Having read lots of threads recently and over the last year or so I am struck by the total lack of any form of design review for the classic replacement project. Everybody on BOINC criticises classic users for being dinosaurs and against progress, etc, etc, presenting dubious and totally erroneous statistics to support their view.

It is quite clear that there are significant aspects of the BOINC structure that are excellent on paper, but which are not working well in practice. Along the way the people who supported this premature(?) drive to a new unproven system fail to understand what all the critics are saying. And because they have committed themselves to the new system refuse to countenance anything other than blind praise for it.

The facts speak for themselves: Network and server loads were way underestimated. The practicality of centralising client settings was not thought through. The simplest way of resolving validation and cheating (perceived or otherwise?) was dismissed. The functionality of the new GUI manager is orders of magnitude behind the standard established by classic command line/setistash/setispy/etc.

Distributed computing as in BOINC appears to be a form of commercial enterprise supported by the unpaid efforts of volunteers worldwide. Yet the structure and management is not of a commercial standard. In any such environment after a pilot study then a serious deisgn review would normally be undertaken especially when shortcomings become apparent. At this time a decision would normally be taken to instigate design changes to overcome the shortcomings or failures, or even, depending on cost or how little the proponents want to save face, to scrapping it and thinking again!

I don't suppose for a moment any of the classic users are against technology or new development. However, they have a working application, known and understood, providing them with satisfaction in terms of kudos and consistent statistics and are only against being forced against their wills to a new system that has a lot of problems. Give us a working BOINC application and management system that is at least as good as we have with classic command line, provides us with continuing and therefore comparable statistics and most arguments will dry up overnight.

---

Nick

SETI Classic has been way behind on validateing like BOINC is now for over 5 years according to Matt. But when you are given credit ahead of time in classic you did not see it.

This is a volunteer project, running on short resources at Berkley, and all the staff have other things that they have to get done at Berkley SSL also. If the main goal was to assign you credits you would have a complaint, but it is for the science behind it first.

From our standpoint we do not see the whole process, just what Matt says and what is in the server status page. I am sure that as time permitts at Berkley SSL things will improve. Yes we do see a lot of other projects without seeing the problems like at SETI, but how many problems do they have that you do not see? It is just that SETI has many more people working on crunching the numbers, and letting more info out to the users that you see the delays that you do. I am sure that the other projects would have more problems trying to keep this many volunteers going.

Ray

---

Pizza@Home Rays Place Rays place Forums

Hope you have your flame suit on. This post is civil but I'm sure there will be others :)

The facts speak for themselves: Network and server loads were way underestimated.

According to Matt and/or Rom all the servers are well within their load capacity. The database server is 80% idle. Right now the only problem is number of files in one directory structure. This problem was brought on by some bugs early on that let the files live longer than they should have and now they are orphaned. I think most of these bugs have been fixed but the files they dropped along the way are still there. The network load is only a problem after outages and when classic is turned off, that will no longer be a problem either.

The functionality of the new GUI manager is orders of magnitude behind the standard established by classic command line/setistash/setispy/etc.

I'm still confused when people say this. There are a lot more features in the new GUI. For example "allow no new work". In classic you had to create a text file in a directory - how obscure is THAT? Built-in queue management. Statistics tracking. Log messages. Where were all these in the classic setup? And you are comparing add-ons to the plain-vanilla client. This isn't fair. Setistash and setispy weren't around for the first year of classic either. And there ARE comparable add-ons for BOINC appearing now. BoincLogX, BoincView, etc.

Give us a working BOINC application and management system that is at least as good as we have with classic command line, provides us with continuing and therefore comparable statistics and most arguments will dry up overnight.

In my view we alredy have this. I enjoy BOINCing and have become more involved than I ever did with classic.

---

A member of The Knights Who Say NI!
For rankings, history graphs and more, check out:
My BOINC stats site

... Yet the structure and management is not of a commercial standard. In any such environment after a pilot study then a serious deisgn review would normally be undertaken...

You seem to have quite rightly noticed that we are indeed The Pilot Study.

This is all a learning exercise for all concerned. In parallel, useful data analysis continues uninterrupted.

