Evaluation of the some code by a programmer at cubeantics.com

Follow up

Cliffs: Program will make a straight line into a hockey stick.

Edit: need to read part 2 first
Edit2: Okay, so the programmer seems to think that this doesn't change the implications of this algorithm, but in part 2 he does confirm that the "adjusted" graph never actually got published. The code is commented out and the adjust data never actually got published -- it's entirely plausible that whoever wrote the program was simply testing something.

In the civilized programming world, one would use the revision control system's blame function to see when the code was introduced, and read the checkin comments to know what the checkin accomplished. As played, the code doesn't offer any hints about its purpose.

lol you didn't read part two. the code isn't commented out elsewhere, and you have no idea if the code was published or used.

In part 2, he explicitly states that there is no proof that this data was ever published:

"NO PROOF EXISTS THAT SHOWS THIS CODE WAS USED IN PUBLISHING RESULTS.
Correct! That’s why I am (and always have) taken the following stand: Enough proof exists that the CRU had both the means and intent to intentionally falsify data. This means that all of their research results cannot be trusted until they are verified. Period."

So he offers some flimsy reasoning for why the mere existence of some adjusted data is enough to damn this research, but he acknowledges that the data didn't actually get published.

Edit: However, you were right that similar code appears uncommented elsewhere. Of course, this isn't a particularly important point if the adjusted data never got published.

You don't read well. He doesn't know if this was used in published material because the CRU has repeatedly refused to published what they used. Citing the lack of open science as something that is the authors fault is absurd.

lol. From his piece:
wat? His post reminds me of your "look I found a hockey stick!" post.

It'll only become funnier if you read the comments. The argument quickly degenerates into "no, I don't have any evidence that the data was published, but it's a hockey stick and it's really peculiar that hockey sticks appear in their data a lot."

Hey guise! 13 found something he can argue against!

He'll ignore everything else but not address the main issue brought up.

The "main issue" is that someone else (who claims--surprise!--to be "agnostic" on the issue) is yet again claiming fraud with no evidence. Scientists plug in numbers all the time just to see the effects. Until you have proof that someone published bogus numbers without a valid scientific reason then you have nothing.

Not that you care about the science or reality.

You are mistaken. If you'd like to know where the program "briffa_Sep98_d.pro" is used, you should probably look for a paper that Briffa published in 1998.

Hint: http://rstb.royalsocietypublishing.o...nt/353/1365/65

Keep me posted for any suspicious-looking hockey sticks.

When was that program's source code published?

Well, is that routine used currently? It exists without any hints in the code about its purpose. That's not really a good thing. (I much prefer source code comments about why than about what). Do the users of this software know what their own code does?

Offering anonymous read access to their source code archive seems like a really good idea. When dealing with black boxes, I think of this:

Did you read the article? Was there something in the article that you thought wasn't adequately supported?

You're going in completely the opposite direction. Hmm, here's some code Briffa didn't publish (and, lucky for the skeptics, he helpfully added that the data correction in the code was "VERY ARTIFICIAL" -- the usual mark of someone committing fraud). Maybe it's relevant somewhere. Until I have clearer evidence, I'm going to assume that it's relevant and NEFARIOUS!

This little tempest in a teapot is a great example of why climate scientists might be reluctant to release their source code.

Being concerned about black boxes is entirely reasonable. However, the mere existence of some code that does something mysterious isn't an indication of anything sinister. Considering these were files hacked from Briffa's computer, they could have been simulations ("what ifs" are often useful, as 13ball points out) or something else that never appears in published work.

Now, in this case, I think that code was used in a paper, namely the one I linked to ikestoys above. However, the "corrected" data that you see in the code doesn't appear anywhere in that paper as far as I can tell.

Again, I want to reiterate that "black boxes" are something you should be suspicious of. If there was a completely unjustified (and seemingly convenient) adjustment that appeared in the published paper, you'd be right to look into the matter further. But otherwise, you are coming at this from the wrong direction.

So the code wasn't published? How do you know it wasn't used if the code wasn't published?

I don't *know* if the code was used, and that's a silly standard to use in the first place. All I can say is that I see no evidence that the code was used in any published paper. In fact, the only published paper that seems to fit this code specifically omits the "corrected" data. Now, what evidence do you have that this code was used?

You are contradicting yourself now.
So what did they create the code for? ****s and giggles? There's a definite possibility that it was used. The obvious rebuttal is to release data and methods like any other scientist would if their work was questioned.

No, and I'm not responsible for your lack of reading comprehension.
We've gone over this already. It is often useful for scientists to consider "what if" scenarios -- it's common practice to try making an adjustment to some data and seeing what happens to your model when you do it.

I'm not a climate scientist, but it seems to me that the reason behind this particular adjustment is very well documented ("the divergence problem"). In short, the tree-ring temperature data diverges from thermometer temperature data post-1960. The code you posted seems to perform an analysis on the unmodified data and then also perform the analysis on the artificially corrected data for comparison.
The overwhelming majority of the data has been publicly available for years. You're going to have to be more specific: what data is missing?

I mean, its the internet, it's right there.

Where was the code of this Briffa paper released before the FOIA2009.zip became available?

I emphasized the word "know" for a reason. My point is simply that there's obviously no way for me to be sure that the adjusted data wasn't ever used in any published paper -- however, the evidence clearly suggests that the adjusted data was not used in any published paper. The programmer whose blog you linked to agreed to as much. I guess I was slightly imprecise about how I worded it, but that's being pretty nitty, especially considering I've since clarified what I meant.

Now, if that's really your biggest objection to my posts (and it seems to be, considering you've opted to ignore most of what I've written), I'll assume that you don't have any evidence that the adjusted data was used in any published papers.

We do not know if the algorithm was used because they won't release what algorithms were used in their paper! The fact that you put the burden of proof on me when the methods are not released is absolutely absurd. I mean, wow.

The burden of proof is usually on the accuser. Just because you found some code doesn't mean they used it. Sure, they didn't release code for some of the stuff they wrote but that doesn't prove they used it and it should be up to the people saying it was used to prove it somehow instead of just asserting that's the case.