jimmont

This explains all the things that are happening now and will happen in the performance changes, this is just the beginning. The reduced quality yields more retries and use of other models and all the outcomes and products yielding more token use. It's the road show now and metrics must show a trend for liquidity and perceived market performance if they want the exits to work. Anthropic and OpenAI apparently are 48% of Google Cloud revenue alone so if this doesn't work out in the window of time that any liquidity in the markets remains, it doesn't work out so well. And the product performance is just a footnote when the differentiation is shrinking by the week and it's the end of the road that isn't going to make it to where it must go (apparently we're told it's 2028, if that works out I'd love to know why and how). It'll be an interesting 2027, maybe Q4 as well. I don't know any time in history that a massive compute buildout wasn't massively depreciating at year 3 and dominance never lasted much past 10 years—and we appear to be at year 3 now if my read is correct.

Clearly that article shipped to production without QA. Having no testable definition of good, it must be bad. Fired.

That's an excellent idea I plan to try, thanks—re using structured JSON with schema. The most success I've had is saying "be brief" or an explicit size, like one line, or do not explain, etc. I haven't measured other instructions so extensively. They do work but the more specific the better. Other strategies around outputs that are more natural language seem to be hands-down the direction to take, and get away from the machine language habits we've used in the past. It's super interesting seeing this new practice emerging and more or less inventing parts of it along the way. Right now I'm at the place where my brute force and elaborate explanations were reaching their limit and in the frustration just realized I need to take a few days and try to figure out the tool. Across all these the pattern seems entirely that the constraints bound the probability space, whether it's the format like you suggested, or the instruction we give, including the space we point it toward (Web APIs, runtime, schema, etc). In all instances where it's not working the solution seems to be what does the pattern reduce to, and what specifics are the do/don't to go with that, and most of the time the results improve immediately. Your tip seems excellent for this. An easy-button.

Reviewing my experience using LLMs, to improve results, reduce churn and token usage. Discovering the gap between what they produce and what I'd normally do is a significant source of output cost, regressions and surfacing a bit of why and how to fix it. Notably Claude is remarkably bad at/about this, producing errors even when directed toward modern Web solutions—that cut token use a lot, like toward 90% occasionally, which together with the frustrating churn led me to review how I'm working, what is happening and generate this article.

Turn off the ability to add hyperlinks and use words instead? Is that what comments were intended for anyway? Seems a bit convoluted, but limiting comments, to comments, would be fairly natural. It'll get subverted anyway, but likely affects. A policy could say don't use hyperlinks, feel free to say where to go. Which would be an interesting policy to technically enforce, and still have the discretion to operate on.

Organizations are choosing to eliminate workers rather than amplify them with AI because they'd rather own 100% of diminished capacity than share proceeds from exponentially increased capacity. That's the rent extraction model consuming its own productive infrastructure. The Stanford study documents organizations systematically choosing inferior economic strategies because their rent-extraction frameworks cannot conceptualize workers as productive assets to amplify. This reveals that these organizations are economic rent-seekers that happen to have productive workers, not production companies that happen to extract rents. When forced to choose between preserving rent extraction structures or maximizing value creation, they preserve extraction even at the cost of destroying productive capacity. So what comes next?

I think it's just general lack of awareness of the effect, or in many instances having alternate economic incentives, like academics who want to commoditize their intellectual outputs to all available distribution channels. Tyler Cowen for example. The AI companies are in a race to the bottom.

For those that use AI/LLM's that retrain on your input, I assume you realize this commoditizes your intellectual work? And effectively makes use of it like they already used copyrighted intellectual property. This is effectively the same as the commons appropriations made for railroad development, reinterpreting fair use, etc.

It's interesting to see where the economics align with software. Web advertising paired with search and a browser. And it worked. I can't recall a better outcome supporting modern communication than Chrome and all it contains. The harmony between the business, software and market need was very well aligned. That's falling apart now due to LLM's taking over much of what previously was search, and Google's innovator's dilemma between AI/LLM and traditional search. While the revenues are stronger than ever this seems unlikely to continue long term. So the dynamic is changing naturally. Reminds me of how a strange contemporary religion where believers families take over a planet, resulted in this group building an amazing genealogy web service (familysearch.org). Or how US hegemony naturally leads to it's military, or how the theocratic monarchy of ancient Egypt was good for the masonry industry--even affecting modern tourism, or how Roman roads form the paths leading to today's major European cities.

It's a demonstration of what an entity can do, and will do. The only innovation from Apple since Job's died were unfinished efforts that completed under Cook. There will be no future innovation from Apple. There maybe acquisitions and optimizations around current and existing positions, nothing more. Apple is IBM now and has been for some time. Apple doesn't know how to develop software or deliver it to market. It has no leadership, only management, and managers are not known for relinquishing their position so I don't see any reason this will ever change. This pattern is endemic in Silicon Valley as old products have run their course. It's like the automobile from the 1900s just playing itself out. Phones and computers are a commodity I don't expect any differentiation within anymore. As to Siri and AI it's an emerging space that Apple can't even deliver on a strategic partnership for within 2 complete phone and iOS release cycles. I'm not sure how an institution like Apple can or will ever overcome its own misaligned culture. Which brings me back to it being IBM.