I am in awe of what has been achieved with BOINC and with climateprediction, and I ran SETI from the early days. The sheer existence of creating anything that can be distributed successfully to 100,000s of users was an enormous achievement. Creating climateprediction was an enormous achievement. But (there\'s always a but) while those achievements were awesome technical feats the barrier to adoption isn\'t a technological one any more. Overcoming the real barrier to uptake today requires different skills and thinking.

I have talked with many who tried to run this and gave up, I notice that only 15% of those that ever signed up are still active. The ease of use hurdle for the computer user of today is just too high. All the information is there, all the technology is there but it\'s just too hard to get to it and use it. I know that you did it, I know that thousands have done it but I know that the main reason 10,000,000 people are NOT using it is simply the ease of use. There are a billion computers in the world - just 1% of those would be a computer 1,000 times more powerful than anything that exists today. We have everything we need apart from a simplification and streamlining of what already exists.

The big technology issues have been solved and now it\'s time to push it to a place where it\'s really really easy. Live Earth will provide massive publicity for doing something about climate change. I intend to help with this (I just launched a blog http://tenmilliondesktops.blogspot.com/ to get this started by finding people that can help raise the profile of climate prediction through Live Earth and other means).

If you look at BOINC and you look at www.GridRepublic.com you can see how something can be done to make it easy for a 55 year old Art teacher to use (rather than a 22 year old comp sci major). BOINC is a disaster (an enormous achievement, but a user disaster). GridRepublic is software as people want and like software. Simple. Three steps. Pretty pictures. Hard stuff tucked away. I have no link with gridrepublic, I\'m just a user who got very frustrated with BOINC over the years and think GR are a bolt of light from the blue. It\'s time to make the same transformation in all other parts of the user experience.

My wish is multiple:
1) New client that looks good - not like Win98 interface.
2) New client that hides all the \'messages\' stuff
3) Client that has stats that actually say what they are measuring (what IS the vertical? the chart doesn\'t say)
4) Three-click installation process
5) Screensaver that actually works. The version I have running on 4 computers tells me to CTRL-click various letters to see different views - all that does is stop the screensaver.
6) Much much shorter work units (why not upload at intervals and let multiple people create a single run?)
7) Website overhaul (yes, this website). Created in 2002 I expect nothing has been fundamentally altered - it\'s just grown like topsy and looks old. There is 10x too much information on it for the type of user that you should be attracting now. Also a lot of the information here, at BOINC, embedded in the model and the client is just plain wrong.
8) Client that tells you what you need to do to make things better and helps you do so \"Hey, thanks for all you\'re doing - right now we need a little more disc to run your model. Can you click this link and select a new allowance for us from the drop-down? We promise we won\'t take it all, and you can always change it back later. Thanks!\"
9) Client that pretty much assumes you\'re running just ONE project on just ONE computer (that\'s how most users think). I know you don\'t need a project manager to run one project - but you need a project manager to run one USER - to make it easy for them to understand and control what\'s going on.

I expect that every single one of these features will have removed 10,000 users from the experiment - that\'s 90,000, or about the number that are running today. With Live Earth coming up we should be challenging every band that\'s playing, every reporter, AL Gore, and every sponsor to be running this software. If I can raise the money I\'ll put an ad on TV asking people to log on and download and get started. We can\'t ask them to when the interface looks like it does now.

I know it\'s a major undertaking, and I know you don\'t have the resources, and I know half the issues are BOINC related - but I believe now is the time, and now is the time that those resources should become available - visibility is support and that translates into time, money, and equipment.

We can have 10x as many people running this. 100x times. 1,000x is not a silly idea. The big issue isn\'t whether it runs on Linux or whether it runs on quad-core with maximum efficiency. The issue is why it isn\'t running on my Dad\'s computer. The problem isn\'t the client capabilities or my Dad\'s computer - it\'s my Dad. And let me tell you, there are millions of people like him, and he ain\'t gonna change. So we/you must.

If you want to contact directly me with support or explanations of why this can\'t be done do so through :
http://tenmilliondesktops.blogspot.com/

I\'ve registered some other URLs that should be a bit easier as soon as I can get them all linked together right.

