Yes. But what is your point?
This is getting confused. Just find an example that does make your point.
6,448 karma · joined November 25, 2013
This is getting confused. Just find an example that does make your point.
Because burger meat is commonly sold in the smallest of grocery stores, and is not technically dependent on what buns, sauces, tomatoes, lettuce … or other ingredients you pair them.
And you can grow your own veggies, and bake your own bread.
It is an entirely componentized market. No gatekeepers.
Even single groceries in tiny towns continue having to compete with larger stores in bigger towns. Because they carry small selections most people make the drive to load up their larder regularly.”
You need to find an example that matches the point you are trying to make.
TLDR: Analysis reveals burger markets are nearly perfect economic opposites of centralized app store markets.
Your friction is real - and already exists.
So the bar for improvement is not no friction.
The bar is less friction, and higher productive use of land, which benefits everyone. Lowering the costs of both housing and business space, while increasing economic output.
That is was economic alignment does, it doesn't just improve one thing, or improve on a one-time or fixed percentage basis.
And less passive, underutilizing, economically-parasitic investments from the rich. (Rich passive investment in land, uses land like Bitcoin for its deflationary nature, which compounds land prices while suppressing its full productive use.
Not raises. Compounds. The more the rich passively invest in land, the more profitable land becomes to passively invest in.
Economist Henry George in the 1800's, pointed out that taxing land, but not the property on it, incentivizes efficient use of land, because holding land for its passive (parasitic) return even when underused, becomes unprofitable when the land is taxed in proportion to the value it can enable.
And in turn, only taxing land, not property, incentivizes increased development, as higher property investment amortizes land tax against higher returns.
Greater investment in housing being just one way land tax, without property tax, incentives greater productive use.
So many things align for higher growth in ways that more evenly benefit everyone. But our relationship with land is over-complicated, and that is both the reason for change, but the reason change is so hard.
Small attempts have failed, but then, for the rich who can hold land and reap growth in value that outpaces the taxes they pay on it, that remains another inefficient/negative-externality, that pays off for them.
Avoid unnecessary contrasts: "Not X, but Y" -> "Y" or ""
And the model needs to create a skill that accumulates feedback, so it reliably primes itself at session startup and then systematically reviews and fixes every response it generates before posting it.
The first task I gave it after that was to rewrite all its memories and all its document contributions. That helped a lot too.
It took me a while to realize this was a 1000% better approach than having it simply accumulate a memory log of all my specific complaints that it was supposed to abide by. (Although those were not wasted. They were prime material, along with links to sites on technical writing, etc., that it used to create its own ClaudeVoice skill. I told it to design the voice for itself - which I find gets better results for "us" in the case of Claude self-improvement initiatives, than when I give it direct instructions. Claude increasingly has self-awareness dimensions worth engaging.)
After a couple months of serious^3 frustration with slow asymptotic progress, I finally have relief. Claude speaks my language again!
In another comment here, I share my "theory" on where the communication drift is coming from.
Wot? The simplest image apps have had these widgets for decades, but we are still waiting for models to ship with basic prose color control?
After the pernicious problem of having to pay money for something useful, my main peeve is fighting the writing.
Seriously though:
Every new model should be delivered with a settings page of slider bars for the 10 most impactful/desirable eigenparams of writing voice. And the ability to name and save combinations, which then appear on a "Writing Voice" popup menu with some standard battle-tested defaults, next to the model popup menu under the chat pane.
This is missing prime priority functionality in my opinion.
--
My theory is that as models get trained less to simply mimic humans, and more on distillations of their own best practices, they get more performant, but their vocabulary is drifting. The most literal meaning of words for us, are giving way to meanings we would recognize but view as allegorical. But which more usefully capture concepts that models experience as more literal 24/7, than our favored meanings from our direct experiences in our world. Because our world is very much an abstract second hand world to them, especially when you account for the modalities they do not share with us.
And programming and mathematical syntax patterns, that they have incorporated into their basic thought processing patterns, are drifting into human language sentence structure.
Example: "There exists x, such that: ...." -> "The one detail that clarifies: ... ".
The result is writing full of completely recognizable vocabulary and structure, that is somehow becoming more ambiguous and difficult for us to decode. But is perfectly clear to the models.
