Jake Gloudemans

Predictions #5

August 6, 2023 7:50 PM

This was a pretty taxing week in the prediction tournament. A bunch of new questions opened up, several of them relating to breaking news events and requiring some vigilance to keep track of. We’re also close to 20 active questions now, so just checking for updates on each of these is getting to be a big task on its own.

On the other hand, a handful of questions resolved this week - all of them favorably for me - and I’m now inside the top 10. So at least the effort is paying off!

I’m going to hold off on doing a deep dive this week since there are so many new questions and resolutions to cover (plus I was a little behind already last week), but the pace of new questions seems to have cooled off the last few days, so hopefully I’ll do one next week.

Current Metaculus Stats
Total Points 378 (+174)
Overall Rank 2545 (-1652)
QC Resolved Qs 8 (+3)
QC Open Qs 19 (+4)
QC Rank 7 / 349 (-21)
QC Take 1.7% (+0.6%)

New Predictions

Coup in Niger

This is the biggest geopolitical news of the week and has the potential to become the biggest geopolitical news since the Russian invasion of Ukraine. Here’s the basic context:

  • On July 26th, Nigerien soldiers detained president Mohamed Bazoum in Niger’s capital city, Niamey, and declared they were taking power
  • This is just the latest in a recent string of coups in the Sahel region, which has been one of the most coup-prone regions of the world over the last several decades
  • This completes the “coup belt” - an uninterrupted string of African nations spanning the continent from coast to coast currently controlled by a military junta
  • Niger’s prospects had been relatively promising for countries in the region. It had a democratically elected leader, and was making progress on education and economic revival
  • The US and France both have 1000 or so troops stationed in Niger, and had been working with the Nigerien military on counter-terrorism efforts against Islamist groups in the region.
  • ECOWAS - the Economic Community of West African States - is comprised of a dozen or so countries in the region, and coordinates economic and military efforts between the countries
  • ECOWAS has at times organized military interventions in the region in response to other coups. However, in the past, they have only intervened in small countries where they have overwhelming military force.
  • On July 30, ECOWAS issued an ultimatum to the leaders of the coup, stating that if Bazoum was not restored to power by August 6th, they would face consequences, potentially including military action
  • Mali and Burkina Faso, two of Niger’s neighbors which are currently under military rule following their own recent coups, have threatened to go to war to defend Niger, if ECOWAS intervenes.
  • ECOWAS has allowed coups to unfold in the past without taking action - Mali and Burkina Faso being two clear recent examples. However, they appear to be serious with their current threats. They are now in a situation where if they don’t act, they could lose credibility, potentially trigger more coups in their member countries.

Thus, there’s now a strong possibility that a multi-nation war is about to break out across West Africa.

There were several questions this week around this event:

  • Will ECOWAS launch a military intervention in Niger before August 12, 2023?

  • Will Mohamed Bazoum, Nigerien President, return to power before August 31, 2023?

    • Conditional on ECOWAS intervention, what is the chance?
    • Conditional on no ECOWAS intervention, what is the chance?

This event is unfolding very quickly (as I am typing this, the ultimatum is expiring and I’m seeing some reports of military prep going on), so don’t put too much stock in these numbers. They’ve already changed considerably the last few days, and will change a bunch more over the next few.

Right now, I’m at about 40% chance of an ECOWAS military intervention. ECOWAS is stuck between a rock and a hard place here - if they don’t intervene, more coup attempts are likely to follow, but if they do, they risk a large scale regional war. Right now, indications are that ECOWAS is preparing to intervene and intends to do so, but they of course could still be bluffing or could wait beyond the question deadline to begin.

If ECOWAS doesn’t intervene, it seems very unlikely that Bazoum would return to power. Looking at coups in Africa since 2010, there’s only 1 instance where a deposed leader quickly returned to power, a base rate of about 5-10%. Notably, there are no anti-coup protests going on in Niger (at least from what I’ve seen on social media), so there’s no real reason to think that absent military intervention, there will be internal pressure to return Bazoum to power.