Meanwhile think about what I\'m saying here, and feel free to copy it around. If you get a chance find an *average* person (someone who thinks an iPod is a bit tricky to master) and ask them to run climateprediction. Give them just one URL to get them started and then sit and watch what they do, and listen to what they say and what they think. No helping. See if they succeed.

All of what has been done here is totally awesome - the most powerful computer on earth. It is just incredible. And it can be 100x more incredible just by tuning into the next phase - the people with the VCR (yes they still have a VCR) that flashes 12:00. If we want users we have to go to where they are - physically and mentally. The good news there are many many millions of them who would all be happy to help if we can help them.

Thanks for listening

Quentin

---

Have you tried Boinc manager version v5.8.x\'s \'simple view\' user interface?

It doesn\'t address all of your points, by any means, but it does address some.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

Have you tried Boinc manager version v5.8.x\'s \'simple view\' user interface?

It doesn\'t address all of your points, by any means, but it does address some.

I have. It does show an awareness of the issue, but it really doesn\'t go much further than that. It hides some stuff but hasn\'t made it go away for ever - if you need it (you shouldn\'t) you have to go to the other interface. What needs to be changed is the entire experience from someone sending you a URL you successfully finishing your first (short) climateprediction run. The whole thing needs to be simple, understanding what\'s going to happen, selecting the project, downloading, running, selecting disc space, having it not fail when you shut your laptop and run to class, watching a pretty screensaver that actually works instead of stopping when you follow the \'CTRL-T\' instructions that are right there on the screen (and I don\'t care if it takes 1% or even 10% of the processor to run graphics - without it the average user just won\'t bother or believe anything is appening), to getting meaningful statistics, getting online help when you need something, getting a fanfare when you complete, and getting a new project to run.

I know that a bunch of this is exists, but some of it doesn\'t and all of it is hard (for the average user) to both find and use. The average user owns a little under 1,000,000,000 computers now!

Quentin

---

One of the problems that you haven\'t addressed is storage space.
LOTS of it.
Nobody who wants to see millions of people running this project thinks of this.

At the moment the project is having serious problems because it can\'t get/afford the hardware to provide the multi terabyte storage needed. And this is with 50 thousand or so users.
You\'ll need to solve this problem first.

Having said that, enough money has been found for a new raid system, which should be in place next week.
But lots more storage will be needed before it\'s worth considering a big change such as you suggested.

PS
Have you heard about the BBC\'s Climate Change Experiment?
This did some of what you\'re suggesting, attracting half a million people in the UK at the start.

DLB,

Your goal is admirable. It\'s limitation is hardware and a pot of money to develop the software. The older the hardware, the more severe the limitation. The limited IT budget means that there is little left for feel-good software enhancements after science and fire-fighting considerations.

Further, one can simplify a super-computer Model and its interface only so much. One can compromise the integrity of the calculations only so much. Some of us recently finished Beta tests for an SSE2/SSE3 optimized Model and for 40-year Runs. Though they ran beautifully, not to mention faster, the tests failed, thanks to the sensitivity of the Model. Further still, the differences in floating-point operations among CPU types are not without consequences in a 160-year Run (200 years when we did the Spinups). Those consequences appear when one 40-year segment is processed on one CPU type, the next segment on another, the third on something else, ... Break it into yet smaller pieces and quality control suffers even more.

Yes, in past tests, comparisons were done for the same Model processed on different CPU types. (Differences weren\'t huge but there were differences.) If the long Runs are completed on one machine, the science team deals with an known entity. Pandora\'s Box, anyone?

It\'s all well and good to want simplicity and for the thing to run on grannie\'s abacus and grandpa\'s slide rule, and meet their skill set, but integrity of the result is paramount.

You weren\'t around through the growing pains in the \"Classic\" version and, then, conversion to boinc. Perhaps that\'s good because CPDN is easier to run now.

The BBC-sponsored part of the Project was simplified a bit though, unfortunately, not as much as advertised. Thousands of people, unsophisticated in the ways of computers and even less sophisticated in the ways of distributed processing nonetheless managed to complete thousands of Runs. Could it have been done better? Undoubtedly, given a sufficiently large and skilled IT staff and enough time for comprehensive testing. (Our Beta test was severely constrained by BBC\'s schedule.)

