And it's sad that DevOps tools like this are tied to a specific language, Puppet especially. I suppose the engine has to be written in something, and the use of classes I guess lends itself to CM somewhat, but I think the learning curve cost is high. When you start wanting to do complex things with Puppet, it starts to degenerate quickly if you are not careful with the manifests->templates->heira to dereferencing variables in the correct context and yadda, yadda, yadda. Although it's simple to write puppet modules with barely any Ruby knowledge at all, it's dependence on Ruby a steep price of entry. Python's whitespace syntax is just irritating at any level. I thought that shit was left behind in COBOL. And yes, I'm a neckbeard.

It's funny because it's true...

So, scientific papers shouldn't try to do the best work they can before publication? The article(to me) seems to insinuate that just like "news", paper publishing is rushing to get out regardless of the claims. True, this sample size is no where near what it needs to be to imply anything factual, but a 10% verification rate seems a little bit more than "a few slipping through." I guessing if you did a follow-up study of how many of the unverified papers had follow-on studies, the rate would be alarmingly low. So, we end up with a lot of published scientific papers that people can point to that are inaccurate and kicking the verification can down the road. The just seems bad all around.

So, this information source is infected the same as any other information source today. No one cares if they are right or true anymore, just if it gets views. I think we all know most science isn't sexy...