What you're describing is a failure of the government (and 'the market') to price our taxes and externalities correctly.

> So, the economically sensible thing is spend $100 to make factory B more efficient, instead of spending $10000 to undo the effect of factory A

This is where my understanding breaks down. I don't think this is true at all. I think the opposite is true - it's economically sensible to remove the pollution immediately from the first factory, not to (conditionally promise to) reduce the future output of another factory.

I don't get it. The planet is heating up and each degree we go is going to cost untold trillions. Saying it is economically sensible to try to reduce future output over not making new output at all is the part I don't get. It's just not true that you can choose to pollute more now and promise that it'll be a reduction in pollution output later and come out ahead. You will come out behind on this. The climate is changing faster than we are able to clean it up.

> So the system can work, as long as everybody's honest (or as long as there's some system that verifies that they are honest).

There is no such system. And in the US, the current Republican government is dismantling all oversight that may have existed or could exist to help this.

Another example: Exxon knew about the damages their products were doing to our health, but they lied about it for 50 years.

So I stand by my points. This idea does not work, and Cloudflare is taking positive credit for "reducing their carbon output" but I think it's just a lie. It simply cannot be true.

It kind of depends on what your target is. If your target is absolutely zero emissions, you do have to turn off all the factories, regardless of cost. If you have a target with some emissions allowed, it makes sense to first reduce the emissions where the cost per unit reduced is least -- and ensure somehow that emissions don't just get moved outside of the measurement regime.

If your target is absolutely zero emissions, you also have to ban volcanos, limestone erosion, permafrost melting, cows, humans, and really all animal life on earth. There are plenty of natural processes that result in CO2 being released into the air, not least of which is breathing.

Grandparent post has a "perfect is the enemy of the good" fallacy. Banning all emissions is simply not possible. That doesn't mean that efforts to limit emissions aren't worthwhile. And that's what cap & trade programs do; the government sets a cap on carbon emissions based on the best available science for what the planet can tolerate before we get unacceptable levels of warming, and then companies can trade carbon credits amongst themselves to allocate who reduces their emissions and who pays other people to reduce their emissions.

As far as I know, cap-and-trade programs were successful in reducing SO2/NOx emissions.

https://en.wikipedia.org/wiki/Acid_Rain_Program

> Since the 1990s, SO2 emissions have dropped 40%, and according to the Pacific Research Institute, acid rain levels have dropped 65% since 1976.

...although it seems debatable whether it is more effective than plain regulation.

> However, although it reduced emissions by 40%, the US Acid Rain Program has not reduced SO2 emissions as much as the conventional regulation applied in the European Union (EU), which reduced SO2 emissions by more than 70%. Therefore, the effectiveness of the emissions trading element as a mechanism has been criticised, ...

Would you rather spend the 10,000 to capture 1 ton of carbon or instead pay 10 factories to reduce their output 10 tons each.

We can make it a concrete example.

Build a nuclear plant first.

Alternative 1: use the nuke plant to do carbon capture from the atmosphere while next door ten coal plants run and pollute as before.

Alternative 2: produce electricity so all the ten coal plants stay off.

The latter approach has about 10x the CO2 reduction effect.