Given imperfections in the Model and in personal computers, a plug-n-play CPDN Model is but a chimera.

In a perfect world the Model really would be plug-n-play and there would be no crashes to deal with and these boards would be unnecessary. Last I checked, this is not a perfect world. Hence ...

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

\"Have you heard about the BBC\'s Climate Change Experiment?
This did some of what you\'re suggesting, attracting half a million people in the UK at the start.\"

Yes indeed, and I was very excited when it launched - I thought that it really would be a much easier process and interface - but it was a pretty incremental change as far as the end user experience goes. And the issue is exactly what you said - \"at the start\". Something like 10,000 people reached the end after millions had been invited through some very compelling TV programs and a national awareness campaign. I think the progress is amazing, but this really showed how the interface isn\'t up to the interest level of the average person.

I know more now (thanks to the following post) about the challenges, and I\'m not belittleing them at all - I\'m just pointing to where the challenges are (and perhaps we are all in agreement here) : user interface / ability to split the model across 40 year or shorter runs / storage.

Quentin

---

Your goal is admirable. It\'s limitation is hardware and a pot of money to develop the software. The older the hardware, the more severe the limitation. The limited IT budget means that there is little left for feel-good software enhancements after science and fire-fighting considerations.

Thank you! My point is that this is the exact opposite of \"feel good\" enhancements. Fire fighting has to happen, of course. My personal take is that if someone loses a few seconds of processing everytime a project switches (BOINC), if it can\'t use all the processors on a quad-core, or if it won\'t run properly on a Linux dual-boot (I am making these up) then frankly \'tough\'. Better spend time helping the would-be user sign up and actually run the software. Each user can DOUBLE their throughput in one easy step - sign up just one more person. If the average person can\'t use it then the number of users is capped. And are you referring to old hardware at your end? I am sure you\'ve have gone cap in hand to HP, Sun, IBM, etc but maybe it\'s time to go again, with the awareness of global warming ramping up enormously in the US - what a great opportunity for them to burnish their credibility as
1) green
2) suppliers of cutting-edge hardware
?

.....the tests failed, thanks to the sensitivity of the Model. Further still, the differences in floating-point operations among CPU types are not without consequences in a 160-year Run (200 years when we did the Spinups). Those consequences appear when one 40-year segment is processed on one CPU type, the next segment on another, the third on something else, ... Break it into yet smaller pieces and quality control suffers even more.

I am assuming that this is a sensitivity not exactly to initial conditions, but a similar chaotic sensitivity? I am amazed that this is a factor (and that shows how much I know!). Matching processor types for a second user to pick up a run will/would obviously be a tricky thing to do. Thanks for the info.

It\'s all well and good to want simplicity and for the thing to run on grannie\'s abacus and grandpa\'s slide rule, and meet their skill set, but integrity of the result is paramount.

Well, of course the results must be valid, and thanks for the background on why this is a challenge on smaller work units. And, just to be clear, my Dad has a perfectly respectable PC, and any of the 10s of millions of new PCs being shipped this month that don\'t have BOINC or climateprediction pre-installed as the screen saver are all perfectly capable of running this software. So it\'s not grandpa\'s slide rule that\'s the issue here, and not even so much the skill set that is the limiting factor. I majored in Physics at Bristol University and got a 2.1 - that doesn\'t make me any kind of genius but I expect I have the \'skill set\' for running climateprediciton on my computer. However, I don\'t have the time to learn where to find the messages in BOINC that tell me that I need 4.5 more Gb of storage, and how to find the link to the website from all the other links that offers me different configuration models (Home/Work...) and which one is applied to which computer I\'m running and whether this is going to take all of my HDD, and whether 10% is enough to get the space I need (i wrote it down here somewhere) and then check whether the change actually got back to my computer.....

You weren\'t around through the growing pains in the \"Classic\" version and, then, conversion to boinc. Perhaps that\'s good because CPDN is easier to run now.

Actually I was - the fact it doesn\'t appear in my profile is that across many PCs and many years I have repeated tried to get this to work and have repeatedly failed. I have been running this software for more than 7 years (when did it start?) and have never finished a climate model run.

