I very much agree here. Have recently ran part of the data through some NLP (incl. AWS Comprehend) and nothing signficant came out. Ended up doing simple free-text or keyword search and only landed on one interesting finding so far: https://twitter.com/yazijys/status/1240465780715683841
Been compiling this list à la patrickcollison.com. The core goal of this is to consolidate good practical and scientific knowledge on the topic in one easy-to-access place/repo, especially that there seems to be a lot of noise currently. Would love to get [critical] feedback from HN and/or connect with anyone who's interested in helping out.
Not sure if it's a direct causal relationship. dang can better tell.
The graph in this case ended up being just a base layer for some interesting internet stories, while paying some tribute to HN for the great place it is.
I don't think the problem should be approached from a "great market" perspective. The market for Hacker News is quite niche. Yet, it has become a great community for the curious of us. Almost hardly replaceable by any of the communities/websites who have great market potential and wider audience. So niche is probably a good thing here.
Driving a critical discourse (on something like Health or Hacker News) between scientists, journalists and the general public can potentially create more appreciation for scientific progress (which might often be dull and slow indeed) and provide a feedback loop for how science is/should be communicated or even done.
tldr; I'd like to get in touch and do a deeper dive into the problem so that the solution can be refined. How can I reach out to you?
> I am a little surprised, given the title, that the issue being highlighted is blogspam and the brokenness is basically that science journalists can't fairly capitalize their effort.
I think this was slightly highlighted later in the post but agree that it wasn't stressed enough.
> Science journalism is broken not because spam blogs copy-paste articles. It's broken because the entire ecosystem has the wrong incentives.
I think the two are closely related; the copy-paste behavior contributes greatly to minimizing the window of capitalization. However, it's definitely not the only factor.
> Researchers, many of whom have inadequate statistical training, are academics facing publish-or-perish and weird publication barriers (p-value filtering, premium on sexy new results over well-researched incremental advances). The more senior authors who have tenure and are safe from some of the incentives still rely on lab funding, which is similarly competitive, and also tend to people whose methodological training is substantially out of date. Most journal reviewers are subject matter experts, some with limited statistical training.
That's basically a problem of science, perhaps at the core of it, but not science churnalism I think.
> I see very little evidence this process stems from the article's implied undertone that blog spam is costing legitimate journalists money and starving them of resources they need to do a good job.
You don't think aggregators that built some traction (i.e. Phys.org, MedicalXpress) are taking away opportunities of good journalism? I don't have quantitative evidence for this but the qualitative one is pretty strong (talking to journalists, for example).
> There is no notion of including comment from critical authors, and there is certainly no notion of critiquing the design, replicating the experiment (either a limited replication using the author's code / data, or a more thorough replication to obtain the substantive result)
[...] The problem is the incentives. Imagine a science website that posts only 1/10th the number of posts, but deep-dives all of the ones they do.
[...] This should be especially followed in disciplines like social psychology, applied economics, neuroscience, evolutionary anthropology or evolutionary psychology, medical research, nutrition research, and other fields where experimental or quasi-experimental design are more difficult.
Precisely the raison d'être of something like Health News. I guess this was pointed out near the end of the post. Although realizing it will take some time and traction, it looks like a good starting point.
The URL to Scripps Research is obviously working now, after the story got its largest share of public attention from other websites. Truly disappointing.
Was chatting about this with someone on Health News. I doubt it's lucky - the two compounds have been examined together in several previous studies; https://www.nature.com/articles/hr2005119