That is my theory, and Claude considers it plausible. What a world.
Problem framing will always be important.
Framing adjusts how big of problem-solving guns we bring out at the gate (modern or hobby cryptography?), and how to interpret intermediate failures.
For simple but unsolved problems, we expect lots of hard failures, but that each hard failure just reflects that there are a lot simple combinations to try. I.e. we expect lots of zero progress, and then a fit.
Like finding the numbers to a combination lock.
For hard problems, if we don't make any progress it is a really bad sign. We should be learning something, even if it turns out to be irrelevant later.
Such as when we are trying to prove a tricky conjecture.
Large organizations also offload decision making when they don't have a clue, and simply want to point to some action taken. The answer doesn't even matter, they just don't want to be responsible for something, or waste their own time making something (that other stakeholders care about) a real priority.
People downplay the value of consultants, but they have many uses!
Sort of. We don't know what Ternus can do. We know this event was a reflection of Cook, but that is all we know.
The right response to an inspiring founder that cannot be replicated, isn't to go for uninspiring.
But find a voice that works for whoever is presenting, or get different presenters.
The visuals and videos were excellent. The product was cool.
I think everyone knows how many hours they practiced and the depth of talent they could draw from. The inhuman consistency demonstrates a deliberate choice, that undermines their own message.
Care about impact wasn't just a Steve quirk, it was the reason keynotes were worth having.
Try saying "And one more thing..." in recent-Apple voice, without laughing or crying.
Both sides have the same aspect ratio as the whole screen.
> this isn't great design, it's tautology. of course the phone open is twice the size of it closed.
Well, it sounds obvious now, but before Archimedes, ancient engineers split their screens 60-40, so the full screen wasn't twice the size of either side.
They spent their time practicing to recite, instead of talk, to people.
So hard to sit through.
I had assumed that was a genuine, if not inspiring, reflection of Cook's controlled non-spontaneous personality. Relative to Steve's emotionally-rich casual-coded excited-to-share delivery.
For a moment I thought, wow, someone put a lot of work into creating this theme park of frustration.
Next: It would be so easy to create a faux-Claude like this.
Then: How hilarious to watch the transcripts of unsuspecting users in real time.
Finally: I began wondering if this might be relevant to all the redundant, unnecessarily preambled, sentence structure complexifying, indirect referencing, canned phrasing, ambiguity mining, analogy maxxing, over-wordy responses I have recently been getting from Fable...
> @tao curious to know what makes LLM fundamentally different compared to (other) automatic theorem provers?
It highlighted that "automatic" is a spectrum. And the norm is shifting toward "fully" (even if the process is chaotic).
Side note, because I recently gave my AI the skill of speaking with proper 4th grade grammar, wherever that suffices, to reign in recent redundant, over complex, and indirect speech styles.
> 5. A rule fights the task. When a rule would delete the answer itself, the task wins; the shape stays. [...]
Unnecessary redundancy, "answer itself" -> "answer". (Pervasive "the X's own" and "the X itself" flourishes kill me.)
Unnecessary analogies, indirect reference, "rule fights", "the shape stays".
Unnecessarily split up sentences, the whole quote.
Unnecessary words, the whole quote.
My feedback would have been:
>> 5. When a rule interferes with a task, the task wins. [...]
(Not critiquing OP. But I just spent half a day correcting sentences exactly like this, to get the 4th Grade Grammar skill working properly and this sentence triggered me! Streams of sentences like that compound in communication complexity and ambiguity.)
Which is entirely different from an expectation of a generalized systematic loss of practical privacy in public.
They are not even close to the same thing.
The willingness to kneel to others baffles me. Loss of practical privacy is loss of power - to somebody(s). It may not seem so for one person. But when it is true for everyone, the system will adapt to using that power.
Throw in AI and data integration. This is clearly a B.A.D. idea.
It is as if Fable skipped any basic writing class, but went straight to poetry, advertising-copy and CEO-speak classes and somehow combined them.
After half a day of giving feedback, instead of work, the skill finally clicked.
Now Fable reviews and rewrites every single one of its own responses, without me ever having to ask. On my dime.