And I guess that your credits might be a little slow to catch up...

Regards,
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

pundit

DEFINITION: (noun) someone who offers opinions in an authoritative style.
EXAMPLE: The Seti forums are filled with pundits, each with his or her own theory about how and why Boinc Seti runs or does not run.......
SYNONYMS: authority, expert


Or maybe this is more appropriate

This 'SPACE' Rented.

The anonymity of the Internet, brings
forth, yet another EXPERT.

---

Semper Eadem
So long Paul, it has been a hell of a ride.

Park your ego's, fire up the computers, Science YES, Credits No.

Having read lots of threads recently and over the last year or so I am struck by the total lack of any form of design review for the classic replacement project.

Nick is assuming that, since such review meetings aren't posted on the site that they must not happen.

---

edit up front: I must say, I didn't rate the original post and won't rate it either. Just seeing all the negative ratings going on there isn't really necessary, in my opinion. If you don't follow his line of thought, just answer in here why not. Rating (negatively) anonymously is going a bit too far.

Having read lots of threads recently and over the last year or so I am struck by the total lack of any form of design review for the classic replacement project. Everybody on BOINC criticises classic users for being dinosaurs and against progress, etc, etc, presenting dubious and totally erroneous statistics to support their view.

Then read again over the last month or so. Not everyone criticises the Classic users for using Classic. All that may be criticised is the use of the word BOINC, when they mean Seti. Same goes for you. BOINC has its own server and news. If you want to learn more, write in on the developer's email list.

It is quite clear that there are significant aspects of the BOINC structure that are excellent on paper, but which are not working well in practice.

Windows Millennium, Windows XP pre SP2, playing 3D games on a bug standard Windows 2000 Pro version, or maybe up to SP2: impossible. You'd need SP3 at minimum, SP4 for later games.

Along the way the people who supported this premature(?) drive to a new unproven system fail to understand what all the critics are saying. And because they have committed themselves to the new system refuse to countenance anything other than blind praise for it.

Whereas otherwise in your view, we should all not go for BOINC since it has or lacks what, exactly?
My Boinc, an alpha version called CC4.72, has been running steady since it was released. I learned not to change anything, just let it do its business of figuring out which projects were attached, what they wanted and it's still running smoothly.

That Seti at this moment has problems with validating and LHC may be out of work, doesn't matter to my computer running Boinc itself. It can always switch to a new Seti unit, to Einstein@Home, Seti Beta and SZTAKI DG.

The facts speak for themselves: Network and server loads were way underestimated. The practicality of centralising client settings was not thought through.

As told by others here, there's not a problem with server loads. Maybe that the 100Mbit ethernet connection is a bit overloaded after an outage, but with a new switch that should be solved.

The simplest way of resolving validation and cheating (perceived or otherwise?) was dismissed.

Which was what exactly in your opinion? Closing Seti/Boinc? ;)
Or just give credits as in Seti Classic and validate afterwards?

Better yet, why no option for no credits at all on both the versions? No 1 WU= 1 credit for Classic and no 3 WUs make a quorum, middle WU's points are given to all in the 3 and there after. That would be a fine solution to me. No credits. Not on Classic and not on Seti/Boinc.

It would take a lot of firing material away from you.

The functionality of the new GUI manager is orders of magnitude behind the standard established by classic command line/setistash/setispy/etc.

Seti Classic came as a CLI without screensaver and a GUI with screensaver.
All other programs were optional and 3rd party based.

Boinc comes with a lot of the 3d party program options included.
It still has a great 3rd party based program variety.

For someone who has read here for over a year you're badly informed.

Distributed computing as in BOINC appears to be a form of {snip} to scrapping it and thinking again!

It's Open source. If you think you can do better, why don't you show it to the world?
It's way easier to break a program down by sitting on the sideline, having only tried it for a minute, than it is to go for it full fling, or heck, even make your own version. Boinc gives you the option to make your own option, so why the comment about feeling it's a commercial product, when you haven't paid top dollar for it? Can you run it? Did you have to pay for it? Did it stop working after an hour telling you you had to pay $300.- for it to see all of what it could do?

It didn't, did it?

