Also a significant number of palantir senior leadership moved over to Andruil years ago. I would imagine this “fellowship” between the two companies has been in place for quite some time already
I bet users won't pay for the power, but for a guarantee of access! I always hear about people running out of compute time for ChatGPT. Obvious answer is charge more for a higher quality service.
They are red states, but without the blue state tastes that might pull the state budget in other directions. (I don't know anything about the budgets of the states of Colorado or Kansas or Wyoming).
The trees move in the wind when you zoom in. Factorio takes sprite graphics to the next level! I felt the same as you, I've always wanted a brighter reskin of Factorio but I've realized it's way tooooo much work for anyone to do for free...
I think of writing as similar to a linear extension of a partial order. Your brain doesn't think a single letter at a time, instead, all of your neurons are doing neuron things all at the same time. But writing is linear. This forces order and I think is partially responsible for the "clear thinking" ascribed to writing!
Hopefully I'll live the couple of decades to find out if PG's prediction is correct, I would bet against it.
Here however, I do agree with his articulation -- "writing is thinking" -- and like you, I've thought a bit about the linear nature of writing.
My view is that the "jumble" of ideas/concepts/perspectives is just that -- a jumbled mess -- and the process of linearizing that mess requires certain cognitive aspects that we (humans) generally consider as constituting intelligence. IMO, the rapid generation of grammatically-correct + coherent linear sequences by LLMs is one reason some folks ascribe "intelligence" to them.
I liked his analogy about how the disappearance of widespread physical work meant that one now had to intentionally invest Time and Effort (at the gym) to maintain physical health. The facile nature of LLMs' "spitting out a linear sequence of words" will mean fewer and fewer people will continue to exercise the mental muscles to do that linearization on their own (unassisted by AI), and consequently, will experience widespread atrophy thereof.
As someone working on linear extensions of partial orders (some of the time), I found your observation very insightful, a perspective I haven't considered before.
To add to this, when I think of ordering I’m reminded of the NP complete traveling salesman problem. It’s easy to make a program to visit all locations, but optimal order is so much harder.
I suspect thinking is similar, which brings up questions about LLMs as well. We all can now quickly write hundreds of generic business plans, but knowing what to focus on first is still the hard part.
