Unless you actually expect your environment to spend more than 5% of its time on average getting suspended and resumed. Maybe you work on some tooling that is scheduled to run right around migration time (for servers) or sleep/resume time (for clients).

The entire point is that monopolization confers the same advantage that this scheme does.

2. Clearly, _you_ don't know the API, as you can get output up to the total context length of any of the GPT-4 32k models. I've received output up to 16k tokens from gpt-4-32k-0613.

3. I am currently violating my own principle of avoiding correcting stupid people on the Internet, which is a Sisyphean task. At least make the best of what I am communicating to you here.

Your mindset would mean that Windows would have next to no backwards compatibility, for instance.

Why?

Optimizing for engagement is not inherently bad, nor does it necessarily result in socially suboptimal outcomes. My TikTok feed is very engaging without having to resort to triggering my anger.

A recommendation algorithm that only sticks to a handful of given topics (rage bait and furry porn?) is not a very good one.

It also involves demanding access to proprietary ICs and information like schematics. A component level repair might become impossible if you don't have access to a vendor-specific replacement for some burnt battery charging IC. You can't really fix up a silicon die like you can a dead pixel.