I’ve been in the industry a long time. Even before LLMs, I think web design would get stuck in cycles. So I just want it to spit out something that looks novel and interesting, which pushes me to coach it even better.
the biggest lever is discard strategy. medium and hard bots should track which cards you pick up from the discard pile and avoid feeding cards near those ranks or suits, while dumping high-value deadwood early. they should also judge each draw by how much deadwood it actually removes, and whether it sets up a knock, instead of just counting current melds. the hard bot can guess your hand from your pickups and discards to play defensively. if they're still losing to easy, the bug is probably in the scoring function itself. have it log win-rate per rule so you can see which one is dragging it down.
I am a runner and have a standing desk. When I run, my mind is more on than at the computer. These days when I run I mentally compose prompts for the LLM when I return to my computer. So beware the illusion that simply walking away is inherently, and unintentionally, meditative. Likewise at my standing desk, the physicality of standing turns all at-desk time into an almost combative wrestling match with my tasks. Just sharing… some optimizations from 15 years of life hacking but still can’t escape the deeper psyche stuff.
Recently in La Jolla, where presumably you have high wealth and IQs and tourism, and experienced the same thing—mostly fidget stuff, puzzles, and Lego resellers. The kids were so excited to go in but pretty quickly realized it’s effectively just a physical Amazon home page of mediocre plastic stuff.
It took me a few vids to warm up to his presentation style and the production style, but now anyone else doing coffee vids seems like a total amateur/influencer/YT rusher. Early Hoffman coffee competitions were the proof point for me that he’s not simply an influencer entrant.
His presentation still is still a bit forced and orchestrated to me, but at least I believe he’s smart and has interest in the craft.
totally agree, and fwiw, nothing in this implementation requires that the agent verifies it themselves. the hope is something that ultimately exists as a verification mechanism on one side of an agent-to-agent interaction/delegation.
because it is true that, even though we've got some adversarial aspects built into the verification, that's not truly blind from the actor (unless you explicitly design the use of these in that way, which is what I've considered as the better design)