I spent 4 hours on the phone to an airline this morning to change a small detail. I was passed between the same team three times until I got someone good who resolved the issue.

My thoughts on service industry post-GPt are…

LLM with api can replace 95% of the people I talked to.

Businesses should keep the one good person I spoke to to deal with the edge cases (everyone in their team probably knows who it is).

All the people not good at their job will likely be good at something. They need to find out what that is. It’s just going to be an unprecedented disruption to labour market until we can get it right. How do we support those affected?

Baysean inference is based on probability. It’s hard to be 100% certain of anything so you weight EVERYTHING you can think of on both sides of the issue. This is different to the normal scientific publications (with some exceptions) but often drives scientific thinking and consensus but isn’t communicated well. Sean Carroll has some good podcasts about it.

It allows for opinion, skeptics, qualitative data, popular belief can also be included but you need to be careful about bias. You hold all beliefs as potentially true but try and weight them. Weight can be based on, for example -how qualified the person is to make the claim (so CNN and FOX could be low) and how biased they might be (CNN and FOX -not doing well here).

You add up all the weights and it gives you a probability of something being true or false. You then update the probability as new evidence comes in. The important bit is you can include anything - you just need to give it an agreed weight. Even if people completely disagree it’s a helpful way to open discussions as the assumption starts that all information adds value.

For example flat earth (numbers are just for fun)…

I look around me and it looks flat (99.999999% weighted flat)

I bump into someone with a telescope and say they think the earth is round (I still think my senses are correct but they seam convinced and the sun/moon are round so maybe I give it a little weight - 99.99999% flat)

I read that a lot of people think the earth is round (ok sheep -99.9999% flat)

I see a photo from the moon that the earth is round (fake I don’t trust the government I’m increasing my weight now -99.999999% flat)

I start using gps and a lot of really smart people are making cool stuff I like. They say it only works if the earth is round (hmmm - I think - how can they get this tech so good but get that so wrong - it still looks flat 99% flat)

I see a YouTube video of someone who presents a flat earth theory that lots of people have liked and my Twitter feed confirms a lot of people hold this view (back to 99.99% flat)

I start to learn some physics and i see how some formulas can predict motion and explain a “theory” of how it the earth might be round (still looks flat but ok I’ll go 95% flat)

I go up in an airplane and actually the horizon is kinda curved (90% flat)

I go in a space ship fly around the earth land on the moon - look back at the earth and update my probability (0.0001% flat).

GPT5 unifies quantum and general relativity and points out that the 4th dimension (time) just gives the illusion of curved space and that actually the universe is all flat (99.999999% flat)

Note it’s never 0%. The probability can go up and down. It’s also not about being right or wrong - just trying to agree on the weighting.

What you do with the probability then is up for debate but at least there is some agreement on the problem. For climate it might be helpful to think about:

1. Low probability + low risk = don’t sweat

2. High probability + low risk = we can deal with it

3. Low probability + high risk = better do something -just in case

4. High probability + high risk = better act fast

You can do the math but it’s actually something you can do in your head. What people usually find hard (me inc) is weighting esp when considering my own bias and I tend to overweight other people that agree with my stance. So you need to talk to people who you don’t agree with. You need to not fully trust anyone -including yourself.

Hey - I missed this post. I can’t get Damus to show a decent time line. I think we agree on this. Have you seen Adam Curis documentary on hypernormalisation and the fall of the Soviet Union?

Let me know how you get on. I’d like to try Linux but never really had a reason to. Do you mind if I post a long post to explain my number thing? It might help shape your response.

As in nost (like lost) and r like a pirate 🏴‍☠️ ?

This any good ? https://github.com/mikedilger/gossip

So long as it’s cheap and accessible. How do peeps with little still get a voice? How about a universal income of tokens linked to KYC (needs thought through 🤔)

Yes phone sucks. And although Damus is good I’m getting lost in the thread replies. I don’t know about Linux - something I’ve always wanted to explore. I have a VM I can instance so if you find find something good let me know.

Thanks. I don’t know a lot about this. Is there a specific peak oil model that was wrong? It would be good to go back to some specific widely published models maybe before 2000 and see what it says. They often come with side notes and references so it might be good to check those out too.

Thanks. This is helpful. I don’t get to chat to folks with your side of the argument much- which I recognise is part of the problem. Maybe nostr is better than Twitter echo chamber? Tbc I agree with you that climate “catastrophe” is overstated and govs using it as a power grab BUT equally I find it hard to put my risk estimate so low. If you ignore the doom sayers who are a small (but loud minority) there is a body of evidence that makes me give some weight to the risk. I find the scientific community to be the most dissident and sceptical and overly righteous folks out there. If the evidence is poor they’d be all over it. Proving things wrong - that’s kinda the method. What’s hard is that it’s complex. I know my field but wouldn’t comment or peer review outside that so that leaves us to weight how much I “trust” that group. That leaves the probability that 1) everyone is an idiot (possible but I would weight that around <1% as they do get other stuff right sometimes and the models they use do work in other fields very well) 2) They are all colluding (again possible but knowing scientists I would put that at <0.01%) 3) They are looking at the wrong data (this is related to point 1 but also includes the smart ones just biased / blinkered - maybe ~10%) 4) they are mostly right but it’s not as bad (maybe ~50%) 5) they are bang on or under (maybe 50%).

Have I missed anything or made any incorrect assumptions? What’s your take on the weightings? (They don’t need to add up to 100 as they are relative).

Ok, let’s say 0.1%. What evidence would you want to see to increase that by one order of magnitude to 1%?

If you were to take a baysean approach what probability would you give on the risk being true? <1%

Sorry I just wasn’t sure why you posted it. Are you assuming carbon dioxide levels haven’t risen? I’m trying to be a “good baysean”.

Are you suggesting we turn the EU back into a forest?

Given we’ve been warned for at least 30 years and done nothing and the models all appear to be coming true. Even if you are only 99% sure about climate change scam, given the risk of what they are claiming, might it not be better to at least hedge?

You should get yourself a lightning wallet for zaps or are you a doge maxi?

Thanks. This was a helpful checklist.