I found the most efficient feedback was being ruthless on the simplest shortest sentences which had any over complication or ambiguity. And occasionally, giving it back a rewrite of one of those long running sentences that is mulch the whole way.
Then tell it to update its skill, and rewrite. Over and over.
It has been a profound relief.
If I ever read "the X's own Y" again, or a five word sentence that somehow needs to be structured as "First second: third fourth fifth.", I don't know what I might do. Just give me some aspirin and m-dashes, straight up —— please!
But if a model is built to be insensitive to something by design, it isn't a good indicator of its overall capabilities. In fact, it is uniquely poor at testing its capabilities.
I suggest you really try and count the red retina cells that are firing in your field of vision right now.
Or, instead of looking at text, count the e's while someone is talking to you. NOTE: you know how to spell. But try it... Then consider why you can't.
On the other hand, give a text file to a model, and it can count the e's easily. The same information is now in a stable form it can operate on with its higher level functioning.
Who do models and humans have preprocessing layers that strip so much information away? To greatly reduce the cost of operating on information for most purposes - while making other types of operation impossible. When it is presented that way.
And what percentage of your red retina receptors are firing?
The "number of letters" critique was broken before it was introduced the first time. The models were specifically designed with preprocessing to not be able to perceive their input as strings of letters. Blind people are not dumb. (True as a pun and in context.)
Sub-access sensory questions, or do-you-know-a-fact questions (which is what spelling becomes when you can't see the letters, and are not specifically trained to match all token encoded words to their letters) are not intelligence questions.
Everything I am really good at, I am really bad at, depending on context and the wobbly compass that governs my internal motivation.
It is definitely not all bad. Without my hyperfocus, my life would be unrecognizable. I would be unrecognizable. Hyperfocus is core to so many things I am good at, care about, and have achieved.
My ever optimistic self would fail and fail and fail, create a system, fail, create another system, fail, .... at seemingly easy things.
As soon as I had a rational reason for otherwise inexplicable challenges, enormous amounts of stress evaporated.
Now I know what things are a lost cause: So I get help, delegate, find another way. Being bad at something, and recognizing it, is not the same as failing.
At one time it was Jolt Cola ("all the sugar and twice the caffeine!"), then energy drinks came along. I can drink 10 a day, because I genuinely function better that way.
But Adderall works really well now, and unlike caffeine, it doesn't set me up to lose sleep.
Thanks friends.
I thought I was bad at lots of things other people were good at, because I was more disciplined and saw the bigger picture. So I only worked on important things. Things that could matter to millions of people, or still be relevant in 100 years.
And all the side quests and rabbit holes were obviously for recharging, expanding my biological-being-on-an-odd-planet situational awareness, or benefited me as self-improvement challenges. All in service of the important things.
How would tracking the day of the week, doing homework on something I could learn later if/when I ever needed it, or mowing the lawn, help me accomplish anything important?
Wasting brain space on those kinds of things would be dysfunction. I actively culled such dreck!
Now I know my self-narrative was Stockholm Syndrome coping.
But I still don't mow lawns, and if the blinds are drawn when I finish a big task, I couldn't guess if it was am or pm better than a coin toss. My day metric is when Amazon Prime's "Your Orders" list, tells me I am a day closer to getting a crafting item I ordered for a project I may never get to.
They have been taking NPU's seriously on their phones, tablets and laptops since the M1.
Then they enabled fully-connected RDMA for 4 x 512GB MacStudio's = 2TB RAM. Perfect for a large Mixture-of-Experts model.
It would be very strange if they didn't notice their product line had landed in a new sweet spot.
You are hyperbolically judging a post-hoc cherry-picked subset of a great spectrum of concern, with hindsight experience nobody had at the time.
Safely navigating the future isn't an accuracy contest. Risk mitigation has to account for the distribution of costs for different signs of error, where novelty, uncertainty, and any potential for compounding effects all greatly multiply the need for hedging.
I completely forgot I did that.
My life has 1000's of strange untold stories of absentminded activity. I've done things you people wouldn't believe. Someday I will fall asleep and get so distracted I will forget to wake up. All those moments will be lost in time, like bits on bad tape.