Preferences

psyklic
Joined 2,127 karma
I am interested in AI, mathematics, and programming languages. Co-founded Hyphos (social discovery/dating), coded numerous startup products, teach & consult in AI/ML, and am a Caltech alum.

website: https://danwil.com/

email: dan@<website>


  1. Mentorships are simply how most junior scientists get started. Even in grad school, most students initially take on projects their advisors have pre-qualified as interesting.

    To make high school-level competitions more fair, we should likely prioritize access to researchers for all smart, hard-working high schoolers rather than only those who are nearby a university or have wealthy parents.

  2. RAM too. They want +$1K to go from 64GB to 128GB, with no other spec changes. It's a way of segmenting the market -- those who actually need it are willing to pay a lot more (e.g. for AI / 4K videos).
  3. When I'm shown a pre-recorded demo, I suspect the product isn't very robust.
  4. Nice job! As a US-based consultant, I've found it's always best if you immediately suggest a contract. This way you can ensure it starts with terms favorable for you.

    I've also found it's pedagogically helpful to have two versions of each contract, a consultant-favored and consultee-favored. This way you can understand how each clause may be tweaked to benefit each party. For example, this book does this (US-based): https://www.amazon.com/Consultant-Independent-Contractor-Agr...

  5. Also Trump asked to direct where the investment goes himself. It looks more like a gift than an investment.
  6. Sure, let's suppose they go down some. For the same reasons you'll still end up paying more for the same Chinese products, which will raise domestic prices.
  7. In actuality, the Chinese company can't half their prices. So instead of paying $0.50 for a Chinese bolt, consumers now will pay $1.00.

    Unfortunately the US bolts will not be plentiful enough. They'll also have to import steel to meet new demand, increasing their price. So ultimately you'll still buy the Chinese product but it will now cost double the price -- $1.00 after tarrifs. Hence the price of everything that has a bolt will increase.

  8. If you don't know the foundations well, you don't belong in a postgrad program. That's the reality and how it currently works. Undergrad teaches you those foundations.

    Anyone can try doing research, even undergrads who half-know the foundations. However, trying research doesn't mean you have the background to do great research or to succeed in a postgrad program.

  9. I'm not sure that's a simple argument and can't imagine many would agree.

    Undergrads who do research generally aren't very good at research yet. A major reason is they either lack or don't fully understand the pre-reqs, which they progressively and cumulatively learn during undergrad. A student can be incredibly smart, but acquiring a strong rigorous math background will still take years.

  10. Undergrads who care about learning and research will take the most challenging classes, do research with professors, and surround themselves with other strong students who will push them.

    Even at top universities, very very few freshmen are capable of doing high-quality research immediately. They'd be better served learning the foundations inside and out with a cohort of similarly strong students to challenge them.

  11. Learning a specific technology for a single project may have a short half-life. However, good coders aren't defined by tech knowledge, but by their deep understanding. If you can make great presentations in Powerpoint, everyone knows you'll still make good presentations in Google Slides.
  12. This is like saying if you have nothing to hide, you should consent to police searches. What is found only provides more possibly coincidental evidence to use against you (just as DNA provides evidence about your potential health).
  13. The article's main point seems to be that students cheat with AI; hence, universities must "resist" AI to preserve critical thought.

    Although universities are certainly against cheating, the responsibility has always been on the student not to cheat. Universities do not oppose useful technologies simply because they may be misused for cheating.

    Put another way, the role of a university is to "discover and invent the future." In this light, universities will be more interested in developing AI than so-called "resisting" it. This is especially since it has already yielded breakthroughs in science, e.g. a Nobel Prize in Chemistry for protein prediction.

  14. It's the same with math -- those with a deep understanding will be able to do it by hand.

    Everyone else will take the easy route -- higher immediate productivity but overall capable of less.