The BBC-sponsored part of the Project was simplified a bit though, unfortunately, not as much as advertised. Thousands of people, unsophisticated in the ways of computers and even less sophisticated in the ways of distributed processing nonetheless managed to complete thousands of Runs. Could it have been done better? Undoubtedly, given a sufficiently large and skilled IT staff and enough time for comprehensive testing. (Our Beta test was severely constrained by BBC\'s schedule.)

And this *was* a huge achievement - and I am sure that the BBC didn\'t really get what you were doing or had to do, and you don\'t have the resources or time you need. All I\'m saying (I expect we\'re agreeing) is that this is the issue to be working on, and that\'s a very different issue from the ones in the past. The technological achievement is epic - it\'s a first-ever type of achievement. It\'s something of which people will be proud to say \"I was there at the beginning\". It seems that the challenge now is scaling it - hardware, support resources, interface redesign (BOINC and cpdn), complete website overhaul, sponsorship etc etc. This is a lot of work, and the computing power available out there will scale by orders of magnitude - 1 2 or even 3.

Given imperfections in the Model and in personal computers, a plug-n-play CPDN Model is but a chimera.

In a perfect world the Model really would be plug-n-play and there would be no crashes to deal with and these boards would be unnecessary. Last I checked, this is not a perfect world. Hence ...

Again, I don\'t know enough to say this with any certainty, but I have to believe that plug and play is *ultimately* possible. BOINC and cpdn must have looked impossible a while back. And, while that dream hangs out there are many other advances on the way that would make a huge difference to use acceptance.

I\'m just (re)starting the conversation and seeing where it can go.

Who knows anyone at HP / Dell / IBM (I know they have their own grid thing going) / CISCO / Sun / Teradata / NCR who might feel like some massive publicity around Live Earth in return for a bunch of hardware, end-user support, and programming assistance?

Sebb

---

One of the problems that you haven\'t addressed is storage space.
LOTS of it.
Nobody who wants to see millions of people running this project thinks of this.

At the moment the project is having serious problems because it can\'t get/afford the hardware to provide the multi terabyte storage needed. And this is with 50 thousand or so users.
You\'ll need to solve this problem first.

Having said that, enough money has been found for a new raid system, which should be in place next week.
But lots more storage will be needed before it\'s worth considering a big change such as you suggested.

Interesting. Is there any way to distribute storage requests too (like torrents) - or is that completely missing th point - you need real-time access to all the data at your end?

In the existing model how many tetrabytes are needed for 50,000 users, and how much does this cost for the type of storage you use?

Sebb

---

The sulphur version (Dec \'95/Jan\'96) of the climate model did exactly this - people were supposed to keep the results files on their hard disk so they could be retrieved if needed.

In practice people weren\'t prepared to do that. Hence in the next version of the climate model (the coupled model), the climate data was transferred back in the form of trickles.

Incidentally you can now browse the uploaded data via the new results server (http://results.cpdn.org/), although it looks like that server is down at the moment.

Milo mentioned current disk space was (I think) a couple of terabytes, and they\'ve got a new 3TB server. The cost of disk space isn\'t so much in the actual disk drives (SATA drives are dirt cheap now), but in the resiliency and backup systems and so forth.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

The sulphur version (Dec \'95/Jan\'96) of the climate model did exactly this - people were supposed to keep the results files on their hard disk so they could be retrieved if needed.

In practice people weren\'t prepared to do that. Hence in the next version of the climate model (the coupled model), the climate data was transferred back in the form of trickles.

Incidentally you can now browse the uploaded data via the new results server (http://results.cpdn.org/), although it looks like that server is down at the moment.

Milo mentioned current disk space was (I think) a couple of terabytes, and they\'ve got a new 3TB server. The cost of disk space isn\'t so much in the actual disk drives (SATA drives are dirt cheap now), but in the resiliency and backup systems and so forth.

I wonder if now that disc space is dirt cheap for end users too now whether they might be more willing to store data than before.
And, out of interest and in case I can raise a sponsor - how much is storage, resiliency, and backup for 50,000 users?

Quentin

---

Probably around £10,000 or so ($20,000), but that\'s a guess. A typical high reliability server would be 2/3 of the cost, and the rest being the other infrastructure.

