Buy

Books
Click images for more details

Support

 

Twitter
Recent comments
Recent posts
Currently discussing
Links

A few sites I've stumbled across recently....

Powered by Squarespace

Discussion > Where is Rhoda's Evidence? (plagiarised by Dung)

You have got me all wrong Martin A. I learn by watching people who know more than I do discuss things. I have massive respect for you in every way. In quoting EForster I am not trying to speak from authority I am trying to watch you argue with his statement so that I can learn more sir

Nov 14, 2015 at 8:21 PM | Registered CommenterDung

TBYJ
I wonder if the problem is that, just as "greenhouse effect" is a useful shorthand which is scientifically not absolutely correct so the idea that the direction of travel of radiation is from hot to cold may not be scientifically accurate (if you're what I might call a scientific pedant) but the net effect is that the warmer object does not get warmer in that situation and as far as the layman is concerned the direction of travel is from hot to cold.
The fact that any object above absolute zero is emitting radiation really is irrelevant in the context of the end result surely.
If the earth is warmer than the atmosphere above it then the earth will cool. The rate it which it cools depends on the extent to which water vapour and these 400ppm of CO2 slow down that process but cool it surely will.
If we actually want to make progress in the increasingly irrelevant science of global warming (I doubt more than about 10% of those heading for Paris at the end of the month give a damn about the science; we're past that stage) perhaps it is time to look closer at the hypothesis that the IR bands at which CO2 absorbs are "over-subscribed" at 280ppm let alone 400 and that the additional CO2 released in the last century is not the cause of waming.

Nov 14, 2015 at 8:41 PM | Registered CommenterMike Jackson

Martin A

You have an odd idea of impossibly complex.

You can define all the main parameters of Earth's climate with thirteen flow parameters , and seven energy reservoirs (land, atmosphere, clouds, freshwater, ice, biomass and ocean.)

Your telephone network is probably more complex.

The difference is that you designed that network, controlled it and thought that you fully understood it.

Nov 14, 2015 at 8:44 PM | Unregistered CommenterEntropic man

EM

You forgot the sun, I know its only a small thing and so I am nit picking.

Nov 14, 2015 at 8:54 PM | Registered CommenterDung

IIRC Hansen's 1984 prediction of a 1C rise between 1880 and 2010 was almost spot on.

Nov 14, 2015 at 11:41 AM | Unregistered CommenterEntropic man

Well, I can correctly "predict" most things you care to name and measure between 1880 and 1984, and raise you to 2015.

But those would not be predictions in the normal sense of how the English language is employed. So, in the sense of time, most of what you recall Hansen was predicting (projecting?) is irrelevant.

In terms of amount, ~1/2 of the warming had already happened by then, so he wasn't predicting that either.

That leaves the remaining 1/2. Given the large variance in the data from year to year, and choosing your start finish dates, you can make Hansen's predictions look better or worse, according to taste. Selecting 2009 or 2012 make a large embarrassing dint in the 'predicted' data which was not exactly pretty to begin with. It's uncertainties, error bars, and large natural variations again, EM. Again.

But the main thing your are probably failing to remember correctly is that, even allowing for using suspect data-sets such as those Hansen was himself in charge of, the warming came in at the very low non-alarming end, while emissions were at the high end of his scenarios. A serious model fail. Again.

Nov 14, 2015 at 9:00 PM | Unregistered Commentermichael hart


You have got me all wrong Martin A. I learn by watching people who know more than I do discuss things. I have massive respect for you in every way. In quoting EForster I am not trying to speak from authority I am trying to watch you argue with his statement so that I can learn more sir
Nov 14, 2015 at 8:21 PM | Registered CommenterDung

Do you mean BYIJ Dung? - he spoke about you arguing from authority - not me.

I did say Dung, dunno who E Forster is but what he says is dodgy - it is either unclearly expressed or it is outright wrong. It sounds suspiciously as if he may well be a member of the Brotherhood of Dragon Slayers.

There seemed to be many people who talks about "back radiation nonsense". And E Forster appears to be one of them.

I think that explaining the GHE in terms of back radiation has lead to huge confusion. Whenever I read such an explanattion of the GHE, i am struck by the impression that the person attempting the explanation has not understood physically what is happening. (All the warming that happened was

Nonetheless back radiation from air to ground does exist when there is a GHE and anyone who thinks it does not exist falls in the category of Skydragon Slayers.

