From	Phil Jones	Date	Fri Feb 6 10:58:17 2004
To	Peter H. Gleick, Mearns Linda O
CC	Stephen H Schneider, Nigel W. Arnell, Christian Azar, d.camuffo@isac.cnr.it, scohen@sdri.ubc.ca, pmfearn@inpa.gov.br, jfoley@facstaff.wisc.edu, harvey@geog.utoronto.ca, ahssec@ansto.gov.au, Thomas R Karl, rwk@ucar.edu, rik.leemans@wur.nl, diana.liverman@eci.ox.ac.uk, mccarl@tamu.edu, lindam@atd.ucar.edu, Richard Moss, ogilvie@spot.colorado.edu, pfister@hist.unibe.ch, Barrie Pittock, pollard@essc.psu.edu, nj.rosenberg@pnl.gov, Cynthia Rosenzweig, j.salinger, Ben Santer, Schellnhuber (E-mail), dgvictor@stanford.edu, F. Ian Woodward, Gary Yohe, yurganov@hotmail.com
Subject	Re: MBH Submission (fwd)
Dear All, So now it seems that we're separating 'providing the code' from 'running the code'. I can't see the purpose of one without the other. Even if Mike complies I suspect there will need to be several sessions of interaction, which neither side will be very keen on. As I said before I know the code will involve lots of combinations (for different periods with different proxies). Also I would expect, knowing the nature of the PC-type regression approach, that there will be library routines. If the code is sent, there needs to be conditions. We don't want McIntyre (MM) to come out and say he can't get it to work after a few days. So, it is far some simple. I'm still against the code being given out. Mike has made the data available. That is all they should need. The method is detailed in the original paper - in the online (methods) and also in several other papers Mike has written. As an aside, Mike is now using a different method from MBH98. Also, as an aside, whilst we've been deliberating, MM have submitted another comment on MBH98 to another journal. In this they say they have a program that replicates MBH98 (although it isn't very convincing that they have it exactly right, as they never show a like for like comparison) , but most of the comment goes on about the results being different due to different combinations of proxies. The latter isn't surprising. It might appear they want the code to check whether their version works properly. If this is the case, then there are issues of IPR. So, if they get the code, how do we stop them using it for anything other than this review. Cheers Phil At 11:40 04/02/2004 -0800, Peter H. Gleick wrote: Yes, excellent point. This should be what we do. Further, we can point out that we've bent over backward here and provided more than typically necessary in order to satisfy persistent but inappropriate demands. Peter At 08:46 PM 2/4/04 +0100, Mearns Linda O wrote: Peter et al., Thanks for reminding me about the new email list. My point about the code is still that 'providing the code' can be interpreted alot of ways. I have thought about this, and imagined if in one of my larger and more complex projects, I was asked to provide all code. I could do that just by sending the pieces with a summary file explaining what each piece was used for. It still theoretically allows someone to see how coding was done. And I do think that is a far sight easier than providing stuff that can be run, etc. I am suggesting that one could do the minimum. Then the point is, one isn't faced with garish headlines about 'refusal to provide code'. I think it is harder to come up with a garish headline about 'refusal to provide completely documented code with appropriate readme files and handholding for running it'. Linda Dr. Peter H. Gleick Director, 2003 MacArthur Fellow Pacific Institute for Studies in Development, Environment, and Security 654 13th Street Oakland, California 94612 510 251-1600 phone 510 251-2203 fax [1]www.worldwater.org (World Water site) [2]www.pacinst.org (Pacific Institute site) Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email p.jones@uea.ac.uk NR4 7TJ UK ---------------------------------------------------------------------------- References 1. http://www.worldwater.org/ 2. http://www.pacinst.org/