So why the snarl then?

they have a working application, known and understood, providing them with satisfaction in terms of kudos and consistent statistics and are only against being forced against their wills to a new system that has a lot of problems.

Please show me pictures and video of persons in black UCB gear pointing guns at people's heads forcing them to run Boinc software. I'm sure the news media would like to know of this attrocity as well.

If you cannot produce the imagery, then it might be because you're volunteering your computers, your electricity, your multi CPU systems, your game time, your grey hairs and the well being of your whole family to crunching Seti through Boinc.

Give us a working BOINC application and management system that is at least as good as we have with classic command line, provides us with continuing and therefore comparable statistics and most arguments will dry up overnight.

Read on for another year and a bit. Was Rome built in a day? Will you allow us who have volunteered our computing time to Seti/Boinc (and a load of other projects) to make it better? Or do we have to throw in the towel now and just start over?

---

In advance I will warn tender readers that I do not think that I managed to keep my customary aplomb and good humor ... apologies in advance ... but I am tired of the complaints that are nothing more than unsupported opinion presented as holy writ delivered to skeptics who cling to an unfounded belief that BOINC actually performs any task which might allow productive work and that any criticism of the criticism is unwarranted, unfounded, hostile, erroneous, and unsanitary.

Having read lots of threads recently and over the last year or so I am struck by the total lack of any form of design review for the classic replacement project. Everybody on BOINC criticises classic users for being dinosaurs and against progress, etc, etc, presenting dubious and totally erroneous statistics to support their view.

Actually, there are some design discussions and reviews, although they do not always take place, and when they do, they usually take place in the mailing lists.

Since BOINC is intended to be used, for the most part, in the support of academic research projects, the main support and design drivers will be other project personnel. Not the participants. We do have a voice. But we do not have a veto, other than to vote with our feet. But, suggested changes are subject to approval by the guys that own the baseline. If you don't like the direction, as said elsewhere, get the source and roll your own. You feel that you have a better handle on how to make BOINC slick, which is cool, either write up you detailed design proposal, or write up the code. Either way, it is up to you ... oh, and, sorry, I don't buy the excuse that "I don't know how to ..." Learn.

None of us criticizes the Classic users. IF that floats your boat, why are you here? Why not wait for the last gasp? Time and again we say that BOINC is work in progress. For all that, and for all the problems that the projects have, and I have memories of when all but one project was dead ..., it works pretty well. If you don't like the way it works, well, see the above.

It is easy to say, and dismiss, any other statement with a glittering generality. What statistics? Who said it? What were they answering? Making a general, blanket, and unsupported statement like that just about kills any chance you have of making a point.

And, yes, if you wish, you can now pout because you are being "put upon" for your opinion... Actually, not hardly. Nary a curse, or a disparagement of you or the ideas you have not presented. I am just asking, "Where is the beef?"

It is quite clear that there are significant aspects of the BOINC structure that are excellent on paper, but which are not working well in practice. Along the way the people who supported this premature(?) drive to a new unproven system fail to understand what all the critics are saying. And because they have committed themselves to the new system refuse to countenance anything other than blind praise for it.

Actually, the funny part is that we understand you, and your complaint very well indeed. The real question is why you cannot see the converse? Yes, there are aspects of BOINC that do not work well. Many places need improvement. And I defy you to search through my entire set of posts to find one where I said BOINC is perfect, does not need improvement, or does not have problems.

If you read my posts carefully, and look through the history, I, and many others have been scathing in our criticism in the sense that we do not hold back our opinions. However, in contrast to the earlier parts of your post here, when *I* complain, I start off with research. I look in the code, I test the system, I document the results and then report them. But, I neither condemn the developers nor waste my time with recriminations. If there is a specific thing wrong, so state. Saying "its broke" is neither useful or meaningful.

And for one of the "papers" we, the "we" being the community did you can look at Benchmarks which we did back during BETA Test. Another study that got interrupted because we really did not get good data capture was this study of issue rate. We have done others. JM VII published his design document for example, we made comments, and, several versions later, well, *I* think it does what he intended it to do ... but, you would probably say that I am just blind to the fact that it does not work ... yet, BOINC cooks away on my systems ....

