I read Dr. Seeley's Honeybee Democracy in 2021, and then again early 2023 (after using ChatGPT). It was my favorite non-fiction read 2021 (Steinbeck's East of Eden for fiction).
The last chapters touch on how simple rulesets can create incredible complexity in the overall supraorganism — it had me thinking of the robots and generative AI that will likely power their micro- & collective- thoughts.
A beautiful book overall, with and without computer comparisons:
Is this what it looks like? A specific concept like time units being defined as members of more general types like numbers? I.e. if I type `1.` to get auto-complete, am I going to see days, and all the rest, as options?? That API design pattern would be a nightmare!
We're a zero trust cloud infra solution for power users.
It solves problems like prompt injection and secrets exposure. For host security you're right cloud is the only way to secure those heavily and one of the reasons we went that route with enclave attestation.
We offer a way for you to use AI agents without the AI provider ever able to see your sensitive information while still being able to use them in a minimized permission environment.
AI has a tough time leaking your credentials if it doesn't know them!
This argument doesn’t make sense to me. Insurance companies are structurally incentivized to minimize payouts across the board. They want hospital bills lower, physician compensation lower, and patient payouts as small as possible. If insurers had unilateral power, total medical spending would collapse, not explode.
The real source of high medical costs is the entity that sets the hospital bill in the first place.
The explanation is much simpler than people want to admit, but emotionally uncomfortable: doctors and hospitals are paid more than the free market would otherwise justify. We hesitate to say this because they save lives, and we instinctively conflate moral worth with economic compensation. But markets don’t work that way.
Economics does not reward people based on what they “deserve.” It rewards scarcity. And physician labor is artificially scarce.
The supply of doctors is deliberately constrained. We are not operating in a free market here. Entry into the profession is made far more restrictive than is strictly necessary, not purely for safety, but to protect incumbents. This is classic supply-side restriction behavior, bordering on cartel dynamics.
We see similar behavior in law, but medicine is more insidious. Because medical practice genuinely requires guardrails to prevent harm and quackery, credentialing is non-negotiable. That necessity makes it uniquely easy to smuggle in protectionism under the banner of “safety.”
The result is predictable: restricted supply, elevated wages, and persistently high medical costs. The problem isn’t mysterious, and it isn’t insurance companies. It’s a supply bottleneck created and defended by the profession itself.
I built this for a product planning tool (https://plan.fyi) I have been working on where I wanted users to define timelines using fuzzy language. My initial instinct was to integrate an LLM and call it a day, but I ended up building a library instead.
Existing date parsers are great at finding dates in text, but I needed something built for extracting context, understanding business time (EOD, COB, business days), parsing durations, and handling fuzzy periods like Q1 or "early january" or "Jan to Mar" etc.
It returns typed results (date, duration, span, or fuzzy period) and has an extract() function for pulling multiple time expressions from a single string - useful for parsing meeting notes or project plans.
LLMs by their very nature subsume software products (and services). LLM vendors are actually quite restrained - the models are close to being able to destroy the entire software industry (and I believe they will, eventually). However, at the moment, it's much more convenient to let the status quo continue, and just milk the entire industry via paid APIs and subscriptions, rather than compete with it across the board. Not to mention, there are laws that would kick in at this point.
> Have you really never voted for a candidate who went on to do things you didn't agree with?
I personally haven’t, no. And I definitely have never voted for a candidate who claimed they’d do horrible things if elected. There is no one who voted for Trump the second time around who has the excuse of him “doing things I don’t agree with”. He told everyone what he was going to do, and people still voted for him. Either they agree with his actions or they’re stupid. If the former, they’re irredeemable; if the latter, they need to take responsibility and act. This was not an “oopsie” you can simply regret and vote better next time, irreparable damage is being caused to the world.
Getting more thinking time tends to give surprisingly small improvements to playing strength. For a classical alpha-beta search based engine, for a given ply (turn) you might have ~20 moves to consider each depth of the search tree. If you're trying to brute force search deeper, a 10x increase in compute time or power doesn't even let you search an extra ply.
