) and robust team is challenging to build since the balance of personalities and skills requires intent and access to amazing hiring opportunities.

Thanks for the recommendation. It looks like an interesting way to meet random technical people.

What are your thoughts on the hospitalization of infected people? Hospitals are at capacity or overflowing.

This article is a well-written deep dive on parallel performance debugging that ends up identifying the difference being due to how caching is handled on multi-CPU vs single-CPU systems. Thanks to the author for taking time to write up their troubleshooting process description.