38 ms·

I admit I’m probably being a bit too picky, but I honestly don’t think this is how science works, I think this paper is a good example of the problem of low qua

I admit I’m probably being a bit too picky, but I honestly don’t think this is how science works, I think this paper is a good example of the problem of low quality papers clogging up academic publishing due to perverse incentives. I had a quick look through the first half dozen citations on the paper, and it is a lot of stuff like this: “The effects of modern climate change are occurring more quickly in grasslands relative to many other ecosystems (Loarie et al. 2009;” This quote doesn’t really use any of the data/methods of the original paper, it just uses the paper to support the fairly obvious idea that climate change has variable effects that differ by environment. Another: “ On the other hand, the climate is shifting rapidly (Loarie et al. 2009)” again nothing is the paper is used, and they claim the paper says change is rapid, but to determine what is “rapid” we would need to compare our velocity to a value we think is “not rabid”. This would provide us with some testable hypothesis (eg “2% of regions move velocity faster than the fastest moving population in the region”). I don’t think my analogy was very good - I’m not critical because they didn’t look at my string, I’m critical because i don’t think their paper can truly help other teams measure their bits of string (or my piece). I agree their method might be useful, but it seems pretty obvious and I don’t understand why they wouldn’t apply it to a scientific theory themselves. Currently they have a model looking for a theory to test, but without any indication that their model is useful for any particular hypothesis. So many climate papers are just models with no link to any tangible hypothesis/predictions. This means they are unfalsifiable and so unscientific in Popper’s epistemology. I would even argue they are unscientific in terms of Lakatos’ Research Programmes because they are likely to be used to support auxiliary climate hypothesis in degenerating programs (since if they were part of a progressive program they would have explicit theory/predictions).

Timon3 3y ago
> I had a quick look through the first half dozen citations on the paper, and it is a lot of stuff like this: “The effects of modern climate change are occurring more quickly in grasslands relative to many other ecosystems (Loarie et al. 2009;” This quote doesn’t really use any of the data/methods of the original paper, it just uses the paper to support the fairly obvious idea that climate change has variable effects that differ by environment. I don't see a reference to Loarie et al. 2009, and I can't find that quote in the article. > Another: “ On the other hand, the climate is shifting rapidly (Loarie et al. 2009)” again nothing is the paper is used, and they claim the paper says change is rapid, but to determine what is “rapid” we would need to compare our velocity to a value we think is “not rabid”. This would provide us with some testable hypothesis (eg “2% of regions move velocity faster than the fastest moving population in the region”). Same goes for this.
- brutusborn 3y ago
  I was referring to citations of the paper (theoretically if these researchers are citing the paper, then they should be using the ideas, model or data from the paper). Have a look here: https://www.researchgate.net/publication/40768520_The_velocity_of_climate_change https://www.researchgate.net/publication/40768520_The_veloci... It has 2172!! citations! And in many(most?) cases these citations are pointless. It’s essentially used to pad out other ‘water is wet’ papers and/or hack researchers indexes. If it were a scientific article, the citations would reference it to use/improve/critique their hypothesis/model/data. I’m sure there’s a couple of these among the 2172 citations, but good luck finding the seed among the chaff! I think this stuff annoys me because of the huge opportunity cost. Climate change is such a big problem with so many critical questions unanswered, yet tomes of research get funded that tell us nothing about how to solve the problem. And worse: we get such volumes of drivel that serious researchers have to wade through heaps of nonsense to find quality papers.
  - brutusborn 3y ago
    I decided to have a deeper look at this during my lunchbreak. I found a bit of a smoking gun: they claim that "probability distribution function of our temperature-based velocities are consistent with those described previously when uncertainty is accounted for (Supplementary Table 2)" If you look at Supplementary Table 2, it looks like their estimates are in line with existing studies (Malcolm 2002). If you actually look at the source of this comparison table (Table 2 Malcolm 2002), you will find that their estimates do not align well. E.g. for 0-315m/year they estimate 37.6/90/15 % for mean/lower/upper. The equivalent measure in Malcolm 2002 is 71.1 but has uncertainty of only +/-1.23%! So their model is "consistent" with existing models because 71.15 is within their range of 15-90%. Figure S23 looks even worse. So essentially they have proposed a model which doesn't really match existing literature and can't really be easily compared or used by others. They explain the differences away with assumptions but don't explain why their assumptions are better/worse. They might have produced a worse model but we don't have any data to tell! But maybe I'm completely wrong and this paper is super-useful to those in this field. I'd still have to see at least 1 paper actually use their model to be convinced.

Timon3 3y ago

> I had a quick look through the first half dozen citations on the paper, and it is a lot of stuff like this: “The effects of modern climate change are occurring more quickly in grasslands relative to many other ecosystems (Loarie et al. 2009;” This quote doesn’t really use any of the data/methods of the original paper, it just uses the paper to support the fairly obvious idea that climate change has variable effects that differ by environment. I don't see a reference to Loarie et al. 2009, and I can't find that quote in the article. > Another: “ On the other hand, the climate is shifting rapidly (Loarie et al. 2009)” again nothing is the paper is used, and they claim the paper says change is rapid, but to determine what is “rapid” we would need to compare our velocity to a value we think is “not rabid”. This would provide us with some testable hypothesis (eg “2% of regions move velocity faster than the fastest moving population in the region”). Same goes for this.

brutusborn 3y ago
I was referring to citations of the paper (theoretically if these researchers are citing the paper, then they should be using the ideas, model or data from the paper). Have a look here: https://www.researchgate.net/publication/40768520_The_velocity_of_climate_change https://www.researchgate.net/publication/40768520_The_veloci... It has 2172!! citations! And in many(most?) cases these citations are pointless. It’s essentially used to pad out other ‘water is wet’ papers and/or hack researchers indexes. If it were a scientific article, the citations would reference it to use/improve/critique their hypothesis/model/data. I’m sure there’s a couple of these among the 2172 citations, but good luck finding the seed among the chaff! I think this stuff annoys me because of the huge opportunity cost. Climate change is such a big problem with so many critical questions unanswered, yet tomes of research get funded that tell us nothing about how to solve the problem. And worse: we get such volumes of drivel that serious researchers have to wade through heaps of nonsense to find quality papers.
- brutusborn 3y ago
  I decided to have a deeper look at this during my lunchbreak. I found a bit of a smoking gun: they claim that "probability distribution function of our temperature-based velocities are consistent with those described previously when uncertainty is accounted for (Supplementary Table 2)" If you look at Supplementary Table 2, it looks like their estimates are in line with existing studies (Malcolm 2002). If you actually look at the source of this comparison table (Table 2 Malcolm 2002), you will find that their estimates do not align well. E.g. for 0-315m/year they estimate 37.6/90/15 % for mean/lower/upper. The equivalent measure in Malcolm 2002 is 71.1 but has uncertainty of only +/-1.23%! So their model is "consistent" with existing models because 71.15 is within their range of 15-90%. Figure S23 looks even worse. So essentially they have proposed a model which doesn't really match existing literature and can't really be easily compared or used by others. They explain the differences away with assumptions but don't explain why their assumptions are better/worse. They might have produced a worse model but we don't have any data to tell! But maybe I'm completely wrong and this paper is super-useful to those in this field. I'd still have to see at least 1 paper actually use their model to be convinced.