The facts speak for themselves: Network and server loads were way underestimated. The practicality of centralizing client settings was not thought through. The simplest way of resolving validation and cheating (perceived or otherwise?) was dismissed. The functionality of the new GUI manager is orders of magnitude behind the standard established by classic command line/setistash/setispy/etc.

Perhaps the loads were underestimated. What would you have done differently? It is fine to say they made a mistake. Perhaps they did. As a matter of fact, I did another less public study of the database design which was, is, and always will be very critical of the way that the BOINC database was implemented. Yet, if the physical DBA says that the DB is not heavily loaded, what difference does it make? Oh, and in their defense, some of the capabilities that would have enhanced the design do not yet exist within MySQL (Triggers and Stored procedures).

WHAT simple way. If you are going to make an unsupported statement like that, how can you expect to be taken seriously?

Lets see what this defective BOINC can do that "classic command line/setistash/setispy/etc." does not (and note, I was devoted to SETI Spy in my classic days - I still have logs if you want to see them as proof - I don't throw much away):

* Connect to multiple computers
* Allow other GUI programs to substitute for BOINC Manager (like BOINC View)
* Prevent external access without a password
* Prevent external access from non-approved IP Addresses or Machine names
* It can allow simplified installation as service, single, or multiple user
* Allow other third party tools like BOINC Spy, BOINC Log X, etc. to be used just as in those golden days of classic ... oops, forgot, they are still here ...
* Run projects other than SETI@Home at the same time with no manual intervention
* Allow you to run the command line version if that floats your boat.
* BOINC Manager also can connect to other computers as above.

There are a couple things that BOINC cannot do:

* Have a single storage queue that services multiple computers which may or may not be connected to the Internet.
* Satisfy people that can find nothing good to say about it.

You know it is funny, when you make your post, you belittle those that seem to have good things to say about BOINC, but hold yourself up, well, I shan't go there... but think about it ... what might that say about you?

Distributed computing as in BOINC appears to be a form of commercial enterprise supported by the unpaid efforts of volunteers worldwide. Yet the structure and management is not of a commercial standard. In any such environment after a pilot study then a serious design review would normally be undertaken especially when shortcomings become apparent. At this time a decision would normally be taken to instigate design changes to overcome the shortcomings or failures, or even, depending on cost or how little the proponents want to save face, to scrapping it and thinking again!

Actually, DC using BOINC is almost all academic. The commercial DC projects are not using BOINC yet. But, it would not surprise me if that did not happen in the near future. We already see something along those lines with BURP. And, don't get us started with commercial is always great ... I will simply say Microsoft Windows (Disclaimer: I have been a bitter critic of Microsoft since the late 1970's time frame when I was working with their compiler/OS and its improper handling of the exception stack of the 80287 co-processor - which had only 4 slots and would crash the system because they never cleared them after "handling" them ... MS "Help" desk told me that it only occurred with IBM Dos ... which was interesting as I had traced the program execution with a debugger... but I digress).

So, we are an open source, volunteer, primarily academic (read as NO MONEY. Yet, in spite of that, and quite a few posts, including a very well done note in the Technical News page telling us what is know, what they HAVE tried, what they PLAN to try ... and some personal notes from Matt in other threads. The one that frustrates me is that there is this assumption with no foundation other than personal belief that "Classic was/is/always will be better than BOINC". You offer no proof of your thesis, other than "I don't like BOINC and it is broke" with no specifics.

Especially annoying is that Matt has stated in other threads, and I also (from my own observations of SETI Spy + SETI Driver and how they showed many connection problems), the old technical news lists, etc. and it still makes no impression. Just because you do not see them, does not prove that there are not problems. Your thesis is that Classic is superior to BOINC...

Prove it!

No data, no facts, no nothing is what you have presented to this point. You want your opinion on the problems with BOINC to hold water and make a difference. Well, make them other than just opinion. THEN and ONLY THEN will you be taken seriously as a contributer to the design of BOINC.

I don't suppose for a moment any of the classic users are against technology or new development. However, they have a working application, known and understood, providing them with satisfaction in terms of kudos and consistent statistics and are only against being forced against their wills to a new system that has a lot of problems. Give us a working BOINC application and management system that is at least as good as we have with classic command line, provides us with continuing and therefore comparable statistics and most arguments will dry up overnight.

