Kevin Lacker (@lacker) 's Twitter Profile
Kevin Lacker

@lacker

Aspiring alien hunter. Formerly: Parse cofounder, Facebook eng manager, Google search quality engineer, college mathlete

ID: 14254499

linkhttp://lacker.io calendar_today30-03-2008 00:48:48

15,15K Tweet

6,6K Followers

1,1K Following

Kevin Lacker (@lacker) 's Twitter Profile Photo

Every time jamming defeats drones in the Russia-Ukraine war, the case gets stronger for full autonomy. Once a drone can carry out its mission without communication, it can't be stopped by jamming. Fully autonomous drone warfare seems inevitable.

Every time jamming defeats drones in the Russia-Ukraine war, the case gets stronger for full autonomy. Once a drone can carry out its mission without communication, it can't be stopped by jamming. Fully autonomous drone warfare seems inevitable.
Mark Farrell 🥥🌴 (@markfarrellsf) 's Twitter Profile Photo

Our reputation as a leader in technology & innovation is because we embrace new ideas & serve as a sanctuary for research & development. With the slowest post-COVID economic recovery of any major US city, we shouldn't over regulate an industry so early. nytimes.com/2024/08/14/tec…

Kevin Lacker (@lacker) 's Twitter Profile Photo

Quite a weird error message to get inside my text editor while I was sitting there doing nothing. Surprisingly enough, following the link made it perfectly clear to me what was going on.

Quite a weird error message to get inside my text editor while I was sitting there doing nothing. Surprisingly enough, following the link made it perfectly clear to me what was going on.
Kevin Lacker (@lacker) 's Twitter Profile Photo

This fall, our society is starting a large-scale Turing test. Thousands of students taking media studies classes will write essays for their homework. Thousands of other students will use ChatGPT to write those essays. Can their professors tell the difference?

Kevin Lacker (@lacker) 's Twitter Profile Photo

The mistake here is thinking that if you prevent large companies from working on open source software, it only hurts large companies. When you stop open source software, you hurt everyone who would be using it. Startups, academics, and individuals.

Kevin Lacker (@lacker) 's Twitter Profile Photo

Good example of when prediction markets go wrong. 1. Emotional issue which brings in irrational bettors 2. High risk of “insider information”, which discourages rational bettors (even if nobody actually has insider info)

Kevin Lacker (@lacker) 's Twitter Profile Photo

Is there important but lost knowledge? Something that some segment of people used to know, or used to be able to do, that now we don’t or can’t?

Kevin Lacker (@lacker) 's Twitter Profile Photo

What if a Moore's law happened for military drones? Every few years, you get twice as many of them for the same price. Eventually there would be more drones than people. Instead of "zone defense", you'd have "man to man". Each person with a drone tracking them.

Kevin Lacker (@lacker) 's Twitter Profile Photo

Which would be a better political system: A. Two random Americans are chosen as presidential candidates. Then people vote like they do now, to decide the winner. B. The Republican and Democratic candidates are picked like they are now. Then the winner is decided by a coin flip.

Kevin Lacker (@lacker) 's Twitter Profile Photo

Reading a book translated from French, I was first puzzled and then entertained to read the word “gigaoctet”

Reading a book translated from French, I was first puzzled and then entertained to read the word “gigaoctet”
Kevin Lacker (@lacker) 's Twitter Profile Photo

Reading about the 1980 elections, it seems clear that we’re still missing something. How exactly did the race go from tossup to Reagan landslide? I wish we had prediction markets back then, that could now offer us more historical clarity.

Kevin Lacker (@lacker) 's Twitter Profile Photo

The o1 model is evidence against the "bitter lesson". It's better at some things than others, and is not simply a scaling up of GPT-4. The right idea is the "boring lesson" - scaling up is good, but better algorithms are also good, and the right mix depends on the problem.