Small nitpick. I've been following this blog for a while now and really appreciate it. I realized today I've been confused about what it means when you say "forecasters estimate that..." Part of me thought you were scanning prediction markets or aggregating external platforms. Now I assume it refers to Sentinel's forecasters. Might be worth phrasing it as "Sentinel estimate" or "our forecasters estimate" to make that clearer.
Similarly with the probability ranges in parentheses, I wasn't sure if they represent an IQR or min-max across your team. I'm now assuming min-max, but spelling that out once (maybe in a footnote or FAQ) would help readers like me parse the numbers more confidently.
Yep, and also just if troubles are contained to the middle east. Currently codes a bit undefined, but if does make sense to be a bit more alert this week.
I think "reversal of the supply chain risk designation by the end of 2027" is really the wrong timeline. The regulatory uncertainty would be very damaging. A better question imo is whether the designation is somehow annulled within ~2 months (either through court injunction, or the DoW never files it formally, or some other reason)
Good point! We were thinking about it in terms of whether Anthropic would get proven right by the courts,but you're right that they do have to be proved right fast enough.
Do you have enough predictions vs realized events that you can issue calibration metrics? I'd like to see those regularly (if only occasionally) in these
Is there a legend for the color codes in the title? Some rough meaning attached to it being yellow now instead of red or green?
Small nitpick. I've been following this blog for a while now and really appreciate it. I realized today I've been confused about what it means when you say "forecasters estimate that..." Part of me thought you were scanning prediction markets or aggregating external platforms. Now I assume it refers to Sentinel's forecasters. Might be worth phrasing it as "Sentinel estimate" or "our forecasters estimate" to make that clearer.
Similarly with the probability ranges in parentheses, I wasn't sure if they represent an IQR or min-max across your team. I'm now assuming min-max, but spelling that out once (maybe in a footnote or FAQ) would help readers like me parse the numbers more confidently.
Yes, this is Sentinel's own forecasters. The aggregate is the geometric mean, the range are the min max of individual forecaster estimates.
Do the forecasts come from just the four authors of this article or are there more than four forecasters each week?
Is the code yellow due to potential escalation of the Iran conflict, or something else?
Yep, and also just if troubles are contained to the middle east. Currently codes a bit undefined, but if does make sense to be a bit more alert this week.
I think "reversal of the supply chain risk designation by the end of 2027" is really the wrong timeline. The regulatory uncertainty would be very damaging. A better question imo is whether the designation is somehow annulled within ~2 months (either through court injunction, or the DoW never files it formally, or some other reason)
Good point! We were thinking about it in terms of whether Anthropic would get proven right by the courts,but you're right that they do have to be proved right fast enough.
Do you have enough predictions vs realized events that you can issue calibration metrics? I'd like to see those regularly (if only occasionally) in these
Seconding this
Yep, see https://substack.com/profile/19041656-nuno-sempere/note/c-217680732. Can mention this in the next newsletter.