If ECOWAS does intervene, the chance is probably much higher, but still would require a fairly quick, decisive, and successful result for ECOWAS. Any military intervention would be led by Nigeria, who’s military seems considerably more capable than Niger’s, so quick success is possible, but I have a lot of uncertainty here

  • 4% chance Bazoum returns to power if ECOWAS intervenes
  • 40% chance Bazoum returns to power if ECOWAS doesn’t intervene
  • 18% chance Bazoum returns to power

Will Meta launch a Threads web app before October 1, 2023?

Threads, Meta’s new Twitter clone, launched in early July and rapidly amassed over 100 million users, by far the fastest any platform has reached that number. The vast majority of the early signups have, at least temporarily, stopped using the platform regularly. It still lacks lots of basic features that Twitter users are accustomed to, like a web version (which would enable desktop use), a search function, or direct messaging. The one major feature they’ve added since launching was a chronological feed only showing posts from people you follow.

I’m very confident that a web version of Threads will be released in the next two months. A web version is by far the most requested feature I’ve seen - pretty much every day when I open the app, I see some high profile person begging for the feature. The Threads product lead, Adam Mosseri, has repeatedly indicated that the team is working on this feature.

From a technical perspective, this doesn’t seem like an enormously difficult feature, compared to setting up the actual backend of the platform and preparing to support 100 million users. From a business perspective, it seems very important for them to launch a web version quickly, even if just a barebones version. Threads really can’t be taken seriously as a threat to Twitter until it has a web version. Getting people to use a platform like Threads requires there to be lots of new content being posted from popular accounts, so that whenever you check your feed, there are new posts to read. And it’s just hard to get that volume of posts without a web version - writing posts from a phone slow and cumbersome.

I suspect that this is the single highest priority feature for the Threads team at the moment, and that they’ve been working on it for a month or more already. I opened at 90%, then raised this to 98% after Mark Zuckerberg posted on Threads stating that a web version was coming in the next few weeks.

Prediction: 98% Yes (opened at 90%)

Will the Israeli High Court issue a ruling on the “reasonableness” law before October 1, 2023?

Israel is mired in political turmoil, as Prime Minister Benjamin Netanyahu and the Israeli Knesset (the country’s legislature) are attempting to dramatically weaken the country’s Supreme Court. Here’s my basic read of the situation:

  • Israel has an unusually powerful Supreme Court, when compared to other democratic nations
  • On the other hand, Israel has unusually few checks on the power of the government. There’s no executive with veto power over legislation and the Knesset itself is unicameral (or rather, the executive is selected by the legislature, and so not independent in the way a US president is, for example).
  • Compare this to the United States where we have a powerful, independent executive branch and a bicameral legislature (both chambers have to agree before anything gets passed), as well as powerful state governments which themselves typically cede significant authority to county and municipal governments.
  • Netanyahu argues that Israel’s Supreme Court is unusually powerful and not democratically elected, and so weakening it is reasonable
  • Opponents argue that weakening the Supreme Court will give unchecked power to the government, which could then very easily rewrite Israeli law to further entrench its power. This could mark the beginning of the end of democracy in Israel, and is similar to the strategy Viktor Orban has used to solidify his rule in Hungary.
  • The governing coalition of Israel is currently very right-leaning, and its members have very plainly expressed a desire to relegate Arab-Israeli citizens to second-class status and move the government in an explicitly religious direction.
  • The Supreme Court has typically been far more left-leaning and secular than the government, which has made it harder for the government to move the country in this direction. Weakening the Supreme Court would remove this barrier.
  • Netanyahu’s insistence on proceeding with these judicial reforms has triggered widespread protests throughout the country

This is the larger context in which the “reasonableness” law has been passed. The reasonableness law is the first piece of the broader set of Supreme Court-weakening reforms that Netanyahu is working to enact. I don’t really have a great grasp of legal/judicial topics, but generally speaking, this law limits the scope of cases that the court is able to issue rulings on.

