RC Correctness Censors
Posted by Jeff Id on July 31, 2009
This morning I received a comment from Laymen Lurker calling my attention to a conversation about moderation on RC. He indicated this post by Steve Fish discussing the non-response to Ryan’s post tAV to REALCLIMATE: YOU CAN’T GET THERE FROM HERE. A significant number of questions and attempts to post on RC were deleted from the current thread at the time regarding this post. It didn’t go unnoticed by those who regularly peruse blogland and is still being questioned by alert readers.
Steve Fish wrote:
Steve Reynolds #218. I checked out the Rank Exploits Blackboard link (same one I tried previously) and found your posts regarding censorship. You referred to a discussion on the Air Vent (tAV) as a worst case example entitled- “tAV to RealClimate, you can’t get there from here.” Several posters complained that posts on RC didn’t get through. These folks claimed that their posts were relatively simple and polite, and the discussion there was relatively low key so I have no reason to doubt them. On the other hand, some of their posts did get through and they, and examples of what didn’t, were so innocuous that I don’t see any reason to think RC would care, much less prevent them from appearing on their site.
What the whole issue was about was the RC article by Eric Steig, “On Overfitting,” and the following posts there by RyanO regarding his amateure (I didn’t see any credentials) reanalysis of Eric’s data that he presented on tAV. This was also a polite discussion and Eric gave RyanO a lot of help, encouragement, and advice regarding his desire to publish his version of the analysis. Eric’s inline responses were as comprehensive as many research article reviewers comments I have seen, so the actual topic under discussion was not controversial and comments were not heated. One of the posters complained in an Overfitting post about not getting through and Eric said he didn’t know why. Would this happen if there was censorship?
I have had one post to RC not make it and it was just a simple comment and question. I didn’t take this as censorship and just assumed that some of the glitchy behavior of the RC site (changes, spam filter, CAPTCHA) was responsible. When making a claim that information is being suppressed, one should consider what the information is and ask the question – to what aim? We all think that our own ideas are important and I suspect that some of the Air Vent guys were just hyperventilating a little. Steve Reynolds, like RyanO, post what you think is important and see what you get. Be persistent.
Since I cannot respond to Steve’s post on RC with any degree of certainty, I chose to do it here. Let’s start with the ‘credentials’ issue. The math of this paper isn’t the simplest thing in the world but it isn’t that severely complex. I’ve run into worse and typically do more complex math for my own job in optics. Ryan’s background is in physics and he’s also been exposed to very complex forms of math which likely exceed the detail of this paper. However, he, myself, SteveM, Ross M and Lucia are not climatologists so that may make us laypeople, however amateur is an unfair characterization of the level of work Ryan put forth. I understand that non-technical individuals cannot tell the difference in work quality of technical folk so it’s no offense for myself or I suspect Ryan, it just needs to be said.
Jon P put a short one or two sentence comment up on the RC thread which initially passed moderation. I thought I had a copy somewhere but can’t find it. It showed up in the thread and was deleted in minutes, I suspect a grad student passed it through and someone overrode the decision. Clearly the post had to be on topic and polite to be allowed through moderation the first time (it was), the thing we need to remember is there was a reason for the deletions.
Laymen lurker tried this comment which also didn’t make the cut for moderation.
Dr. Steig’s Antarctic warming paper is an example of recent work overturning conventional views Antarctic climate trends.
Ryan O, has just completed an alternative Antarctic climate reconstruction based on iterative truncated SVD rather than RegEM. Some objections raised by Dr. Steig on Ryan’s previous analysis included 1. the affect of calibrating AVHRR and surface station data and 2. overfitting due to inclusion of additional AVHRR PC’s. The latest analysis eliminates calibration and demonstrates improved verification with the inclusion of additional AVHRR PC’s.
So the question Steve alludes to is the correct one from above. Why wouldn’t RC take questions about Ryan’s work? It’s also the same question you should ask yourself as to why the RC thread in reply to Ryan was closed down so quickly. That thread had legs if you look at the number of comments which came in one day and a number of things were going on. I disagree with Seve’s interpretation of ‘encouragement and help’ but you have to understand the detail of what was being said to get that interpretation. Dr. Steig was polite and didn’t insist that I take any Matlab classes from him so it gave some hope for a reasonable discussion, however it was cut amazingly short.
Ryan’s work since then has been an even more dramatic improvement over the previous reconstructions which came through ever deeper understanding of the mathematics. He actually ended up re-inventing Dr. Beckers work on the DINEOF algorithm simply by making one improvement at a time until he converged to the same result. This happens to engineers more often than you can imagine but it also demonstrates the superior quality of the work Ryan is doing. At the same time it validates the work by Dr. Beckers.
Mann made claims of victory on another thread after the fact pointing out again that the verification stats Ryan presented weren’t correctly calculated, ignoring repeated explanations that everyone understood the nuance it had been done both ways and it made no difference. I believe Mann knows it would make no difference. In the end it resulted in another post which has been the topic of this thread.
Ryan’s work is incredibly clean, you can understand it by reading the code which is cleaner than RegEM, works better than RegEM for most things and is being thoroughly vetted by him and NicL in the background on a daily basis. Nic is another professional of some sort who also does careful fantastic work.
So as a professional non-climatologist (thank god!) I’ll answer the question above now. Why doesn’t RC answer?
The reason the posts were deleted is very clear to me…. Ryan’s right!
Everything points to the same answer, better verification statistics, good match to simple area weighted methods, no Chiladni patterns being mistaken for weather. Everything about it points to the same thing. If you want to understand why RC won’t let comments about this be discussed, just ask yourself:
What do they have to gain if Ryan’s right?
and the opposite question as a test of the theory,
What would they have to gain if Ryan’s wrong?
There are many excuses they can use for not replying to work that is an improvement on their own (i.e. time, publish it, not a climatologist,… whatever) however, which of the above fit’s the facts?