However, thanks to an internal reorganisation last week (CPDN moved from one department to another) I don\'t think funding storage is a problem any more. PM Milo on the main forum if you need more specific information (http://www.climateprediction.net/board/index.php).

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

DLB/Sebb/Quentin, as you question the the implementation of CPDN, some comments and questions for you. Fair enough?

Actually I was - the fact it doesn\'t appear in my profile is that across many PCs and many years I have repeated tried to get this to work and have repeatedly failed. I have been running this software for more than 7 years (when did it start?) and have never finished a climate model run.

There is a reason it doesn\'t work on you PCs. Most don\'t meet the minimum spec. (And, yes, minimum specs are documented.) Since your June 2006 registration, your four machines have been issued 47 Models. No successes. Not a good record, eh? Whose problem? Ours? Or yours?

More than seven years experience with this software? A gross \'typo\', surely! How? I was part of the original Beta . . . in mid-2003. (That was for the \'Classic\' Model, a year before CPDN went to boinc.)


The sensitivity on which the latest Beta tests stumbled was the hard-won Sulphur parameter, lost to compiler optimizations. Sulphur is a small value but ...


As to your issues with not having time to read, etc. How many things in life do you undertake without knowing what you\'re getting into? Once committing to something challenging, how many things do you start and then walk away from, assuming all will go without a hitch? On your job, how many things are as reliable as you want this Cray supercomputer model to be on a vast array of PC types, running different OS types?

As I wrote in my first sentence to you, your goal is admirable (and we\'ve seen it stated more than a few times before). You must recognize, however, that it is utopian and we don\'t live in a utopia (at least, not those of us with ordinary means). You show the utopian horizon to us, we hold the mirror of reality to you.

And are you referring to old hardware at your end?

No, actually, I alluded to old hardware on your end, and mentioned above.

\"...your end?\" My hardware is in Washington state, USA. Les\' is in Australia. Mo\'s, Marj\'s, Mike\'s and others are scattered across the UK. Honza\'s is in Czech Republic. Geophi in central US. Others in France, Germany, New Zealand, ... (You get the picture; the clock is reasonably well covered much of the time.) None of our hardware is connected with Oxford except by the license under which we all run the Climate software, and by occasional Internet links.

By \"your\" do you mean the Oxford crew? That isn\'t us. The Mods here are volunteer participants. We receive no remuneration, except for an occasional word of thanks (and the too-frequent blasts of vituperation or spleen -- once is too often for that.) That\'s not to mention the joy of responding to repetitive posts by the many who, like you, can\'t be bothered to make use of the wealth of information already on these Boards, much of the pertinent stuff conveniently collected at the tops of the Threads. So, we\'re here because we believe in what we\'re doing, not to build pebble piles of pseudo-credits or to fatten our bank accounts. It certainly isn\'t a task to fatten one\'s ego! We try to help interested folks over some rough spots because the Project is, unfortunately, imperfect -- as is our hardware. There is, however, precious little we can do for those who are offended by the necessity of a bit of effort on their parts. Can/Will you offer some suggestions as to how we can deal with those who expect perfection, short of having the world\'s only perfect project?

By the way, in re. the current Server issue which has exercised many, as documented on the Boards, one reason for recent hardware shortages is a project underway to make ALL processed data available online for use by Climate research scientists, worldwide. That project is in incipient stages but will be enormous -- especially when we start on (in-development) hi-res Models. (We\'ve done one: Seasonal Attribution. Others will have different structure; all will require storage.) If you\'d been around and read the complaints about Model size and having results left on machines, I doubt it would be suggested again, as in your 18:02:59 post.

Issues are manifold; expecting a plug-n-play environment, especially on inadequate equipment, is truly beyond reason. Your own words say it better than I can:

Again, I don\'t know enough to say this with any certainty, but I have to believe that plug and play is *ultimately* possible.

Jim

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

DLB/Sebb/Quentin, as you question the the implementation of CPDN, some comments and questions for you. Fair enough?

Completely fair! BTW: DLB - name of the first computer I signed up this time round, Sebb - nickname I used online and used here by mistake, Quentin - my real name.

