• 0 Posts
  • 202 Comments
Joined 6 months ago
cake
Cake day: November 5th, 2025

help-circle

  • No, not from a startup. It might be worth mentioning that I’m in the U.K. so the terms and lingo may differ from yours.

    I’ve had RSUs from series D scale ups. I’ve had options from start ups. Typically in the UK they’re wrapped as an investment you make on entry (using a loan the company offers you) to only pay capital gains tax (25% in the UK). And I’ve had RSUs from big, established enterprises.

    What’s common to them all, for me, is that they’ve broadly not paid out what they were advertised to, either because the stock falls (Unity springs to mind), you leave before anything material vests (and the hiring company matches your RSUs) or there’s no liquidity event.

    I trust cash, paid into my bank account. The rest, IMHO, is just trumps (US: farts) in the wind.





  • My first computer was an Amstrad 664, with a green screen. I’m old. And I’ve been around Vim and EMacs from time to time and I love the console but for the love of god, since GUIs became the normal way to interact with computers, I just install micro now and have the same hotkeys across all the modes of interaction.

    Speed of typing really isn’t the defining productivity measure for code.

    Now I use VS Code in a GUI and micro on the console and that provides a reasonably consistent way of interacting with text.







  • You propose we do what exactly with AI?

    And I’m not ignoring “rapid deskilling”. I’m specifically arguing against the very specific point that TFA made that “we will forget how to code”. That’s clearly not what will happen.

    Some people today still know assembler. Some people still how to design CPUs. Some still know lithography. Not as many as before, because we just don’t have the need.

    I marvel at the number of engineers who don’t think about memory allocations at all. But it’s ok - they probably don’t need to, for the task they’re solving.


  • I’m sorry but I’m not buying it. We’ve been sitting on top of abstractions for years.

    What, you think the average engineer knows assembly? You think they know how to design gates? You think the gate designers know how to make lithography work?

    How do you build a new factory, of anything? By using machines that have been built in other factories! We’ve got a highly redundant, interwoven mesh of things that rely on others things to be made. There is no “starting point” that you can trace today - all is done with something else that’s also complex.






  • That’s not quite my understanding of EEE.

    • Embrace - adopt something that someone else has done
    • Extend - add proprietary extensions on top of the original, quicker than the original owner can
    • Extinguish - Kill the original owner off by moving quicker then either slow down or kill your own support for the product

    What the AI model owners are doing seems to me just to be normal loss-leading with a view to gain market share.


  • They are currently selling it at a huge loss, agreed. They’ve got plenty of runway for specialised hardware prices to come down, for companies to get hooked and plugged into the ecosystem and for real value to be demonstrated.

    When this happens they’ll raise prices and companies will gladly pay it.

    Profit at this point is not relevant, seen from the perspective of investors.