> We build 8 core CPUs and blow e-fuses in the factory to make them 4 cores. As resources dwindle and the planet fills with e-waste this disgusting, unethical wanton destruction of value continues because we've normalised it.

This is a bad follow-on example because artificial price differentiation is sometimes what it takes to make a business viable. Making an actually worse model in the processor case has far more fixed costs than re-using the existing pipeline. The business wouldn’t be viable selling all of the processors at lower nor would it be as strong leaving out the lower income segment.

The alternative world is you get no affordable processor at all and the 8-core version costs 200% more due to the volume lost.

Finally, this ignores that processors selected for lower tiers can be chosen precisely because they didn’t meet the bar for the high performance batch. So if they had proceeded to treat it as an 8-core it would have had thermal issues and an 80% reduced life.

In your scene you would see increased buying and volume at the higher price point which would offset the money lost at the lower tier.

> The alternative world is you get no affordable processor at all and the 8-core version costs 200% more due to the volume lost.

No, you are making up a false dichotomy. There are many alternative worlds, including the ones where CPUs have no proprietary "intellectual property".

> artificial price differentiation is sometimes what it takes to make a business viable

Sometimes wage theft is what it takes to make a business viable, so I don't really buy this as an argument.

> processors selected for lower tiers can be chosen precisely because they didn’t meet the bar for the high performance batch

This is a better argument

> The alternative world is you get no affordable processor at all

You probably do, just maybe their performance doesn't double every year. I would say that's probably an interesting tradeoff to make, depending on how catastrophic one thinks the state of the world is today.

If binning of CPUs were disallowed, it would result in even more e-waste because any imperfect chips would have to be tossed out instead of sold with fewer cores enabled. Bad example.

binning != intentionally crippling working silicon to meet quotas of binned parts

>Welcome to modern tech, where the management, perception, control and tactical destruction of real value is the only place left to eek out a profit margin.

It's almost as if there's a consistent economic process that leads companies to do this kind of thing again and again. We could call it "the tendency of the rate of profit to fall".

Consistency may be running out. It may be time to turn this elegant observation (not immutable law) into an anti-pattern. As an optimist I prefer "tendency of systems toward failure of innovation" (AKA sloth of the entrenched incumbents). This is the last crisis. Because it's not a crisis of capital, it's a crisis of the very substrate (the planet) which allows for the possibility of capital. Schumpeter's creative destruction is now literally environmental destruction. A new broom is required. Let's hope it's not the communists again. A few simple bits of regulation could go a long way to correct things.

> We build 8 core CPUs and blow e-fuses in the factory to make them 4 cores.

Not "deliberately". We build 8-core CPUs, discover four of them are a bit sub-spec, and blow e-fuses to disable those and sell it as 4-core. It's possible that those four disabled cores would be just fine, but I'm not selling you a chip that gets its sums wrong sometimes.

I guess people complained about how chip manufacturers sold CPUs rated at 25MHz that could be overclocked to 33MHz, but they too were not guaranteed to meet spec even if they worked.

Is it really the case that dies/packages are tested and batched individually?

I mean, you make a good point on reliability/quality, but there are surely cases where to create a product differential perfectly good (and known to be good) devices are crippled.

That seems ever less ethically defensible as we move into scarcity.

Depends on the value. If you test half a dozen off a die and they're all sub-spec, they're all binned the same even if some are okay.

This is why overclocking used to work - if even a small proportion of the chips tested were good to 25MHz, that was a 25MHz batch, guaranteed good to 25MHz. If you got a "lucky penny" that'd go to 33MHz as per the design spec, then good for you! But it wouldn't be guaranteed to do that.