There is a reason it doesn\'t work on you PCs. Most don\'t meet the minimum spec. (And, yes, minimum specs are documented.) Since your June 2006 registration, your four machines have been issued 47 Models. No successes. Not a good record, eh? Whose problem? Ours? Or yours?

Terrible record! Then add in the other ID\'s I\'ve had over the years - none of them ever finished either. All I\'ve managed to do is drain resources from you, and give nothing in return. I don\'t think that fault is a relevant question here, though. I never saw anything about minimum spec, and assumed that if it ran then it ran. It did run, so I stopped worrying about specs. Didn\'t realize until now that I was under spec and that this would create crashes. My assumptions were probably bad. Some way of stopping people with my specs signing up would seem to be a good idea. BTW - here\'s a simple process that is used for automatically polling hardware for gaming sufficiency, I am sure they would add BOINC to the list with a polite request (I may have already suggested it to them):
http://www.systemrequirementslab.com/referrer/srl

More than seven years experience with this software? A gross \'typo\', surely! How? I was part of the original Beta . . . in mid-2003. (That was for the \'Classic\' Model, a year before CPDN went to boinc.)

Only that long? It seems like ages - I\'m pretty sure I was on before it went under BOINC, but as you say, less than 4 years in any case. I remember I was running it when Larsen B calved and that has to be 4 years ago (but my memory has been shown to be fallible)....

As to your issues with not having time to read, etc. How many things in life do you undertake without knowing what you\'re getting into? Once committing to something challenging, how many things do you start and then walk away from, assuming all will go without a hitch? On your job, how many things are as reliable as you want this Cray supercomputer model to be on a vast array of PC types, running different OS types?

I am not looking for something challenging. And the software that millions of potential users will compare it to is Google, GoToMeeting, and InstantMessenger. You know and I know that it is vastly more complex below the surface (though Google is no technological slouch), but the average end user doesn\'t and they hold 10, 100, 1,000 times the computer power you are currently being allowed onto. I know you couldn\'t support that many users at your end tomorrow, but this has to be the direction you\'re going? I am looking to help, like the 100,000s who have signed up and not made it to the end for whatever reason. It\'s not their fault, it\'s not yours. As you say this is an historic achievement - all those OSs, processors, all done distributed and over the net. Incredible. What I am saying (and I suspect we agree) is that the/a next major step in increasing model Runs is in ease of use - from the first look at the first website (technically easy) through to shorter model runs (technically difficult).

As I wrote in my first sentence to you, your goal is admirable (and we\'ve seen it stated more than a few times before). You must recognize, however, that it is utopian and we don\'t live in a utopia (at least, not those of us with ordinary means). You show the utopian horizon to us, we hold the mirror of reality to you.

I think we are agreed on the basic points.

By \"your\" do you mean the Oxford crew? That isn\'t us. The Mods here are volunteer participants. We receive no remuneration, except for an occasional word of thanks (and the too-frequent blasts of vituperation or spleen -- once is too often for that.) That\'s not to mention the joy of responding to repetitive posts by the many who, like you, can\'t be bothered to make use of the wealth of information already on these Boards, much of the pertinent stuff conveniently collected at the tops of the Threads. So, we\'re here because we believe in what we\'re doing, not to build pebble piles of pseudo-credits or to fatten our bank accounts. It certainly isn\'t a task to fatten one\'s ego! We try to help interested folks over some rough spots because the Project is, unfortunately, imperfect -- as is our hardware. There is, however, precious little we can do for those who are offended by the necessity of a bit of effort on their parts. Can/Will you offer some suggestions as to how we can deal with those who expect perfection, short of having the world\'s only perfect project?

Please register my deep awe. And many thanks. This is the cpdn website and there is a wish list discussion board - looked to me that I was talking with Oxford. I\'m also not interested in credits (I still don\'t know what they even represent, and as we\'ve seen whatever I\'ve done with my PCs has been worse than useless - taking resources and not completing a run). My suggestions on how you treat those that expect perfection? Distinguish them from those that don\'t in fact expect perfection, and then tell them to %$#$-off.

Issues are manifold; expecting a plug-n-play environment, especially on inadequate equipment, is truly beyond reason.

