Given the base rate data and lack of any positive outlook in the future (at this time at least), 2,000 unfortunately looks like a pretty low threshold.
No Scores Yet
Relative Brier Score
0
Forecasts
0
Upvotes
Forecasting Calendar
Past Week | Past Month | Past Year | This Season | All Time | |
---|---|---|---|---|---|
Forecasts | 0 | 0 | 0 | 0 | 30 |
Comments | 0 | 0 | 0 | 0 | 18 |
Questions Forecasted | 0 | 0 | 0 | 0 | 13 |
Upvotes on Comments By This User | 0 | 0 | 0 | 0 | 8 |
Definitions |
Active Forecaster
Why do you think you're right?
Why might you be wrong?
The situation could of course change and this is early to be forecasting it.
Why do you think you're right?
The 2020 activity, at 461, is by far the highest and represents an outlier. To me, the question centers on what happened in 2020 and whether it represented an aberration or change of direction. Based on the 2021 number, it seems like it's the former.
Looking at the data, it's clear that nearly all the 2020 entries were in Syria and related to the Syrian civil war. Given that the conflict is now in its final stage (and at least some reputable sources even treat the conflict as having functionally ended in 2021; see The Economist). Assuming that the 2020 spike was related to the Syrian Civil War and to the extent that that conflict is de facto over, I see the likelihood of such a number as relatively low.
Why might you be wrong?
Given the newly elected right-wing government in Israel, it's possible that the conservative government will take a harder stance and be more aggressive in the conflict.
Why do you think you're right?
Given that about halfway through this window and, most importantly, after Bolsonaro lost the election and challenged the election results, the total stands at 23, it seems highly unlikely that the total would even surpass 60-70, nonetheless 100. In the absence of a sea change (which is still possible if Bolsonaro rallies the military, but seems less likely every passing day), topping 100 would seem exceptional.
Why might you be wrong?
As mentioned, Bolsonaro could still attempt to rally the armed forces, as he's hinted at doing in the past.
Adjusting up my forecast for between 200 and 400 given that the Kremlin has cracked down even more harshly on public displays of protest against the government and further because at this point, the number of those most likely to protest on a large scale (politically engaged educated young men) has shrunken as a result of (1) the draft or (2) the mass exodus to neighboring countries in order to avoid the draft.
Why do you think you're right?
Given the recent mushrooming covid outbreak in China at this time, and given that predictions don't see it peaking until the spring, even the most jingoistic plan to invade would be put on hold at this rate.
Why might you be wrong?
Xi could always use an invasion to create a distraction and a "rally around the flag" effect in order to distract Chinese from the CCP's poor handling of the covid outbreak.
Why do you think you're right?
Much like @belikewater, this question to me is all about solar. In the decade from 2012 to 2022, we see that biofuel consumption in the U.S. is about steady, whereas solar continues to increase. The EIA itself predicts that solar will be up and expects almost all the increase to be in wind and solar. A simple regression puts biofuels' share at 19% using data from 2012-2022. With that in mind, if solar moves up even a hair, it wouldn't be expected to break 20%.
Why might you be wrong?
Could always be missing something and it's possible that solar installation stalls, which would vastly increase the chances of biofuel's share increasing.
Why do you think you're right?
Using data from the California DMV, I constructed a dataset of the total counts of vehicle reports by month and year since 2016 and plotted the totals (see first image). In the absence of other data, I relied solely upon this. (There were excellent-looking market research reports about the forecasted growth of the autonomous vehicles market, but they were paywalled.)
As we can see, total aggregate reported crashes have increased most years. See table:
Year | Total Crashes |
2016 | 15 |
2017 | 29 |
2018 | 75 |
2019 | 63 |
2020 | 42 |
2021 | 117 |
2022* | 138 |
* data through December 2, 2022
Further analysis, however, showed that, in the same period (2016 - 2022), certain months had higher averages than others across years (albeit without statistical significance testing).
I subsetting the dataset to the three months of interest for each year and plotted them (see second chart). As we can see, the 2022 figure was 47. So, unless there were a decline next year (which would be unprecedented), the first three bins (0 - 39, inclusive) seem highly unlikely.
The remaining options are either staying within a 9-point range of the 2022 figure (7 below for 40 or 2 above for 49), which seems reasonable, or increasing by at least 3 points, which seems the most likely based on historical data: the average increase from 2018 was 8.5 points, and even ignoring the large 2021-2022 jump, the figure would be 3.3.
Why might you be wrong?
I haven't had any good correlative data or leading indicators, so it's hard to know if there's a real relationship or pattern here. Operating under the assumption that the number of reports is a function of the number of tests being done, which itself is a proxy for the vitality of the industry, and to the extent that the broader economy would affect the vitality of the industry, it's possible that a slowing economy will drain resources and fewer test drives will be completed resulting in fewer crash reports than otherwise would be expected. In fact, we already see 2022 figures beginning to trail after April.
Why do you think you're right?
Over the past six years, we haven't seen the annual cost decline necessary to get the costs down to $100 in the next year. In order to get that kind of cost fall, you'd need to see totally unforeseen declines in costs. Given that the marginal gains of data have begun to slow, we would expect slower, not faster, declines in costs (which is borne out in the data from 2019-2021). I have averaged the year-on-year decline in costs for 2020 and 2021 to do a rough forecast for 2022 and 2023. We see that we'd be nowhere close to $100, and that assumes no decline in year-on-year cost changes (in other words, the same annual cost decrease was applied to 2022 and 2023). A real breakthrough would be necessary to get this kind of cost drop.
edit: table footnote should read "August 2022 and 2023 are forecasts."
Why might you be wrong?
There could always be an unpredicted breakthrough, specifically in the relationships within the genome data already available and being analyzed.