Did I miss the thread about the CRU Fraud?

[BlackCat]

Can you summarize it, BC?

Summarizing a poor sods multiple year agony of trying to reproduce research results ? Well, I'll try

What version of the progam is needed to run.
Which library version is the rigth.
what files are needed.
wich parameters shall I use.
Why does the programs manipulate the data how they do.

That is what I got out of skimming half a dozen of pages.

[KrazyHorse]

So....it took him a while, but he finally managed to reproduce the results?

[KrazyHorse]

If that's the case, then it sounds like what I'd have to do to reproduce most phenomenology papers' results.

[BlackCat]

EDIT: in other words, what's the key part that I have to read in order to demonstrate that the climate people are doing stuff which isn't adequately described in their papers?

Well, I don't object against climate alarmists when they produce their results, I just find it a bit worrying that their data handling method isn't documented. If that was described in their papers, there wouldn't have been this frustrated programmer.

[KrazyHorse]

Dude, yes there would have been. There is a pressure to make papers shorter in order to convince people to actually read/publish them.

If somebody were to read one of my papers they could probably manage to get something reasonable within a few months, but to get my exact results would take a very long time.

[Lorizael]

Just a minor question assuming you don't use computers - do you need to provide a description of how you manipulate data and why ? I think that you would have to do such before the invention of computers.

Can you please explain why such isn't any longer nessecary after the invention/use of computers ?

Further, how can use of "shoddy" models be considered scientific ? Well, you claim that it's because it's done by a computer, but would you accept such if it was done by hand ?

Data "massage" do make sense, but viewed scientific, wouldn't that demand that you have a precise description of when and why you do it ? If for nothing else, to remind why you did it ?

Honestly, what you have described sounds more like data manipulation than science.

I'm a bit dissapointed that KH apparently agres with you on this, but maybe I have a wrong perception of the scientific method.

I honestly don't know what questions you're trying to ask me here, but the reason "shoddy" was in quotes is because I don't actually believe the models are shoddy. I believe that the programming may be shoddy, in that it doesn't meet commercial standards and such, but that doesn't mean that it does not accurately interpret the data.

No, that is not the case. It's true that it has been reproduced, but only by some that can be expected to find the same. The data hasn't been available for scientists that doubt the primary results wich is against the scientific method.

I'm not sure it's useful to generalize the specific instance of CRU avoiding FOIA requests to the whole climate science community being evasive or something like that.

Who really cares why they want acces ? That is totally irellevant - either your data and model can hold against atttack or it can't. You can't claim that the "skeptist" have a wrong opinion or are "paid by oil" - either your research are sound and can take the beating or the theory are wrong.

What you really are talking about is that opponents can't be allowed acces because that will disturb the political process.

You can claim that certain skeptics aren't actually scientists. And in many cases, that's completely accurate. Non-scientists shouldn't be allowed access to data, because there's nothing useful they're going to be able to do with it.

[BlackCat]

So....it took him a while, but he finally managed to reproduce the results?

If I'm getting it right, it took him three years. Though, that isn't really a proof of validity - just that he made the same moves as first run.

IMHO, you can't claim that you have used the scientific method if it's nessecary to have some guy guessing and trying to reach the same result for several years.

[KrazyHorse]

BC, you have no idea what you're talking about. ANY scientific paper in which there is a lot of data manipulation is not going to be straightforward to reproduce exactly. The key is how dependent the broad conclusions of the paper are on the exact manipulations you did.

[BlackCat]

Dude, yes there would have been. There is a pressure to make papers shorter in order to convince people to actually read/publish them.

If somebody were to read one of my papers they could probably manage to get something reasonable within a few months, but to get my exact results would take a very long time.

Just out of curiosity - wouldn't you want others to verify your findings ? Yeah, if you suspect that you migth find more, then you may want to hide it, but your data would actually be a part of what you publish.

[KrazyHorse]

My raw data CANNOT be part of what I publish. It takes up several terabytes. I can't even store it; I manipulate the "data" as I simulate it, then perform one of several transformations on the data, then store the results (which only take up gigabytes)

I condense it down to a table of 50 numerical values or so for publication.

[BlackCat]

BC, you have no idea what you're talking about. ANY scientific paper in which there is a lot of data manipulation is not going to be straightforward to reproduce exactly. The key is how dependent the broad conclusions of the paper are on the exact manipulations you did.

KH, that is idiotic. You can't make any claims on data you have manipulated unless you can document how you did it.

[KrazyHorse]

You document IN BROAD STROKES how you did it. This doesn't mean that you can reproduce EXACTLY what you did.

[KrazyHorse]

Holy crap, you are so wrong it's hilarious. Get to know some scientists and find out what they do in detail compared to what they publish.

[BlackCat]

My raw data CANNOT be part of what I publish. It takes up several terabytes. I can't even store it; I manipulate the "data" as I simulate it, then perform one of several transformations on the data, then store the results (which only take up gigabytes)

I condense it down to a table of 50 numerical values or so for publication.

Well, then compress it

Nah, I don't mind scrapping data if they are replicable, but climate data for the last 40 year or so isn't in that category. No problem in building a new collider and remake an experiment, but it's impossible to make new temperature measurements backing 40 years.

[KrazyHorse]

Dude, I don't tell people everything they need to know to reproduce my **** exactly. Nor will the CERN people give out their raw data. Nor does any other experiment give out its raw data.