*This essay is by* [*Sam Freedman*](https://twitter.com/Samfr)*, who writes on politics and policy at New Statesman, Prospect, Tortoise, Financial Times, Sunday Times, and his* [*Substack, Comment is Freed*](https://samf.substack.com/)*.*

Correctly predicting whether Russia would launch a full invasion of Ukraine rested on your ability to read the mind of one man. Everyone could see the threat, with a huge invasion force collecting at the borders for months beforehand, but interpreting what it meant required knowing whether Vladimir Putin was trying to create leverage or genuinely believed he could win a war on multiple fronts, and across a vast territory. And not just win it but extract a political victory from the military one.

Of course now we know those who anticipated Putin would be bullish were right. It looks like he was misinformed by his own intelligence agencies about how easy an invasion would be; and to some extent had bought into his own propaganda about the weaknesses of Ukraine’s leadership. But he was not the only one to believe that Russia would win a quick and straightforward military victory.

Indeed this was the widespread expectation after the Russian military crossed the Ukraine border, early in the morning on the 24th of February. The initial plan was to rapidly capture major cities, including Kyiv, with paratroopers landing in an airport near the capital on the first day of the war. At this early stage, the Metaculus community prediction gave 90% likelihood to Kyiv falling by the 1st of April.

https://www.metaculus.com/questions/9939/  
However, this initial plan quickly failed, the major cities were defended, with the exception of Kherson in the south of the country, and the leadership of Ukraine stayed in Kyiv. The Russian army hadn’t, it seems, considered this prospect and their attempts to respond by sending a large convoy of military vehicles toward the Ukrainian capital got, literally, bogged down. By the 28th of February there was a 65 kilometer traffic jam stretching from the Belarusian border to the outskirts of Kyiv.

Despite this, and Russia’s failure to take any other cities, even much smaller and less well defended ones, the community's median prediction still held at 69% likelihood Kyiv would fall by the 1st of April. It didn’t drop below 50% until March 5th, by which point the Russian convoy was very clearly stuck, the defenses around Kyiv had been strengthened considerably, and international military aid was flowing into Ukraine.

This seemed strange to me at the time – on the 3rd of March [I said on Twitter](https://twitter.com/Samfr/status/1499507148002320385?s=20\&t=uKoaUpCtwobJfAHHxgiaKA) that I thought it should be “well below 50%” – and it still seems strange to me now. My father, Lawrence Freedman, an Emeritus Professor of War Studies, [had written](https://samf.substack.com/p/a-reckless-gamble?s=w) as soon as the 25th of February that “the initial assaults lacked the energy and drive that were widely expected \[…] it is still reasonable to ask if Vladimir Putin has launched an unwinnable war.” 

By the 2nd of March he [was writing](https://samf.substack.com/p/russias-plan-c?s=w) “If there was ever any possibility that this war would end with the complete subjugation of Ukraine by force of arms this has now gone.” Given that he, and other respected military experts, were saying this, why did it take so long for predictions to correct? Why were the Russians being overestimated even after the evidence had mounted up that they had badly miscalculated? (See also the question about Zelensky’s likelihood of being President on the 1st of January, 2023).

https://www.metaculus.com/questions/9743/  
From discussions I had at the time I think there were three factors. First, many people simply didn’t understand the strategic challenges of the initial invasion plan. The relative strengths of the two armies on paper seemed extremely mismatched, but a large army isn’t much value without a plan to get it to the right place and use it effectively. Once it became clear that Kyiv and Kharkiv hadn’t fallen in the initial assault, the route to a quick victory relied entirely on an unexpected Ukrainian collapse. The huge traffic jam on the road to Kyiv became symbolic of these logistical challenges but it was quickly apparent across the country, outside of the South at least, that they were stuck.

Moreover, it was also apparent that Russia didn’t have the necessary resources for a lengthy conflict across the whole country. Had the Ukrainian army fallen immediately, then 175k Russian troops may have been enough to manage an occupation of key cities while some kind of deal was worked out. But as the Ukrainians held firm it was never going to be nearly enough manpower. Even if some cities had been taken the number of people required to hold them against a hostile civilian and guerrilla population would have been huge. It was all highly implausible. The confusion here wasn’t helped by the way maps were drawn as if Russia were taking and holding significant territory, rather than just advancing down roads (and often getting overextended).

