10 Comments
User's avatar
Michael Magoon's avatar

Yes, when I see a supposed break in a trend, my first thought is:

1) Did the metric change?

2) Is there something the matter with the raw data?

Expand full comment
barnabus's avatar

Wow, so it's 98% virtue signalling everywhere, not just "genocides"...

Expand full comment
Rod David's avatar

Separate, but similar, is how (in my proprietary work) market price chart trendlines are tested -- before either breaking, or else resuming the trend... Picture an orderly uptrend, comprised of alternating higher highs and higher lows. A straight line connecting the lows magically reverses price back into the ongoing uptrend. Until it doesn't, and the trendline support is broken. Mainstream analysis labels this as a trend reversal. But the trendline break itself doesn't reverse the trend, not unless that reversal extends to also reverse below the trendline's last touch, i.d. the previous higher low. And very often, that test holds, and the trend resumes -- often more steeply as the trend reversal sponsorship adds fuel to the prevailing trend's resumption.

Expand full comment
vectro's avatar

Does this logic extend to concerns about environmental contamination? For example, microplastics or PFAS? If so, how do we know we're dealing with an artifact, and not the new lead paint?

Expand full comment
David's avatar

ACLED data on militia attacks before and after the main dataset on militias (PGMD) made this the widely used term for measuring violence by nonstate groups circa 2015 (happy to share an image if it’s interesting)

Expand full comment
Cremieux's avatar

Definitely tell me more.

Expand full comment
David's avatar

ACLED, one of the most widely used conflict datasets, relies on media reports to geolocate events like battles, civilian targeting, and IEDs, classifying actors as state forces, rebels, political and identity militias, rioters, and so on. Violence appears to rise across all groups over time, likely due to better or more permissive reporting—but there’s a notable spike in political militias around 2015 (it looks clear when plotted). That timing aligns with the release of the Pro-Government Militias Dataset (PGMD). Before PGMD, researchers (myself included) used terms like "paramilitaries" or "death squads," but once PGMD became available around 2011-2012 (article published in 2013), “militias” became the dominant label, because that’s what the data captured. I suspect the rise in reported militia activity may reflect this shift in researcher and journalist terminology, not just on-the-ground dynamics.

Expand full comment
Roger Tilbury's avatar

Compare and contrast "global warming" once platinum resistance thermometers started to be used in weather stations.

Expand full comment
Cremieux's avatar

Got a link on this?

Expand full comment
Roger Tilbury's avatar

Try this:

<https://tallbloke.wordpress.com/2025/06/02/platinum-resistance-thermometers-are-they-suitable-for-weather-readings/>

There are lots of other really interesting articles on that site detailing how the UK Met Office are using Weather Stations that are completely unsuitable...

Expand full comment