Nov 14, 2015 at 10:09 PM | Unregistered CommenterMartin A

Martin A

Very sorry for the confusion ^.^

Nov 14, 2015 at 11:57 PM | Registered CommenterDung

Yes, EM, you had already made it clear that you think the Earth's climate is really quite simple and can be adequately modelled as a small number of interconnected containers of liquid.

Nov 15, 2015 at 8:37 AM | Unregistered CommenterMartin A

Sorry about the delayed reply . I got 403'd.

Dung

No problem.

I did not forget the Sun. Solar insolation is one of the thirteen flow parameters.. Look again at the upper centre of the energy flow diagram I linked, labelled Incoming Solar Radiation.

Martin A

In practice it is even simpler than I described above

Hansen's first model used just two containers, atmosphere and ocean. It failed completely, of course. It predicted that 1C warming since 1880 would occur in 2010 instead of 2015.

Since the ocean has 96% of the total heat capacity of the climate system, you can still get explain most the climate using interactions between atmosphere and ocean.

Nov 15, 2015 at 12:54 PM | Unregistered CommenterEntropic man

And there was I thinking that even the Met Office's million lines of Fortran represented a laughable oversimplification of the reality.

Nov 15, 2015 at 1:34 PM | Unregistered CommenterMartin A

Martin A

How many lines of code does it take to run a telephone network?

Nov 15, 2015 at 1:51 PM | Unregistered CommenterEntropic man

Martin A

Serious question, since I have no recent IT experience.

What can you do with a million lines of code?

Nov 15, 2015 at 5:40 PM | Unregistered CommenterEntropic man

Gosh EM - I've got simply no idea.

No doubt orders of magnitude today than in the 1980's and an awful lot more then than in the 1950's.

And it would obviously depend on what you meant by "running a telephone network". Keeping track of customer accounts, voice messaging, billing, marketing, equipment purchase and installation, management of operations staff ...? Or just call handling in switches?

Nov 15, 2015 at 5:54 PM | Unregistered CommenterMartin A

EM, if you posit that every phone call generates 8 call detail records that have to be stitched together to form a billing record. So assume a CDR is 10 bytes. Then add the database that links a phone number to a billing record and then to a call plan. Along with details of payments, complaint calls, etc. And marketing offers. Throw in text messages handled by a different system and calls made and received in foreign countries. That’s a lot of code.

Nov 15, 2015 at 6:20 PM | Unregistered Commenterdiogenes

EM - I missed your 5:40 follow up. 1,000,000 of code for a scientific program is just huge.

I have always thought 'lines of code' is a pretty useless measureng on all sorts of things - depending on programming language used, field of application, quality maturity of the organisation, and other things . But for comparing scientfic programs written in Fortran, 'lines of code' might be less than completely useless.

Surveys have suggested that professional programmers produce somewhere from one to ten lines of code per work day. The Met Office programmers are climate scientists of one sort or another, rather than professional programmers, but if we guess that their output would be about five lines of code per day working in their programmer role, that would mean the equivalent of roughly 900 programmer years. A collosal effort.

[1,000,000 lines of code = 200,000 programmer days = 877 programmer years. (assuming 228 working days/yr for civil servants)]

Nov 15, 2015 at 7:09 PM | Unregistered CommenterMartin A

colossal

Nov 15, 2015 at 7:10 PM | Unregistered CommenterMartin A

I suspect that 5-10 lines of code is an older metric. The project I'm current working on produced 1.4 million lines of code in a little over a year and that's not a particularly complex application. Remember that modern languages come with all sorts of library and stubbed out pregen code which do not need to be written from scratch.

As you said 'lines of code' is not a good metric and was probably only good when people write machine code for a living.

Nov 15, 2015 at 8:58 PM | Unregistered CommenterTheBigYinJames

As you said 'lines of code' is not a good metric and was probably only good when people write machine code for a living.

Yes.

Or Fortran, which the Met Office unashamedly say is what their model is programmed in.

Nov 16, 2015 at 7:01 AM | Unregistered CommenterMartin A

Martin A

The nice thing about Fortran is its transparency. It can be written by one scientist and then understood by other scientists and peer reviewers.

If you use an IT professional to write the software it becomes opaque. Input goes in, output comes out, but the scientist has no way of checking its validity. Nor can the software be debuggd or developed by the user.