The second reason that predictions took too long to shift away from initial expectations was, I think, a belief that the news coming out of Ukraine had to be corrected for optimism bias, due both to the natural desire for the oppressor to be defeated, and the effects of successful Ukrainian propaganda on Western journalists and public opinion. This was a real concern for a community that identifies strongly as rationalist and is aware of how emotional attachment to an outcome can exacerbate biases. As the [analyst Samo Burja said](https://www.city-journal.org/putins-bet) on March 1st: “This impressive feat of consensus-building, however, risks the same mistakes that led internationally supported Armenia to a defeat at the of the militarily superior Azerbaijan in 2020. Wars are fought at the speed of tanks, not tweets. Symbolic victories have their own value, but they cannot ultimately surmount a hard military defeat.”

Yet this underestimated the importance of those symbolic victories both in terms of galvanizing Western support which has been much more significant, and protracted, than most people expected when the war started. Even by early March sanctions had been put in place that have placed the Russian economy under serious stress, and weapons deliveries had been ramped up, which rapidly reduced the chance of any quick Russian victory.

Moreover morale is incredibly important in war. Some of the stories that emerged in the early days of the war, such as the defense of Snake Island (“fuck you Russian warship”); President Zelensky’s “I need ammunition not a ride” quote; and myths like the “Ghost of Kyiv” – could seem to those trying to be dispassionate like corny Hollywood nonsense. They could seem to give the appearance of success when in reality the Russians still had overwhelming force they could bring to bear. But these stories and images did galvanize a resistance far more stout than many anticipated, and contributed to a collapse of morale amongst Russian troops.

This is the first truly social media war, happening in a highly digitally connected nation whose younger generations understand the language of memes. And there is much to learn about the importance of the rapid dissemination of a narrative across sophisticated audiences. The Russian propaganda techniques of painfully obvious false flags and relying on Western stooges to push conspiracist rubbish have looked seriously outdated compared to the more networked efforts of Ukraine. In addition to this the ability of the largely amateur OSINT community to provide detailed confirmation of losses on both sides, as well as satellite imagery of troop progress, has made it much clearer which side’s story is closer to the truth.

Thirdly, and perhaps most controversially, I think that Western political discourse about liberal decline seeped into people’s expectations and proved difficult to budge even when the evidence was piling up in the opposite direction. Many forecasters are broadly aligned to soft libertarian movements like Progress Studies which see liberal Western countries, with some good reasons, as weakened by the regulatory and political constraints of democracy. Which isn’t to say those movements are anti-democratic, more just frustrated at the barriers to progress created by social, economic, and political trends. How many wistful posts have their been comparing the number of airports or power stations built in China to the US or UK?

The disastrous American and British withdrawal from Afghanistan was still fresh in the mind, where an army that had been exceptionally well-resourced and trained simply melted away once the Western powers left. There were obvious differences in context but it certainly played into the wider sense of decline and failure. Richard Hanania [acknowledged](https://richardhanania.substack.com/p/lessons-from-forecasting-the-ukraine?s=r) it was a big reason why he underestimated the resistance to Russia.

Yet this war has ended up as a reminder that, while authoritarian regimes can move quickly in ways democracies cannot, they also have major weaknesses. With no media or parliamentary accountability, institutions can quickly become highly corrupt and inefficient without anyone really noticing. Hierarchical control structures where favor depends on keeping the ultimate boss happy can lead to misinformation and severe optimism bias. A major lesson of this conflict for forecasters is that these weaknesses need to be built into predictions as much as the challenges of democratic consent.  

With Russian military ambitions now reduced there is no doubt that Kyiv will remain the Ukrainian capital and Zelensky the President of a free and independent state. Nevertheless there remains a question as to whether Russia will be able to hold any territory beyond that they held on the 23rd of February. Currently the Metaculus community prediction is 61% that they will hold additional territory on the 1st of January 2024. Given what we’ve learnt about how Russian capabilities have been overestimated to date, this, again, seems high to me (I would put it at 25%).

https://www.metaculus.com/questions/10745/  
NATO is now in a position where it has to support Ukraine indefinitely, and there is absolutely no interest within Ukraine for a ceasefire at borders worse than February 23rd's, despite the entreaties of Henry Kissinger and the New York Times. Zelensky has said there will be a referendum on any deal and it’s very clear the population wouldn’t accept it. Russia has expended a huge amount of resources on making some limited progress in Donbas and there are growing reports of soldiers refusing to fight. Meanwhile the economic pressure on Russia is continuing to grow, with a 6th round of sanctions recently agreed to by the EU.

None of this is immediately terminal for Russia and it’s certainly the case that offensive action will be harder for Ukraine than defense has been, even with US and EU weaponry. But if Ukraine won’t agree to a settlement including any new territory, and Russia can only take a bit more of the Donbas, it’s hard to see how it can be in their interests to prioritize that land over reducing sanctions and the cost of ongoing conflict. Perhaps we’re just back to trying to look into the mind of one man. But now he and we know the weaknesses of the Russian position.