- m0lluskOr possibly under regulated. Where exactly is all the radioactive waste going to go? Especially the spent fuel rods pose major disposal challenges. The one site that was looking hopeful appears to have been discarded. It is a bit late in the game to be pending basic stuff that is piling up. Most nuclear power plants are not well sited for long term disposal, though that is what is happening.
- It isn't only companies, it is a mass social movement. Anyone with basic coding experience can download some basic learning apparatus and start feeding it material. The latest LLMs make it extremely easy to compose code that scrapes internet sites, so only the most minimal skills are required. Because everything is "AI" now aspiring young people are encouraged to do this in order to gain experience so they can get jobs and a careers in the new AI driven economy.
- This is the biggest red flag for me. In contrast, as Ruby and Rust were adopted they generated many tools, libraries, frameworks, and full applications which answered unmet needs in clever and interesting ways. What has generative LLMs produced? Apparently it is nearly all slop or minor iterations. In some cases there are significant bug fixes accumulating which is about as good as it gets. Still many powerful tools coming from the generative LLM wave, but probably worth waiting for best practices to emerge and pricing to stabilize.
- > To build it requires companies to invest a sum of money unlike anything in living memory.
Do we know this? Smaller more carefully curated training sets are proving to be valuable and gaining traction. It seems like the strategy of throwing huge amounts of data at LLMs is specific to companies that are attempting to dominate this space regardless of cost. It may turn out that more modest and better optimized methodologies will end up winning this race, much like WebVan flamed out taking huge amounts of investment money with them but now Instacart serves the same sector in a way that actually works robustly and profitably.
- That isn't realistic at all. Customers change and their needs change with them. Sometimes customers die. Products that stay the same change their fit over time and eventually fall away. A business that is not growing is dying. That is okay. It can be fine to let things fall away when they have run their course, but some prefer to endure. But the absolute fact remains that a business that is not growing is dying.
It seems like what you are perceiving is a common market delusion. An unfortunate fact of hiring is those workers who are not employed and satisfied are often less experienced and skilled than those who are well placed and not looking. The same logic applies the other way around to companies. Those who are looking to hire juniors who haven't yet found their way are often companies that lack a solid center and just want to squeeze some money out of whatever customers they can find using whatever tool is at hand.
With the current state of things if your needs are truly modest then there is a good chance that you can get by with some independent offering. Find something you are interested in and make it work for someone willing to pay for it. Make sure to lean more into sales and actually making things work for customers than the engineer tendency to envision mechanisms and focus entirely on that. This way you can set the balance for yourself, and I can absolutely guarantee that you will experience the realities of growth or death up close, though in a more personal way that you can take control of and manage for yourself using criteria that have meaning for you.
- It will be interesting how this goes moving forward. Agents learn from massive scraping. With the newest tools and frameworks there is nothing but documentation and initial examples to scrape. And now that agent output is flooding everything it can be expected there will be a lot of feedback with automated learning early in development cycles.
Lots of applications have a simple structure of collecting and operating data with fairly well documented business logic tying everything together. Coding outside of that is going to be more tricky.
And if agentic coding is so great then why are there so still so many awful spreadsheets that can't compete with Excel? Something isn't adding up quite as well as some seem to expect.
- > I have zero doubt that BTC will hit $1m one day.
Well, there's your problem. Here is a potentially interesting math paper that comes to very different conclusions: https://www.fooledbyrandomness.com/BTC-QF.pdf
- Strange how LLM vendors are flooding the market with reasons not to do business with them. Every paid agentic interaction contributes to all the bad behavior we are seeing. From out of control web scraping to buying up available hardware LLMs are turning out to be highly efficient misery manufacturing mechanisms.
- Turn every potentially useful development tool into some LLM hype bullshit to grow the bubble.
- Then he chickens out on Tuesday as usual, that's what.
- It seems that innovators, researchers, and founders all work at a fast pace, but adoption of new technology, especially LLMs, ends up being done by companies as is convenient. When an open position can go unfilled or a group can scale up without hiring then companies might move forward with a commitment to LLMs.
Even with strong adoption it may take many years for LLMs available now to reach their potential utility in the economy. This should moderate the outlook for future changes, but instead we have a situation where the speculative MIT study that predicted "AI" could perform 12% of the work in the economy is widely considered to not only be accurate, but inevitable in the short term. How much time is needed dramatically changes calculations of potential and what might be considered waste.
Also worth keeping in mind that the Y2K tech bust left behind excess network capacity that ended up being useful later, but the LLM boom can be expected to leave behind poorly considered data centers full of burned out chips which is a very different legacy.
- Family values.
- Labubus got extra kick from being gambling also. Many were sold in boxes without labels or with minimal labels that listed possible contents. That makes the actual product into more of a loot box kind of thing. That might have contributed to the speed of the trend passing.
- Real Estate is currently trading at the highest price to earnings ratios ever recorded.
- > We need a ...
Here's were he loses me. The problem statement is detailed, but proposed solutions need more work. There must be ways to improve the system without abandoning the original intent. There may be way to account for costs, simplify reviews, and so on. Often changing regulations to have specific goals and sunset provisions changes enforcement for the better. Sometimes basic changes like the amount of time allowed for any given step can make a huge difference.
Solving regulatory problems is as real as the engineering and marketing that make products in the first place.
- Nice summary, but in my experience with programming language design the macro usage issues loom large. What about base libraries, use of popular libraries, build tools, performance analysis, debugging, packaging and modularity, and so on. The core design matters and then cascades into all manner of differences.
- This idea that there is only one story and one path is pernicious. Paul Orfalea flunked out of school and couldn't hold a job, so instead he built a business that he later sold for around five billion. His book Copy This! is loaded with interesting and potentially useful stories. His is not the most common story but is an example of great business success being unrelated to either identified genes or familial wealth.
- Inform the user in a minimal way. Probably this means some kind of flag that can be clicked on in order to see a list of what was done and what problems that might address. If the fixes relate to unused features then they can be pended and there is no need to interrupt workflow.
- Telemetry is not mentioned. Many apps nowadays are extremely chatty and record a wide range of details of use. In theory this helps to optimize usage and discover blocking bugs, but in reality it is mostly just an annoyance that enables service providers to spy on what users are doing.
- All of the serious problems I have seen with Waymo navigation so far have had to do with busy urban streets. Trying to make use of blocked non through way alleys, turning around in driveways when other vehicles are exiting, coming to a complete dead stop on busy one way streets, failing to brake predictably for pedestrians walking into lanes, suddenly backing up a half block from stopped at a red light in order to change lanes, and so on. Freeways are a simplified driving environment that should suit current technologies well.