Nov 16, 2015 at 9:55 AM | Unregistered CommenterEntropic man

There's no point in having a code war here. Fortran wouldn't be my first choice to write something now, but back in the 80s, it was the de facto language for coding algorithms.

You can write opaque code in Fortran, and you can write readable code in any modern language. You might argue it is easier to write spaghetti code in the older languages, and because of the lack of referencing structures, easier to succumb to the temptation to hard-code in parameters. But that would be down to the programmer.

Criticism of the actual code is fine, but there really is nothing to criticize or praise in the choice of language.

Nov 16, 2015 at 11:21 AM | Unregistered CommenterTheBigYinJames

The nice thing about Fortran is its transparency. It can be written by one scientist and then understood by other scientists and peer reviewers.

If you use an IT professional to write the software it becomes opaque. Input goes in, output comes out, but the scientist has no way of checking its validity. Nor can the software be debuggd or developed by the user.
Nov 16, 2015 at 9:55 AM | Unregistered CommenterEntropic man

EM - here is a paper on the Met Office's software develpment http://www.cs.toronto.edu/~sme/papers/2008/Easterbrook-Johns-2008.pdf

Same link, non-clickable: http://www.cs.toronto.edu/~sme/papers/2008/Easterbrook-Johns-2008.pdf

From what you said in your comment, I'm not sure we would be on the same wavelength in a discussion on how to specify, create, and test large computer programs of very high quality.

Nov 16, 2015 at 11:57 PM | Unregistered CommenterMartin A

Martin AA


"I'm not sure we would be on the same wavelength in a discussion on how to specify, create, and test large computer programs of very high quality."

Indeed. You do not strike me as the sort of person who would fit into the Hadley Centre environment.

Nov 17, 2015 at 11:00 PM | Unregistered CommenterEntropic man

I take that as a compliment EM though I'm not sure you meant it that way.

...For example, errors that cause instabilities in the numerical
routines, or model drift (e.g. over a long run, when conservation
of physical properties such as mass is lost) can be complicated
to remove, and so might be accommodated by making periodic
corrections, rather than by fixing the underlying routines.

Nov 17, 2015 at 11:54 PM | Unregistered CommenterMartin A

EM

On various other threads and here you appear to misrepresent what engineers software and otherwise do.

An IT professional covers a wide range of things. With regards to a software engineer working on anything safety critical rigid version control is mandatory. As is clean design and full traceability. Code written by these IT professionals is going to be way easier to read with comments than some internal program.

You forget that some of us have done development and verification so have seen different levels of code practice.

Martin, I agree about that Met Office paper being a bit, let's say, solipsistic. They also seem to put app developers in the same class as say oil rig control software engineers. The idea that only scientists need the exact version control in order to repeat tests is laughable.

Nov 18, 2015 at 8:00 AM | Registered CommenterMicky H Corbett

Mickey H Corbett - Yes.

Years back, I was on my way to Geneva sitting by the side of a colleague who was at that time in the process of creating an X.400 messaging system. As the Boeing 737 (I think it was a B-737) started its takeoff run, my colleague turned to me and said "Martin, you see that engine to your left? I wrote the software that is controlling it".

My immediate thought was "well, in that case, whatever might go wrong on this flight, it won't be the result of a software defect in the engines". Having sat in, for interest, in some of his team meetings as they reviewed the design of their X.400 system and explained the formal specification methods they used, I got a glimpse of what had been involved in their producing a sizeable system that was essentially error-free.

Later, I got involved in quality and productivity assessment in a business software development organisation with aspirations to raise their level in the quality maturity model to beyond level 2. Not exactly, as they say, 'rocket science', but with a level of discipline and organisation that makes the Met Office's climate model development sound quite ramshackle.

Somebody whose sole experience has been writing and "debugging" some bits of Fortran will just have no conception of what is involved in the production of high quality software. Hence my comment to EM about the difficulty of a discussion on the topic.

EM deserves praise for the wide range of things he takes an interest in. But I have pointed out before to EM that he will often imagine what is involved in things that are outside his knowledge or experience, and that will then become, for him, his reality. He will blithely post stuff here that he has imagined, as if it were reality.

At the lowest level, he has told me that I lack the knowledge to do specific heat calculations (O-level physics stuff) merely because he had never seen me mention it. EM always seems to be willing to explain how simple is the work of an electronics engineer (on the basis, it seems, that he once watched his brother try to make a transistor radio).

Nov 18, 2015 at 9:59 AM | Registered CommenterMartin A