Absolutely - so one change would make it obvious to the more casual user that they are failing because their hardware isn\'t up to it.

Your own words say it better than I can: Again, I don\'t know enough to say this with any certainty, but I have to believe that plug and play is *ultimately* possible.

We ARE agreed on this. And while there is tension between \'ultimately\' (more you) and \'possible\' (more me) getting into the gap between these two words is where things can be made to happen. I don\'t know if you have what resources (and what resources Oxford have) and I didn\'t know until yesterday that floating point differences were sufficiently large to disrupt model runs. I am laying out what my naiive end-user view of the next step is, and the fact that naiive end-users have all the computing power.
Re-reading my original post I DO sound angry, and I apologize. I am not splenetic (word?) about this, and I do stand by my actual ideas, even if the delivery sounded \'aggrieved\'. I do get frustrated when I see colleagues push CTRL-T to see the pretty temperatures on their screensaver just like the interface tells them to and then the model stops running. Making changes like updating the instructions, and much of the other \'plug and play\' progress is actually very simple - it\'s just time consuming and a set of design skills rather than computing skills and working with newbies and seeing what they understand / don\'t understand when presented with the website. I don\'t have an interest in anything that comes out of this (bank account - don\'t know how that would even happen, or in credits) - just in your results and support. If there are others with similar perspective/ideas or access to resources/sponsorship then maybe we can be a little more focused on helping.

Quentin

---

Probably around £10,000 or so ($20,000), but that\'s a guess. A typical high reliability server would be 2/3 of the cost, and the rest being the other infrastructure.

However, thanks to an internal reorganisation last week (CPDN moved from one department to another) I don\'t think funding storage is a problem any more. PM Milo on the main forum if you need more specific information (http://www.climateprediction.net/board/index.php).

Milo = mdzbor?

Many thanks
Quentin

---

Mdzbor is Martin Dzbor. Milo Thurston joined cpdn as an additional programmer last year and is working in Oxford together with Tolu Aina.

---

Cpdn news

P.S. I did see the specs page when I first signed up - years ago. And I\'m pretty sure that the spec then was lower than it is now (1.6GHz, 1Gb seems high for a minimum spec in 2003). I never saw a reason to check back and see if the specs had been increased. I\'m now in the process of stopping climateprediction on my four computers. Getting a new computer in a few weeks which will be well up to the job.

---

Yes, the spec has been changed. My old computer happily ran a Classic model on 256Mb RAM. The new models really need 512Mb. And because the Classic models were relatively short, a slow CPU didn\'t matter much. With the new models, which have more timesteps per model day and many many more model years, the 1.6GHz CPU minimum spec at least gives members a fair chance of completing their model.

BTW, we have been aware for some time that the cpdn website requires an update. The mods have no access to it and Sylvia, who used to look after it, is no longer in Oxford. The Oxford programmers are up against strict deadlines for the completion of many jobs and are always under pressure.

This is why the mods have set up the cpdn project READMEs (thanks to MikeMars) and News threads on all three forums. We know this doesn\'t put the website right, but it does give all members who visit any of the forums immediate access to up-to-date information.

---

Cpdn news

BTW, we have been aware for some time that the cpdn website requires an update. The mods have no access to it and Sylvia, who used to look after it, is no longer in Oxford. The Oxford programmers are up against strict deadlines for the completion of many jobs and are always under pressure.

This is why the mods have set up the cpdn project READMEs (thanks to MikeMars) and News threads on all three forums. We know this doesn\'t put the website right, but it does give all members who visit any of the forums immediate access to up-to-date information.

I know....it\'s the thing about draining the swamp while up to your arse in alligators.....

I don\'t know much about websites and all, but do recognize simple greatness when I see it : www.gridrepublic.org www.carbonite.com www.gotomeeting.com

So I don\'t know if it would be feasible to get some outside assistance (without starting a massive undertaking with huge expense) - assuming some minimal funding was available?

In any case perhaps it would make sense to collect from these boards, and maybe even solicit input / suggestions for what a revised site would look like if and when it happened? At least record in one place the best ideas and priority of issues to be fixed (a sub-wish list)?

Thanks again for listening.

Quentin

---