Computer glitch hits climate prediction project
Tiny software mistake puts results back by months.
A software error has hit one of the world's most sophisticated climate simulations. Participants in the BBC-sponsored project, which uses spare time on home computers to predict Britain's climate in 2080, will have to wait longer than expected to see their work on television.
Read more here
Comments
What morons, to believe in computer models of so-called global warming.
Posted by: Joe Six-Pack | April 19, 2006 07:31 PM
When I first came across this on their own website, I did give a lot of thought to letting them use my PC as part of their project.
But as they could not promise 100% security to myself and of my course my computer. I decided against it.
It seems I had taken the right decision, as a few people that I keep in contact with round the world, say that they have had their security compromised. And now this....
I wonder if the BBC regret entering into this :-)
Posted by: John Vega | April 19, 2006 07:55 PM
AHH! Don't they realise that their assumptions about environmental modeling were wrong from the start?
There is much to consider with respect to environmental change. The ball seems to be in one court only at the moment. I hope someone is watching the other side of the court - and even more importantly - some one is watching the ball!
Posted by: Robert Neal | April 20, 2006 08:37 AM
I think this case ilustrates a general problem in computer modelling. How to KNOW the code is correct? When mistakes are big, they are notorious, but what if they are more subtle? I wonder if the problem of code validation is being addressed formally by research groups.
Posted by: Eneas Aguirre von Wobeser | April 20, 2006 08:43 AM
In his article, Michael Hopkin quoted Myles Allen as saying that 'the mistake probably happened...because of the complexity of the model'. This is not what Myles said. He specifically said that the model software was not at fault and that the error occurred because of an incorrect data input.
Myles' announcement to Climate Prediction crunchers can be read in full here:
http://www.climateprediction.net/board/viewtopic.php?p=42706
John Vega said that 'a few people that I keep in contact with round the world, say that they have had their security compromised'.
I am a volunteer moderator/admin on one of the Climate Prediction forums. Since the BBC modelling project started two months ago I have read almost all of the many posts. At the beginning, several potential or new crunchers brought up their security concerns and we responded to these in detail. To my knowledge, not a single cruncher has posted about any actual breach of security. In 2 1/2 years of climate prediction modelling, I am not aware of any security breaches at all.
Perhaps John's friends would like to post their concerns about security on one of the climate prediction forums so that if there is a real problem, it can be addressed immediately.
Posted by: Maureen Vilar | April 20, 2006 10:24 AM
I am not suprised by the title of this article. That is headline writers for you.
The more interesting question is whether there is any such thing as bad publicity?
The passage
'Reactions on the project's message board range from "I can't believe that this program wasn't completely tested before being released to thousands of people," to "Whoops! Still, that's science for you...."' seems rather selective to me. I have been impressed with the high proportion of people on the boards who have been very supportive of the team.
Most seem to realise that things like that can and do happen. A possible bigger problem than putting some results back by a couple of months is whether lots of people leave. That is one way a slightly bad situation could be made worse.
Anyway this article represents an opportunity to point out the work done was useful and encourage people to continue or join www.climateprediction.net.
Posted by: Christopher Randles | April 20, 2006 04:02 PM
I was a willing participant until I was advised by an IT chap at work that to run the CPU @ 100% 24/7 for 5 months solid was a sure to burn it out. When I tried to remove the program it had a spike that disabled driver installation and consequently I had to empty my pc completely and reload - this took about 3 days - I consider it a lucky escape and will never ever ever volunteer to give 100% security rights to untested software again.
Posted by: Gary McGilligan | April 25, 2006 10:00 AM
hi,
well i absolutely agree with the model runs that are being run in the climate models..but yeah at least..with the plausible scenarios at established with IPCC and the kyoto protocol are well understood..the climate impact due to global warming has several factors..and we, as in the scientific context have well understood the carbon imbalance that is responsible for the huge warming up. Considering the cooling effects, we have to still understand more and this is more related to the aerosol effects, which has a huge negative error bar..and which in the future can compensate a large part of the warming trend. also. with the possible emission scenarios, the climate models are doing so far a good job..modeling the carbon emission and concentration levels for the next 100-200 years .
Posted by: sangram ganguly | April 26, 2006 05:40 AM
think this case ilustrates a general problem in computer modelling. How to KNOW the code is correct? When mistakes are big, they are notorious, but what if they are more subtle? I wonder if the problem of code validation is being addressed formally by research groups.
good text
Posted by: kasim | January 15, 2007 11:07 PM