The goal of diplomacy is to avoid STUPID wars

The goal of diplomacy is to avoid stupid wars, not all wars

Don't blame computers when you model wrong

I don't have any degrees in international security. But I'm seen as a leading global expert in artificial intelligence, and as a consequence I've had an opportunity to learn about international security from people who are global experts in that. I've attended small multi-day Chatham-rules international security workshops in Chatham House in London, the ICRC in Geneva, and the UN in New York City, as well as various larger meetings of the OSCE in Seoul and the EC in Brussels. That's still not a lot of time, but it's enough time to pick up the basics if you're any good at learning.

The most basic thing I've learnt is that the point of diplomacy and international treaties isn't to prevent war. That's seen as impossible. The goal is to eliminate wars based on miscalculations. Stupid wars result in destruction and suffering for no reason. If only you could honestly communicate your commitment and abilities, then maybe any issues would be resolved without violence at all. Although the only way to really, honestly communicate abilities and commitment is unfortunately to periodically display them. This is part of why zero war is not seen as likely to be achievable.

map showing economics not language or ethnicity determine where 'rebellion' happened in Ukraine
Zhukov, Yuri M. "Trading hard hats for combat
helmets: The economics of rebellion in eastern Ukraine
Journal of comparative economics 44, no. 1 (2016)
No one should in any way belittle what the Ukrainian people are doing. But why they are having to do it appears at least in part to be that Putin did not or could not believe what Science shows: that even ethnically Russian people of Ukraine would much rather have the relative liberty, wealth, and stability of the EU versus what's on offer in Russia (see first Figure). It may even seem to Putin that Russia is more stable than democracies, because it has single leaders for longer. But an autocracy can become a tyranny way faster than a well-structured democracy can. And by well-structured here I mean ones that truly, structurally support the rights of minorities. These days the best democracies we seem to see are those with multiple functioning parties and coalition governments. The legitimacy and robustness bought through proportional representation may be becoming as important as the separation of powers, civilian leadership of the military and other 18C Anglo-American innovations.

If we've managed to make it really stupid to fight an offensive war, that's a pretty good outcome. The only really well motivated offensive war I can think of in my lifetime was the invasion of Cambodia by Vietnam, which occurred after Pol Pot has already killed 1/4 of his citizens. Arguably, NATO actions against Kosova and Libya were similar, and also what NATO intended to do in Syria before Obama took that one walk in the White House rose garden. These actions and inaction all created ongoing messes, but perhaps the interventions created more merciful messes than what they replaced, or what did happen in Syria. I'm not informed enough to say.

But I know a lot about modelling. I know it's not useful to blame the miscalculations of Putin – and many other people, including in the West – on some mystic or supernatural capacities of the Ukrainian people. Yet these miscalculations seem to be ongoing. Over the weekend I was enraged to see that Marine Corps Commandant General David Berger had claimed that computer models "didn't and couldn't" account for "the Ukrainian people's will to fight for their homeland." Maybe those models didn't, but they so easily could have. Or rather, the people that wrote them could have set that level correctly.
Click the picture to see this in twitter, or
 click here to see the article Gady is quoting.

The General's statement is more of the dangerously insane "robot rights" anthropomorphised thinking about AI and computers that initiated my entire career in ethics. Here is a general blaming computers for being incapable of understanding human emotion, when all that's needed is to have adjusted the "will to fight" parameter on that model, if that even really was the problem with their predictions. The model may also have failed because someone hadn't updated its estimate of Ukraine's capacity to fight, despite the way they've been arming and training since 2014, or because someone underestimated NATO's willingness to provide money and military intelligence, if not so far actual material help. Someone's model failed, and a war started. And maybe the model isn't a complete failure, because no one has ended that war yet, so we don't know whether yet again Putin is going to be accommodated, like he was in Syria, or with respect to the election interference in the Brexit and Trump wins.

The Ukrainian people know what they want, they know what they've lost, and they know their enemy. They know what life is like in the parts of their country that have been occupied by Russia since 2014, and how it is over their various borders. They also know that their enemy is a set of human beings exactly like them making belittling decisions, and they are rightly outraged. And at least some of them know how much military training and support their forces have been receiving since 2014, and how much intelligence support they are receiving now, and think they have a chance. 

Let's not mystify this process. And I can't say this strongly enough – let's not pretend that computer simulations can't model basic social dynamics because they're so f**king mystical. Scientists model emotions and social dynamics every day.  I personally have published models of this kind of polarisation, and I'm hardly the only one. If there really is this model Gen Berger describes that included an underestimate of the Ukrainian people's willingness to fight, then the people who built it made a mistake. People were wrong about the inputs. At this stage we need to reassess everything those who are surprised by what's happening think they knew, and then use our improved understanding to figure out how to conclude this war with the minimal further damage. When we see our models are wrong, we should then learn something about the world and fix them.

And while I'm ranting – this isn't only about Putin. Putin is being directly supported by many overly-wealthy people in Russia, London, the US, and other places. And he's being indirectly enabled by everyone who thinks it's OK for individuals to have this much power and seeks to improve their own wealth and power by disrupting regulatory efforts designed to ensure that democratic governance can limit these kinds of excesses. This is another way that I have come to expertise about this – because I work in digital governance. Consequently, I have been learning about market concentration and its relationship to "big tech" and the entity formerly known as GAFAM (MMAAA – bleat it like a sheep, then say "wake up sheeple"). I've known since 2016 that inequality was a posited major factor in WWI. I've also known this was even more important ethics issue than bias in AI because this time it's not just a potential world war, it's a potential world war during a climate crisis.

See also my previous blogpost: Paying a cost to harm others more: Subtractive asymmetric balancing, antisocial punishment, and Ukraine's nuclear power plants, also based on some of my published science. Science can explain what's going on, you just have to want to know even more than you want to be rich.


Lukas said…
What exactly is the take-away message from this post? Apart from that the author has a lot of expertise in lots of fields (as well as a novelist's understanding of how models are used?) Would have loved to read something related to the title, i.e. why the goal of diplomacy is (limited) to avoid stupid wars.
Joanna Bryson said…
Thanks for asking. The main point of both this and the previous blogpost was meant to be re-enforcing that what is going on now with the Russian war both is comprehensible and was predictable. I want to lend support to those who are doing a good job of understanding this, and undermine those claiming these are either random acts of a madman or cannot be addressed. I want everyone to recognise we are in a hybrid war, and that the side supporting the rule of law and human rights needs to take stronger action to both weaken the other side's position, to ensure it gets zero benefit from this act of aggression, and that it does not soon repeat the testing of the limits of our resolve that it has been doing not only in Ukraine but Syria, Georgia, the US, the UK, Hungary and so forth.

The main takeaway of this specific post is meant to be boldfaced but doesn't look so in my chrome browser on my laptop (though it does on my iphone safari) "At this stage we need to reassess everything those who are surprised by what's happening think they knew, and then use our improved understanding to figure out how to conclude this war with the minimal further damage." The reason I talk about my expertise is that I've learnt that not everyone knew that some major security organisations thought I was qualified to speak on security, which I bring up just to reenforce this important message. Sorry to throw in arguments from authority; mostly this is meant to be arguments from science or by reason, but it's an important enough topic I threw in everything I had.