In the past, the standard has most often been used to strike down “unreasonable” government appointments. While in the US, the Senate has to approve appointments to key position, there isn’t an equivalent process in Israel. For example, in January of this year, Netanyahu appointed an official who had previously committed crimes while holding that same office and cut a plea deal agreeing to never hold public office again. The court blocked the appointment using the reasonableness standard.

The law was passed on July 24, and in response to numerous appeals against the law, the Supreme Court quickly announced it would hear those appeals and issue a ruling. So we have a situation where the court is ruling on a law which would greatly reduce its own jurisdiction moving forward. If it accepts the law, it’s a step towards unchecked government power. If it strikes down the law, it sparks a constitutional crisis, pitting one branch of government against another.

Most of this context is not actually super important for forecasting the question at hand, except to say that this is a very significant decision, where either outcome will have profound implications for Israel’s future. In terms of forecasting, the main thing here is that the timeline is very short.

The court will hear arguments on September 12th, which leaves just 2.5 weeks for them to make a ruling (assuming the arguments only take 1 day). This just seems like far too short a timeline to expect a decision - from what I could find, it sounds like several weeks to several months is the typical timeline for rulings, and in this case the decision is hugely important, which I expect will lengthen the timeline considerably.

I don’t think it’s totally impossible - maybe they will actually expedite the ruling, given the massive public interest, and the fact that the law is currently in effect and preventing them from using the “reasonableness” doctrine. But even in that scenario, 2.5 weeks is still a stretch.

Prediction: 5%

Will the oil transfer from the FSO Safer finish in 19 days?

The FSO Safer is an oil tanker that was abandoned in the Red Sea off the coast of Yemen back in 2015, containing over 1 million barrels of oil. It’s remained there ever since amid regional conflict and disputes over who owns the oil and ship.

In the years since the ship was abandoned, its hull started corroding, and the inert gas that typically fills the empty space in the oil tanks to prevent the oil from exploding has gradually leaked out. This has led to increasing concern about a massive oil spill and/or explosion, which would shut down nearby ports and ravage the local fishing industry.

In 2022, the UN launched a crowdfunding campaign to raise funds for a salvage operation to offload the oil from the vessel. The UN purchased the vessel earlier this year, and the salvage operation began in July. The salvage company, Boskalis, stated that the offloading of the oil was expected to take 2-3 weeks. This question opened around July 25th, the date that the oil transfer from the FSO Safer to the salvage vessel began. The question is asking whether the transfer will finish within the anticipated timeline.

I forecasted a high chance of this happening, opening at 90%:

  • I expect the salvage company would err on the side of being too conservative with their timeline, rather than too aggressive
  • The timeline for transferring the oil seems easy to predict, relative to other parts of the operation. You know how much oil there is to pump and how fast your pumps go.
  • When the question opened, pumping had already started. So by this point, many of the more uncertain parts of the process were already finished
  • The salvage company has been very transparent about the whole operation and things seem to be going very smoothly.

I’ve raised my estimate a bit after we’ve received a few early updates about the progress of the transfer. It took less than a week for the first half of the oil to be transfered, so things appear to be very much on track.

Chart showing oil tranfser progress for the FSO Safer salvage operation, vs. pace required to finish by question deadline

The last few updates have made me slightly concerned as the pace appears to be slowing down, but it’s hard to tell if this is real or if the timing of each progress report is just off a bit. I’m just going off of periodic Tweets from a few accounts tracking the operation, and the exact timing of each update could very well be off by up to about a day in either direction.

For now I’ve lowered my estimate slightly, but the next few days should help clarify if there’s actually been a slowdown or not.

Prediction: 92% (updates as low as 75%, as high as 95%)

Will the reported 2023 count of TBE cases in Sweden exceed 370 on October 1?

Changes in weather patterns have led to a rise in tick populations across Sweden, which in turn has led to a rise in cases of Tick Borne Encephalitis. This page records case statistics for TBE since 2013 - the number of annual cases has nearly doubled in that span from around 200 cases per year to 400 or more cases per year. The question is asking if at the end of the tournament, on October 1, the number of cases will exceed 370.

