In my current drunken state let me vomit some shit up for ya'all, cause I'm pretty sick of this.
Opened up the current Journal of Systematic Biology today. Back in the old days it was called Systematic Zoology, and the articles were pretty damn good, overall high quality with interesting and useful information about the world of what are species and what are their relationships. Today, the whole thing is bullshit. It's all computer programs, models, statistics, and all of it in distance-based maximum likelyhood and baetesian phenetics.
You see, back in the 80s systematists got ahold of DNA sequences, and more recently its become very easy to replicate and sequence a particular gene. This isn't a bad thing; nucleic acid base pairs are characters just like any other, they can be coded into a matrix and determined to have specially shared sequence parts. Just another tool in the tool box.
But what happened is these nearly dead pheneticists snuck into molecular phylogenetics with their overall similarity methods and called it "distance based". These just so stories with ad hoc assumptions and a priori models that were never made explicit created trees, trees that these people claimed, "Look, this tree looks like this and the DNA said so, so it MUST be true!"
Fuck them all. Not a day goes by that I open to what I hope will be an interesting article and find a maximum likelyhood tree, or a baetesian "inference", completely useless and completely untestable, a worthless pile of junk. This is LAZY systematics, is what it is. The crap methods have caught on because WAAAAAAA, PARSIMONY TAKES TOO MUCH TIME!~
Well, suck it. If you want good testable inferences then you have to use good testable methods which take time and effort to execute. This is not a Rudyard Kipling fairytale, THIS IS SCIENCE MOTHERFUCKERS. You get out of it what you put into it, and this bullshit is not acceptable.
So, I completely refuse to give any leeway to these people anymore. They can do things right or GTFO my science.
Also, fuck you Kai.
FUCK YES, KAI! :mittens:
Burns smells a letter to the editor??
Quote from: Burns on March 26, 2010, 03:12:10 AM
FUCK YES, KAI! :mittens:
Burns smells a letter to the editor??
Unfortunatly, no. They wouldn't publish it and it would probably ruin my career.
However, I can refuse to take their "research" seriously.
"Did you see that article last month?"
"I glanced at it, yes."
"Well, what do you think of it?"
"I think nothing of it, which is exactly how much science it includes, a big. fat. zero."
Thats the cool thing about being a good scientist in your field. I DON'T have to take their word for it. I can choose on good grounds to say that a particular paper, method, or journal is full of shit, cause I understand why.
As much as i like reading your stuff Kai, eventually I come to one of your threads and I'm "durhur" by the end of the second sentence.
I can tell that you feel strongly about it, though. And I liked it. So.... :mittens:
Sounds like this is a case of "Fuck accuracy, we want results."
I have almost no idea what molecular phylogenetics involves besides what you've mentioned in the OP, but based Wikiing a little on it, I think you've basically summed it up with
Quote from: Kai on March 26, 2010, 03:04:44 AM
THIS IS SCIENCE MOTHERFUCKERS.
If it's not useful or testable and there are alternative, better methods, there is no place for it.
GTFO Kai's science.
Also, fuck you Kai.
And I want to be clear, this is a matter of laziness. Using cladistic methods to analyze sequences is HARD WORK, and requires a computer run time of months, sometimes longer. It does the same thing I would do by hand, just testing all the possibilities to find the shortest tree length, of which there are probably sextillion or higher, a lot of fucking possibilities. In science, I get out in results what you put in, in effort, whether that means finding the most elegant method or the most robust results.
The reason this is so important is that tests of systematic hypotheses (the actual topology of the tree) is done by congruence of /new/ characters to the rest of the characters. Basically, I have a tree, and a new taxon within this clade. So I apply this previously unassessed character information to the old tree, and see whether they are congruent and supporting of the previous characters, or whether they are incongruent and non-supportive. If they are /truly/ incongruent, I can say I have falsified this hypothesis of relationships, and should adjust the hypothesis or throw it out and find a new one. This is science right there. I would be unable to test Maximum Likelyhood due to the ad hoc and non explicit insertion of models, but more importantly, it doesn't even use synapomorphy, just ALL similarity. Just like the pheneticists.
Again, they can suck it.
Mad Scientist Monologing, ITT.
And I'm loving it.
Rock on Kai! :mrgreen:
FUCK YES.
LEARN2SCIENCE OR GTFO.
:mittens:
Loved this rant.
I didn't realize that untestable claims were so rampant in biology.
That's the last place I'd expect that sort of bullshit.
Quote from: Ne+@uNGr0+ on March 27, 2010, 04:53:49 PM
:mittens:
Loved this rant.
I didn't realize that untestable claims were so rampant in biology.
That's the last place I'd expect that sort of bullshit.
Only in systematics.
Just the sort of thing I would do, choose the science that needs the most work.
Sometimes reading systematics journals can be depressing due to the overabundant use of just so stories (JSS). It's so bad that I get excited when I finally find where someone has actually included a testable scientific parsimony analysis.