Comment by jjoonathan | Hacker News Reader

Comment by jjoonathan | Hacker News Reader

Parent Full thread

The glyphosate team is a bad example. They deserved every bit of the criticism they received [2]. In fact, I'd go so far as to Monsanto's criticism was highly understated and generally an outstanding example of scientific discourse. I suspect that they (the team that found negative effects) didn't get the results they wanted so they came up with an implicit null hypotheses so shitty that they practically guaranteed p<.05: they first did PCA on a 48-dimension dataset and then did significance tests along the principal components. I'm not a statistician, but that methodology is about as dubious as it gets. You should read it and decide for yourself before using it as an example.

It's a pity that this kind of behavior is par for the course in the category of "controversial environmental findings," because there are plenty of legitimate questions to ask and legitimate issues to raise.

I have no relation to Monsanto, except that I eat their products. We talked about this paper [1] in journal club and were unimpressed (except by their criticism of Monsanto's methodology, which was, ironically, spot on). I wasn't aware of Monsanto's formal response [2] at the time, but from a brief glance it appears to hit the nail on the head.

[1] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2793308/

[2] http://www.monsanto.com/newsviews/Documents/SpirouxdeVendimo...

Well, and this is the very problem with these tactics and the very reason why they are so thoroughly evil.

I do not have the ability read a scientific paper and judge its merits. I'm a software engineer. I can usually get the gist of what the experiment did, what they found, etc, but not judge the merits.

The fact that this amount of scientific critique can be bought, for someone like me, calls into question any scientific discussion anywhere. Because I don't know who to trust to tell me whether a paper or experiment is good or not.

My natural inclination would be to be suspicious of Monsanto and side with the glysophate paper's author. After it was so thoroughly discredited, I believed the critiques of the paper. However, reading about the methods that were used against Tyrone Hayes, I'm now thoroughly questioning that paper's being discredited. Because the manner in which that paper is being attacked looks so similar to the manner in which Tyrone Hayes was being attacked.

Do you see what I'm getting at? It ruins the trustworthiness of scientific discourse for anyone not a scientist. And that's an enormous problem.

What source of information in life isn't like this? You have to trust the news, the history books, everything else. You make a fair point, but you can't simply just say the two studies can both be questioned because of the manner they are attacked.

You don't judge the the paper based on how it's being discredited. You try your best to judge the paper on it's own merits. If you lack the knowledge/information to do so, then you obviously have to live with the fact that your opinion on the matter may not necessarily reflect what is true. It's a tradeoff many of us make considering we can't be experts in everything. If it's something you truly believe in and want a better opinion on...then you have to spend time on doing research into it.

Everyone verifying everyone is O(n) work per person or O(n^2) total work. I think it is possible to do better than that. Web of trusts for instance.

It doesn't have to be everyone verifying everyone - just that a reproducible result tends to be more trustworthy. This is why reproducing results from other studies are also valuable scientific endeavours.

But you know, just verifying that it is reproducible takes time. Ideally you would have a system that enable you to trust everything you read with minimal work. Like if I claim the world is round you would know from my awesome reputation / fancy title / little green WoT indicator, that you can take that at face value.

In this case the poster tried to use fancy title, but found that it didn't work satisfactorily.

> The weight of people -- supposedly credible scientists -- that they were able to bring to bear to speak against him calls into question effectively every controversial scientific finding.

Since you claim not to be a statistician, pardon me if I also ignore your unsubstantiated opinion that the null hypothesis was "so shitty that they practically guaranteed p<.05".

Hey, you don't need to be a statistician to understand the "data mining" fallacy. If you look at 100 different null hypotheses, and you set p<.05 as your threshold for rejecting, then you're going to find ~ 5 ones you can reject at p<.05. Duh. Monkeys typing Shakespeare and all that.

Reply on news.ycombinator.com