I didn’t do much research in terms of the disease itself here, and am mostly just looking at the way cases have historically risen over the course of a year. This graph shows cumulative cases per month for each year since 2013, with the average in blue and the current year in red. The green dashed lines show the end date and the case threshold for the question to resolve Yes.

Chart showing cumulative cases of TBE over the course of each year since 2013

Since 2013, the case count on October 1 has only exceeded 370 once, in 2021. The two years with the most cases are 2021 and 2022, and the other 3 above average years are 2017-2019. 2023 is currently at a slightly slower pace than 2021 and 2022, but is more similar to those years than any prior years. Since 1/2 of those years closed October above 370, I’m therefore opening with a forecast slightly below that, 30%.

Unfortunately, it seems like the data only updates once per month, so I’ll only really have one more opportunity to check if cases are tracking above or below 2021 and 2022. I may look around and see if I can find more frequent case reporting somewhere, which would give a slight edge in terms of updating the prediction as time passes.

I realize this is a very rudimentary approach, but I’m hesitant to try and do anything much fancier than this. I know very little about disease modeling or specifics of this disease that might help me give a more confident prediction. I think this at least puts me in the right ballpark - cases seem to evolve with a similar pattern each year, and the years most similar to this one give a 50% chance. And so far, this year is slightly below those years.

Prediction: 30%

Will any of the listed alleged co-conspirators of Donald Trump be indicted for a federal felony before August 11, 2023?

The latest indictment of Donald Trump named 6 “Co-conspirators” who were instrumental in his attempt to overturn the result of the 2020 election. The indictment doesn’t directly name these conspirators, but 5 of the 6 can be determined from context:

  • Rudy Giuliani
  • John Eastman
  • Sidney Powell
  • Jeffrey Clark
  • Kenneth Chesebro

None of the co-conspirators were charged in the indictment, only Trump, so the question is asking if any of them will be charged in a subsequent indictment by August 11th.

After reading about this, it seems like the different possible scenarios here are:

  • None of them will be indicted at all
  • One or more will eventually be indicted but Trump’s indictment was prioritized / accelerated so that a verdict can be reached before the 2024 election
  • The prosecutors are trying to get one or more of the co-conspirators to cooperate - if they don’t cooperate they will be indicted later

I don’t know enough to really interpret the indictment myself, but from other analyses I’ve read, the co-conspirators are very clearly implicated in the Trump indictment, maybe more than Trump himself. So the first option above seems very unlikely to me. If the co-conspirators are clearly guilty, it would be strange to let them off the hook just because Trump is the priority.

Some articles I’ve read have speculated that the 3rd option could be what’s going on, but others have said this is unlikely, because each of the co-conspirators has given no indication that they will ever cooperate. Even if this option is true, I would expect them to give more than one week for cooperation.

So the 2nd option - they simply prioritized Trump’s indictment and will eventually indict the others - seems the most likely. If this is the case, it seems strange that they would indict the others just one week after Trump. At that point, why not just do them all at the same time?

Collectively this makes me very doubtful there will be an indictment of any of the co-conspirators this week. That said, I’m not super knowledgeable about the legal process and strategy here so I don’t want to get too crazy confident and go to 1% or something.

Prediction: 5%

Will the Antarctic Sea ice for every day in September 2023 be the lowest in recorded history?

Antarctic Sea ice is currently, way, way, way below historical average. To make a forecast here, I looked at this graph and thought, “there is absolutely no way this will not happen.”

Plot showing deviation in Antarctic sea ice extent over the course of different year. 2023 is far below any prior year.
from The Economist

Look there were a lot of questions this week and I have a life to live!

Prediction: 98%


Will Ukraine recapture the center of Bakhmut by the end of September?

Initially 22%, as high as 32% → now 19%

At this point, consensus among the military analysts I’ve been following seems to be that Ukraine’s counteroffensive from the last few months has largely been a failure, at least in the sense that little ground has been re-taken relative to the cost of the effort. They’ve made minimal gains along all axes of attack and have settled into an attritional approach, where the goal is to maintain an attritional edge over Russia for a sustained period, hopefully leading to an eventual breakthrough. For a while, progress in the Bakhmut area was going comparatively better, with Ukraine steadily retaking ground, but this progress has slowed to a crawl in recent weeks.

