

Exactly, the one big issue with the modern world is the algorithms pushing for engagement as the only important metric.
Exactly, the one big issue with the modern world is the algorithms pushing for engagement as the only important metric.
This is spot on. The issue with any system is that people don’t pay attention to the incentives.
When a surgeon earns more if he does more surgeries with no downside, most surgeons in that system will obviously push for surgeries that aren’t necessary. How to balance incentives should be the main focus on any system that we’re part of.
You can pretty much understand someone else’s behavior by looking at what they’re gaining or what problem they’re avoiding by doing what they’re doing.
It makes sense, and in the long term even LLMs will need more efficient hardware if it will stick around in affordable prices.
It reminded me of an older writing about it:
Open source doesn’t make money because it isn’t designed to make money
You’re allowed, but as long as anyone else can do it for free, you can’t build a business model on selling it. At most you can sell something else (support, cloud compute, some solution that makes using it easier etc.).
More processing might result in more water use, but storage? It makes no actual sense. Having more stuff sitting in your storage isn’t making your computer hotter.
In fact, I guess creating a guideline for servers to use more efficient processors would do much more for that.
Which means in the long run the cost will get down to 0.
The moment the code and redistribution rights are out in the open, anyone who tries to charge for it faces competition from people charging less — and eventually from people charging nothing. The economic pressure pushes the price down to the cost of copying, which in the digital world is effectively zero.