These days chatbots are good enough that when I do use a search engine, I really just want pure search results, I’m not interested in getting another AI opinion. I am sick and tired of getting AI overviews for a google search. What’s a better search engine? Heck, it doesn’t even have to be “better”, I’m looking for different results, not just perfect matches.
IIRC ( and it was 20 years ago now that I learnt this) the brain uses 20% of the body's resting energy usage. Most of that is keeping neurons polarised to the outside (ion pumps need ATP!!!).
The body uses 25w resting and thus the brain is about 5w.
Source: biology degree but like I said please take with the same amount of weight as a hallucinating LLM.
GPT says: Unless you're a hamster hooked up to a Fitbit, it's more like 60–70W for a normal adult human. So the brain’s real power draw is more like 15–20W, not 5W
Resting energy usage in humans is ~1200–1500 kcal/day, or about 60–70 watts, depending on the person. Logic holds, estimate is just low
Could you use the vibe coding loophole to eliminate all uncertainty: the AI has the answers you need you just need to develop by continuously prompting and reviewing until the solution is ready for production?
If you are a pure software company (no hardware or other activities) your options are rather limited.
Also, as I said in other posts, at year six you reach steady-state and are amortizing the full amount every year. Example:
amortization for each year
R&D year 1 year 2 year 3 year 4 year 5 year 6
year 1 1,000,000 100,000 200,000 200,000 200,000 200,000 100,000
year 2 1,000,000 100,000 200,000 200,000 200,000 200,000
year 3 1,000,000 100,000 200,000 200,000 200,000
year 4 1,000,000 100,000 200,000 200,000
year 5 1,000,000 100,000 200,000
year 6 1,000,000 100,000
total amortization: 100,000 300,000 500,000 700,000 900,000 1,000,000
Not ideal, of course, but if you are not a "flash in the pan" company, at year 6 it feels like this rule doesn't exist, other words, you are amortizing the full $1MM every year. The TVM on the deductions you could not take until steady-state is reached is part of the hit you take. The other is taxes on profits from operations during the early years.
Most companies don't have profits rise exponentially during the first few years, so it might not be too bad. Also, there are many ways to mitigate this. For example, section 179, which allows businesses to deduct the full purchase price of qualifying equipment and software in the year it's put into service, rather than depreciating it over several years. In other words, instead of paying taxes on your profits, use that money to buy GPU's computers, tools or whatever you might need. Easy.
A tax attorney is essential if you want to minimize tax liabilities intelligently and within the bounds of the law.
But, yes, 174 needs to go back to full annual deductions.
Be honest: do people vibe code because they simply can’t imagine all the complexity and details of what they’re trying to achieve, or is it simply because it’s faster than typing it all out?
If it’s the latter, perhaps it’s a sign that we are making languages too verbose, and there’s a lot of boilerplate patterns that could be cut down if we give ourselves wider vocabulary (syntax) to express concepts.
In the end, if we can come up with a language that is 1 to 1 with the time and effort spent to write equivalent prompts, there will be no need for vibe coding anymore unless you really don’t know what you’re doing, in which case you should develop your skills or simply not be a software engineer. Some may say this language already exists.
Given how addicted people are to using LLMs steep price hikes are almost certainly guaranteed at some point.
When this happens, what we will see is once again the rich and privileged will benefit from the use of LLMs while the poor have to just rely on their own brains. Consider how some students will have to grow up struggling through school without any LLMs while rich kids breeze their way through everything with their assistants.
If they are dividing a few billion dollars in model training between a small number of rich people, it quickly becomes too expensive even for them.
Meanwhile, a free model running locally is good enough for most people. This causes pricing pressure (and I think is probably going to bankrupt most of the AI companies).
More likely IMO is that AI becomes a loss-leader. It'll all be stuff like Grok or DeepSeek where the real profit is in censorship and propaganda.
Not sure if you're being sarcastic. The cost of compute is perpetually going lower, it is getting harder to scale though. I feel like LLM's will become ubiquitous. When I went to University in the 90's, only the wealthy could afford cell phones, pulling one out was a flex. Now they are everywhere. Even Nvidia's sky high margins will someday be eroded.
> It means you can't sustainably release free apps unless you're willing to just ignore that you're burning $100 a year forever.
The fact that this $100 yearly fee has never been adjusted for inflation, in a time when developers are easily shelling out $200 a month to use LLMs, makes this gripe all the more petty. Especially when software developers are still among some of the highest paid jobs on the market.
Back in the days of newgrounds, a bunch of kids learned to program by picking up flash, making games, and sharing them with their friends. That would not have happened if you had to pay even $1/year to publish a .swf file.
Now, their friends all have iphones, and if a kid could hack out an app for their friend group and share it around, it would give them great motivation to learn and hack in that ecosystem.
It's kids, idealistic OSS hackers, and people in the global south who see the $100/year and give up.
It's also just plain dumb rent-seeking that no other OS does. I can make an APK for a friend for free, I can make an exe for a friend, an elf for a friend, etc.
Yet, even if I live in the EU and thus can distribute apps to my friends to sideload directly, even then I'm required to pay $100/year or my app stops working.
> Now, their friends all have iphones, and if a kid could hack out an app for their friend group and share it around, it would give them great motivation to learn and hack in that ecosystem.
edit: To be clear, I'm not defending Apple here, I'm still 200% in favor of "let people run the code they want on the device they own", what I'm trying to say is just if you're a kid who wants to do this, you already can!
> a bunch of kids learned to program by picking up flash
flash does not have a free version (and i recall a trial version is locked after some days of usage). Guess where they got their flash development/program from?
The $100 isn't just required for the things published to the AppStore.
It's required for a sideload app in the EU. It's required to make a testflight app that isn't discoverable on the app store, but people can opt-in to running (though apple still heavily restricts what apps you can publish via testflight, it's not a substitute for sideloading).
> There is no reason for it to be polluted with everyone’s throwaway side projects.
Too late for that, the app store already has a huge amount of trash on it.
If apple insists that the appstore is the only way to get software on a computer in my pocket which I paid for and own, then yes, they are obligated to let me put trash on it.
If apple won't let people put anything on it (as they don't), then they should be obligated to let me install devices in another way.
It's already incredibly silly that if I want to just have my own simple app (like idk, a native app that lets me upload a photo to my personal archive over a custom protocol I use), I have to jump through hoops as if I want the whole world to use it, even if I never intend for anyone outside literally just me to touch it.
It’s a phone, not a computer. It’s not sold as a platform for developers to tinker around doing whatever they want. It’s primarily a lifestyle device and businesses can produce apps for it. That’s it. You don’t get to put your own software on everything that has a computer in it: cars, consoles, appliances, etc.
Even more of an argument for free sideloading then. Ping your friends a URL, they click it and can install your app. All without the need for an App Store.
Yes for some people who used these machines once, they might just think of them as old machines, the same way an ancient Roman still alive today might not think much of mundane Roman tech.
But getting into retro-computing as a hobby is more like being a historian or archaeologist. There is endless lore to discover, and restoring old hardware is an art. Some of these people were never old enough or even existed to lust after these machines.
Someday, all the people who used these machines will be dead, completely dead, and the machines will be all that remains. Blessed are those who keep them running in their memory.
reply