Additionally, it’s not clear the retaking Bakhmut is a priority, or even a good military strategy right now, and retaking the city may not be Ukraine’s current objective in the area. Recapturing Bakhmut would be a blow to Russian morale, but holding the city would likely be very costly, and isn’t strategically significant within the broader context of the war. Right now, the main significance of the battle is that it’s “fixing” Russia’s forces, i.e. keeping them from being somewhere else. If holding the city is more costly for Russia than attacking it is for Ukraine, and the battle is diverting Russian forces from the other axes of attack, Ukraine may be okay with proceeding very slowly and maintaining the status quo there.

Resolved Predictions

Will a nationwide UPS strike be underway in the United States on August 4, 2023?

Plot showing my forecast vs. the Metaculus forecast for this question

Resolved No

I already discussed this briefly last week, but this one did officially resolve No. I made up some ground in the standings with this, as I was well below the Metaculus forecast for half the question, and the community forecast was actually more like 40% for most of that time. I’m honestly not really sure why everyone else was giving this such a high chance of happening and I think my lower estimate was warranted. I think people maybe weren’t appreciating how large of a strike this would have been historically speaking, and so were underrating the incentives of both parties to avoid this. And frankly, the union demands didn’t seem particularly extreme and there seemed to be lots of room for various compromises, or even for UPS to just meet all the demands.

I also think that negotiations in general give people a hard time, because the outward statements each party makes are often pretty disconnected from the reality of what’s actually happening in the negotiation. I think the community was reading too much into things like the “practice pickets” that the union was holding, and not putting enough weight on priors.

Will an attempted replication of LK-99 superconductivity be published before August 4, 2023?

Plot showing my forecast vs. the Metaculus forecast for this question

Resolved Yes

There ended up being one or maybe two papers that met the criteria here and were published by August 4th. I bounced around a bit here as you can see. The lower of the two dots in the middle was the prediction I made after writing out my thinking in last weeks post. I moved downward a bit, just for the passage of time, and then jumped up to the high 90s once the papers showed up an ArXiv. I was generally well above the community forecast though, so again picked up lots of points here. I think the reasoning I laid out last week actually holds up really well. The “it just takes one” nature of the question plus the very tight timeline roughly balanced out to around a 50/50 proposition.

This could just as easily have resolved No. While there were a handful of LK-99 papers published throughout the week, most of them weren’t really replication attempts, and of the actual replication attempts, I think only 1 actually reported all the data mentioned in the question criteria. It would totally have been possible for them to not make those specific measurements, or wait a few more days, in which case this would have resolved No. So I think 50/50 or maybe slightly below that was a pretty good estimate.

Will Crook Manifesto by Colson Whitehead be on the August 13th NYT Bestseller list?

Plot showing my forecast vs. the Metaculus forecast for this question

Resolved Yes

Crook Manifesto came in at #11 on the August 13th list. I ended up getting lots of points here because of my high initial confidence, and since I was above the community almost the whole way. I think I got slightly lucky though - the book definitely did a bit worse than I was expecting (I wouldn’t be shocked if it’s off the list next week), and as you can see, once sales started and I could see some early indicators from Amazon and other lists, I started to lose faith a little.

Maybe my reasoning was good and this was really a worst-case scenario, and so I was right to be so confident. But given the fact I’m not super familiar with the dynamics of book sales, a 94% chance to start with was probably a bit too high. To put it another way - if I routinely make 94% confidence predictions, given the level of information I was working with for this question, I will probably be correct less than 94% of the time, so I’d be better off predicting with lower confidence.

A general observation I have from my early predictions is that I should avoid really high confidence predictions for subjects that I’m more unfamiliar with, to account for the chance that there are important factors I completely missed. I got away with it here, but in the long run it will probably come back to bite me.