At the moment, no one is being forced anywhere.

And by the way, Willy says that you are wrong about BOINC working to the tune of about 3,579,150,592 Cobblestones ... to date ...

And if you gave BOINC half a chance you would find that it is better than the Classic command line... for one thing, I can have the GUI when I want it, I can have the graphics when I want it, or, neither. In Classic, if I want the screen saver/graphics I have to pay for it all the time, or not have them at all... and if you truly love that command line, install BOINC as a service (daemon for the OS-X/Linux people).

Here is the bottom line. You can be part of the solution, or something else ...

In Paul's opinion, well, I don't see you as part of the solution. And if you see this (and I admit I am a LOT more negative at this time) as being uncritical "blind praise for it" and that I am "criticizes classic users for being dinosaurs and against progress, etc, etc, presenting dubious and totally erroneous statistics to support their view", well, so be it ... for the most part I have asked for you to be specific.

Last week I had a conversation with a participant that DID have a legitimate complaint ... it took me awhile to understand the point ... but ... this is/was a design choice made for the moment ... and the point was correct ... BOINC does not do that right now ... there, see, BOINC does not work at all because ... well, no, actually ... it works except for that one place ... and I mentioned it above too ...

---

Just to add my point to this whole subject. When I first downloaded BOINC (v4.09), there were a lot of things that weren't quite right with it. In fact nearly half of the features implemented now didn't exist back then. Then I lived without them, but now I just couldn't live without the Update All button. I couldn't write some of my software until there were GUI RPCs, they didn't fully exist until v4.19.

So, just wait and all the features you need may find their way into the software. The facts that Nick Cole has stated, are somewhat half true. The fact that the server load has got high is a fact, but the fact that it wasn't planned for is not really a fact, it happens to be the way things worked out.

If it weren't for BOINC, I don't think you easily run 6 or seven different projects on each PC. Each of the projects would EAT the CPU cycles of the others! BOINC was a big leap and it is still growing. It is much more stable than in about January this year and so therefore I think the chaps at Berkeley need a pat on the back for managing to design it to this standard already and to keep working on it to bring it even further forward to have the way forward for DC in the future.

Please note: The views expressed above are MY views ONLY and feel free to criticize them.

---

At last some more informed debate. Firstly though we need to re-iterate that people are not against BOINC itself. What also needs to be remembered is that yes this is posted on the SETI BB so actually while we DO all realise that BOINC can do more than one project, the fact is that it is SETI that is probably of most interest. So any critical comments relating to BOINC are based on our experience of it running SETI.

Not everything that the project team need to do has to be publicised, but it is VERY good practice to keep all the stakeholders informed. When there is no visual evidence through information that the project team even read the BBs or take any notice of the critics who are motivated by a desire to improve things after all, then it generates concern. All the people who provide their time and machinery for nothing are stakeholders and without them the concept would not work. I am all the other committed classic operators are NOT against a new system, we just want one that is better than we have at present and when anybody actually looks unemotively at the evidence it is far from better. If of course somebody wants to run several different projects then BOINC is the only practical solution. But that is probably not what a lot of the current classic users are interested in, as evidenced by the sheer weight of numbers. So whatever processing application we use has to be better or cause less problems than what we presently use.

Comparisons with valid work units, repetition and so on do not stand up to independent scrutiny.

pundit

DEFINITION: (noun) someone who offers opinions in an authoritative style.
EXAMPLE: The Seti forums are filled with pundits, each with his or her own theory about how and why Boinc Seti runs or does not run.......
SYNONYMS: authority, expert

Out of the vast hundreds of thousands of contributors there is a very strong likelihood that the collective experience and knowledge of operating, designing and installing large computer systems, major projects and so on which massively exceeds the experience of the small (and not to denigrate them at all) project team. They are doing a job and the only ones who are in a position to do it, but do not seem to tap into the pool of knowledge that is there trying to help them. That also is at the heart of all the comments.

Comments highlighting flaws and problems are not negative. Distributed computing for the greater good is not negative, but it has to work properly. That is all we are asking.