Elo gains for engines tend to come from better evaluation, better pruning, and better search heuristics. That's not to say that longer search time or a stronger CPU doesn't help, it just doesn't magically make a weak engine into a strong engine.
Pure speculation, but I’d guess that an arrangement with Google comes with all sorts of ancillary support that will help things go smoothly: managed fine tuning/post-training, access to updated models as they become available, safety/content-related guarantees, reliability/availability terms so the whole thing doesn’t fall flat on launch day etc.
I stated that I am not naive and am not entirely convinced by Apple's sales pitch that the Private Cloud Compute containers are encrypted with keys in a way that only your hardware device can read in such a way that the PCC is an extension of your device.
I just think it is useful that Apple is trying something along those lines and wishful the guarantees work half as well as they claim they do, because that's a good goal to have in theory even when it fails in practice against dedicated threat actors.
And yes, to be fair my personal day-to-day threat model currently is much more concerned with the evil advertising company known as Google than it is with government actors. Even if Apple's Private Cloud Compute only means "private from Google" that's still a win for me (and most of the information I was looking for when I saw this headline, because my first fear was that the advertising company Google was involved).
The only winning strategy: get him to sell nukes to the other side. He'll bullshit them out of billions for years and fail to deliver. See also the Boring Company, "Full Self Driving", and the Mars mission.
If (as you seem to be suggesting) relativity was effectively lying there on the table waiting for Einstein to just pick it up, how come it blindsided the greatest minds of his generation?
There has to be a way to set permissions right? The demo video they provided doesn't even need permission to read file contents, just read the file titles and sort them into folders based on that. It would be a win-win anyways, less tokens going into Claude -> lower bill for customer, more privacy, and more compute available to Anthropic to process more heavy workloads.
You can use all the math stuff like scientific notation, tetration, etc... but it won't help you make things smaller.
Math notation is a form of compression. 10^9 is 1000000000, compressed. But the offset into pi is effectively a random number, and you can't compress random numbers no matter what technique you use, including math notation.
This can be formalized and mathematically proven. The only thing wrong here is that pi is not a random number, but unless you are dealing with circles, it looks a lot like it, so while unproven, I think it is a reasonable shortcut.
I have done this specifically with the second item in the list in the OP.
Not only did I do it to get free shipping, I got it to get free international shipping.
For extra bonus CO2 points, the other item was coming from a different country.
So I basically paid $0.42 to have a single packet of kool-aid shipped across the pacific ocean.
(I'd never had kool-aid before and I must say I was disappointed.)
> In very simplified terms, the medication works by waking up the brain stem, preventing full muscle relaxation in the throat, while allowing the brain itself to rest during sleep.
I wonder if this is the same level of restorative rest one would get without keeping the brain stem awake?
> The camera is clean. I can see that on the dashcam records.
You won't see it unless you shine light into it.
> And if the system is so fragile that a bit of dust kills it, then it's not good.
It's not dust, it's fog on the inside of the windshield from offgassing.
> The issue with the red-hands-of-death is caused by the forward collision warning, the road there curves and slopes up, so the car gets confused and interprets the car in front as if it's on a collision course
Of fair enough. I've never seen this, and I used FSD (14) all through the Appalachian mountains.
> FSD got better than it was 4 years ago. But it's still _nowhere_ near Waymo
Fair enough, but FSD is still years ahead of any other system you can buy as a consumer.
Companies were still figuring out Linux with servers at the time. Xserve seemed like it might be something of interest to at least academia but Apple never really had their heart in it as I wrote at the time.
I don't understand. On average, for every 4 input bits we will get it right 3 times writing 0.5 bits each time and get it wrong once writing 2.4 bits once. So we write a total of 3 * 0.5 + 2.4 bits = 3.9 bits. The compressed output is 3.9/4 = 97.5% as big as the input. Not very compelling. What am I misunderstanding?
It is reported that the ICE agent who fired the deadly shots already experienced a harrowing situation with a car . . . He was seriously injured last summer when he was dragged by the vehicle of a fleeing suspect whom he shot with a stun gun.
Apparently the driver rolled up the driver's side window on his arm and dragged him.
