Is Earth's Apparent Design for Life Simply the "Most Severe Case of Observational Bias in the History of Science"?
Meyer in Seattle: See Stephen Meyer in a Panel Discussion on Origins, Friday Night at University Presbyterian Church
Host a Worldwide Simulcast with Lennox, Meyer and Metaxas on "Science and Faith: Are They Really in Conflict?"
A milestone: there are now over 60,000 fans of NCSE's Facebook page. Why not join them, by visiting the page and becoming a fan by clicking on the "Like" box by NCSE's name?
Quote (sparc @ May 09 2014,03:36)Even without extremities gpuccio will still call it a draw:
Quote 17 JLAfan2001 May 8, 2014 at 7:27 am
19% functional is long cry from 80%.
Quote 18 gpuccio May 8, 2014 at 7:38 am
OK, but the ENCODE data still show activity for 80% of DNA. Nothing has changed. We will see how much of that is confirmed as functional in independent ways. Science must be patient.
edited to correct links and spacing
There's a nice review of the case for "junk DNA" in PLOS Genetics, which I'm sure won't be mis-interpreted by 'news' at all.
Quote (Wesley R. Elsberry @ May 08 2014,22:18)Gary:
I think that it is rude to demand a citation from someone who is only explaining how Watson works.
Of course there is no mention by star of "temporal correlation" being "the way Watson works" you are the lunkhead who mentioned it.
Here's how you can derive "causation" from Big Data: suppose you have two "factors", call them X and Y. You know, based on data, that X is correlated with Y. But you don't know whether X causes Y, Y causes X, or whether there is no causal relation between them. One example of how they can be correlated, but for which there is no causal relation is if there is a THIRD factor Z that causes both of them. You can render these possibilities as graphical models as:
X --> Y,
X <-- Y, and
X <-- Z --> Y
How can you decide which one it is? Well, you can disentangle the first two, in some cases, through temporal knowledge. e.g. Usually, X --> Y implies that X precedes Y; but the converse is not true. So, you can do some temporal reasoning to eliminate possibilities. And there are other ways to separate them; for example, you can mine text, and look to see how people speak about the relationship between X and Y. Individuals are maybe not to be trusted, but high-quality sources are -- and individuals, in aggregate, are, in some cases (and, you can even do deeper modeling and figure out when they are and when they aren't).
Now what about the third possibility? Well, you need some candidate Zs to work with. And how can you find those? Again, this is where data-mining comes in: you go to the web, and look for co-occurrence of (X,Y) with other factors Z. Once you've located a set of possible Z's, then you need to determine whether the third possibility holds. You can, for instance, ELIMINATE the third case, again, using temporal reasoning. And you can do even more elaborate things.
Gary, what "star0" discusses above *is* temporal correlation. The 2003 Nobel Prize in Economics went to Clive Granger, in part for his development of temporal correlation as the basis for his causality test. What "star0" did not assert was that IBM's "Watson" uses temporal correlation for detecting causality. In fact, "star0" makes no claim at all about any method that "Watson" might be using in the entirety of that post.
To sum up: "star0" does discuss temporal correlation, but does not discuss "Watson" in that post (thus "star0" was not "explaining how Watson works", at least not in the post Gary quoted and I was referring to).
Gary is 0 for 2.
"Lunkhead", when used by Gary, means "perceptive".
Explaining how Watson works was in a number of their threads. That information came from just one of them. This one does maybe go into more than what Watson has but I would not be surprised by it already having that as well.
Star0 used the phrase "temporal reasoning" when talking about "causation" and I'm not sure whether they would agree that "temporal correlation" is a direct substitute. It might be, but in this case the question is why David Ferrucci said “People are so enamored with the data-driven approach that they believe correlation is sufficient."
This evening I'll have some time to go over your leads.
And "lunkhead" was to go with earlier reply about it sometimes getting like Three Stooges in this thread. It's a bit of humor without personal insult like I was responding to that you hurled at me.
As silly as UD is, the denizens there are made to look like geniuses by some of the YECs out there.
Take this guy Nathaniel T. Jeanson, who claims to have a PhD from Harvard.
Quote We can't say that we are closer to, say, chimps than we are to yeast, because chimps and humans are equally distant from yeast.
Holy hell. You're doing it wrong. I promise you that despite all of us being equally distant to my great aunt, that I can prove I am more closely related to my sister than my first cousin. It is beautiful that in his presentation he has the data to do phylogeny right, then discards it, and from the resulting mess, claims to have "disproved evolution."
Here: http://www.icr.org/article...., he applies a molecular clock to mitochondria, multiplying millions of years or 6000 years by the mutation rate. The number of differences works better for the YEC model. He conveniently forgets mitochondria have small (20,000 base) genomes, and that he has predicted 2-3 million coding changes. Oops. (Not to mention mtDNA is probably not neutral, he picked the most rapidly evolving segment "D-loop" to get his molecular clock, and he never states what genomes he's comparing).
I take it back. This guy can't be that dumb. He's just lying.