  15. Nice work and blog! You might consider crossposting some blog entries to LessWrong for more visibility of your research. People are very interested in new ideas on benchmarking.
  16. Removing crypto regulation only benefits those at the top and scammers, since it gives them one-sided control of the market with few consequences. Which is exactly what's happening.
  17. In everything, there is some low-hanging fruit that yields an impactful outcome for minimal spending.
  18. Good news -- if you're not interested in extending state-of-the-art and simply want to call APIs, you don't have to learn ML deeply.
  19. As of Feb 18, WaPo says merely 14% of US adults overall support pardoning people convicted of violent crimes (and only 32% of Republicans).

    For nonviolent pardons, 42% overall support it (and 76% of Republicans).

    [1] https://www.washingtonpost.com/politics/2025/02/20/trump-pol...

  20. You likely don't know anyone because there is luck involved in success. Having to initiate a lawsuit would mean you are very unlucky.

    For example, maybe your business partner fraudulently took most of the shares for himself despite a written equal equity split. Here, you can either sue or just walk away and give him your shares.

    Keep in mind that "successful" people are incentivized to say all lawsuits against them are unfounded and they are merely the "victim". Why assume they are telling the truth? It's at least as likely that they had something to do with causing the lawsuit.

  21. You can win economically, especially if it's a slam dunk case and hence get a lawyer on contingency. That said, it can take a long time to resolve which can prolong any stress surrounding it. And even if the other party is clearly, obviously at fault, they may spread falsehoods about you and the case. Especially if this is done in private, it's practically difficult to do anything about.
  22. Stratechery just published detailed specific examples of when it's useful and not: https://stratechery.com/2025/deep-research-and-knowledge-val...
  23. JetBrains' local single-line autocomplete model is 0.1B (w/ 1536-token context, ~170 lines of code): https://blog.jetbrains.com/blog/2024/04/04/full-line-code-co...

    For context, GPT-2-small is 0.124B params (w/ 1024-token context).

  24. The President can offer pardons for criminal matters. However, he is required to uphold laws passed by Congress, particularly bipartisan ones affirmed by the Supreme Court.

    For example, why would the President have a veto power if he can simply post-facto ignore laws they pass?

  25. No? Simply look at Elon's X account. And that's the point - if there's "controversy" then it doesn't get noted, falsely implying it's true.
  26. Another major issue is that if enough people believe/support a lie, it doesn't get noted. So in practice, you end up having "official" lies that go un-noted, making them appear true.
  27. Not seeing fact checks likely means it's working: "Once third-party fact-checkers have fact-checked a piece of Meta content and found it to be misleading or false, Meta reduces the content’s distribution "so that fewer people see it.""

    The issue with Community Notes is that if enough people believe a lie, it will not be noted. This lends further credence to a certain set of "official" lies.

  28. Everyone must learn new things after university. For example, LLMs, blockchain, and new frameworks likely weren't covered. This is why they say in university you learn how to learn.

    My personal opinion is that universities should teach things that would be useful to an expert which you won't have time to learn at a job. Like math and gaining deep understandings of things.

    That said, I'd assume if you went to university in computer engineering (i.e. embedded systems/digital electronics), you'd have numerous labs and projects designing and using microcontrollers.

    In CS, you might have one lab/class on CPU architecture/assembly but I'm guessing this would be a specialized focus area.

  29. Larger clients will often require you to have liability insurance. I was able to get some quotes for this as an independent consultant myself. However, given the extra expenses you'll have to charge more. I'd suggest using this higher cost as leverage toward the liability they want you to accept (for you, ideally as-is). I'd also explain since it's hourly if they dislike your work they can just replace you.

    IANAL but my understanding is that for contractual work in the US you can only realistically be sued for at most the amount of the contract (supposing non-negligence). Asking you to accept liability beyond this sounds risky.

This user hasn’t submitted anything.

Keyboard Shortcuts

Story Lists

j
Next story
k
Previous story
Shift+j
Last story
Shift+k
First story
o Enter
Go to story URL
c
Go to comments
u
Go to author

Navigation

Shift+t
Go to top stories
Shift+n
Go to new stories
Shift+b
Go to best stories
Shift+a
Go to Ask HN
Shift+s
Go to Show HN

Miscellaneous

?
Show this modal