Hacker News new | ask | show | jobs
by nvm0n2 933 days ago
It's not called fraud in climatology and yes it is how "science" functions. There are no universal standards in academia. Climatological norms involve heavy processing of thermometer readings to "clean" them in various ways. They all do it and do not see any problems with what they do. Every claim you've ever read about temperatures from climatologists is based on that kind of procedure.

Some of this has good intentions, at least originally. Weather stations aren't normally intended to be used by climatologists. They exist for other reasons. So they get moved around, or not moved even as the environment changes around them, get placed in inconveniently unrepresentative places like airport runways, and more. Climatologists scrape this data from the internet or collect it from logbooks and then try to work out what's happening, but the data is super noisy.

Now the way science works is that you characterize the uncertainty in your data and propagate it through any calculations you do, in order to track your uncertainty intervals. Then you communicate those and take them into account when making predictions.

But in climatology they don't do this. Instead they use lots of algorithms and manual tweaks to try and "fix" the data to bring it into line with what they know it "should" be, and then report the data without CIs, as having 100% confidence. For example if a time series at a weather station is stable for 20 years, then experiences a short break, then it returns but the average is consistently 0.3 degrees different than before, they infer that it must have moved and they then "correct" it back to the previous baseline. If there are gaps in the data then they generate fake readings by interpolating between the nearest alternative weather stations, and so on.

Outsiders might expect that they would investigate and try to improve the quality of their source data but they don't. Like, if their algorithms infer a station move, they don't contact the station operator to figure out if that really happened. They just assume their corrections are fine and move on.

Another fun thing they do is alter data that was already published. When they update their algorithms for deciding what data points to include/drop/change, they don't just use it for new data running forwards. They reprocess the entire historical data set. That can yield outcomes that would normally be taken as a clear indicator of scientific fraud, for example where NOAA declared a temperature record, and a few years later declared a new record that was lower than the previous one [1]. Or where scientists invalidated decades of published papers (thousands of them) by deciding that the temperature trend in the first 15 years of the century was totally different to what had previously been reported:

https://www.nature.com/articles/nature.2015.17700

The underlying data on which those papers were built was announced to be all wrong, but nothing was retracted! That's how science functions. And the best part is that they've trained the public so well that for anyone who calls any of this fraud, as you just did, they are instantly ostracised for being a heretical Denier.

[1] https://retractionwatch.com/2021/08/16/will-the-real-hottest...