Address the problems don't ignore them. And the problems are NOT caused by those people brave enough to highlight them.

Windows Millennium, Windows XP pre SP2, playing 3D games on a bug standard Windows 2000 Pro version, or maybe up to SP2: impossible. You'd need SP3 at minimum, SP4 for later games.

???

The point is that BOINC (as is classic) is a complex system. It has clients, servers and communications. While there were functional issues with classic command line client, it worked extremely well, and with the gui add-ons superlatively. Fully controllable, proper queuing, statistic generating, etc. the communications were not a major problem generally other than we have all experienced over the last year or so and affecting BOINC as well (anecdotally far more so in fact). The real problems lie in the server infrastructure. While there were issues with validation, this has absolutely nothing to do with the client software that we run. And a solution to this problem was systemically extremely easy to resolve, even without knowledge of how they have structured it. However instead of concentrating on fixing this, a new client and multi-project capability was built. While some of the server side issues were resolved others have been introduced which appear to be significant and in many respects a direct consequence of the system design decisions taken some years back. So rather than bite problems off in manageable chunks in reality mor ehave been added as the servers still have problems and now there are a massive volume of complaints and issues around the new client as well!

This is where the design review needs to have been kicked off, and it should have happened sooner rather than later. None of this is against distributed computing or BOINC per se, but given the range of issues it is still very much at beta stage.

Unfortunately the user population is being taken for granted, and while there may well be useful add-ons, documentation (which I have read and leaves more questions than are answered), volunteer support, and so on there is NOTHING on this web site that lists or links to any of it. What I have seen, and NOT being a geek who spends all day doing nothing but scanning the net, is limited. I spend some time when I feel motivated enough answering these points in the vain attempt that somebody at project level will do something, other than rely on volunteer responses which do nothing to address the problems as the volunteers are also not in a position to do anything. In fact ignoring us volunteers has been a fairly consistent theme. Back in 1999 I was in a position to supply 500 (then good) PCs available 24hr per day, unused for 16hrs to participate in the project. However I needed to discuss security and a number of other matters, did anybody reply? No.

It's Open source. If you think you can do better, why don't you show it to the world?
It's way easier to break a program down by sitting on the sideline, having only tried it for a minute, than it is to go for it full fling, or heck ....

I am not a developer, I am a systems engineer, I am an experienced project manager. I do not criticise the code, but the structure and how it is put together and crucially how it all works together. And it is the system as a whole that needs to be reviewed, not necessarily small aspects of it. Loading, network performance, UPS (even), database structures are all part of the system and it is visibly obvious that these are the areas where there are problems, and to repeat yes, it is SETI, not BOINC.

I'm still confused when people say this. There are a lot more features in the new GUI. For example "allow no new work". In classic you had to create a text file in a directory - how obscure is THAT? Built-in queue management. Statistics tracking. Log messages. Where were all these in the classic setup? And you are comparing add-ons to the plain-vanilla client. This isn't fair. Setistash and setispy weren't around for the first year of classic either. And there ARE comparable add-ons for BOINC appearing now. BoincLogX, BoincView, etc.

Yes there are more features in the new gui than the old vanilla client. But, less features and manageability than with the add-ons. The add-ons have been around for several years predating the BOINC gui which should by definition be better all round, which it isn't. It is extremely easy to use the Setistash and others) gui to stop processing or transferring wus. GUIs can also generate text files. I haven't written any text files to run 4 instances of classic. Since we are talking about a recent system then it is entirely fair to compare the new with what it is intended to replace.

That is what evolution is about. (Or are the pro-boincers closet creationists?)

SETI Classic has been way behind on validateing like BOINC is now for over 5 years according to Matt. But when you are given credit ahead of time in classic you did not see it.

So what has validation got to do with it? Fix that, don't introduce further validation backlogs and other problems. Credits are a multi-headed issue. The BOINC method is confusing, but is a counting mechanism. Classic is clear and immediately gives credit for the efforts put in. BOINC only gives credit if what you have been given to process does something valid or useful. This will inevitably underscore some people's efforts and boost others, but in a way that cannot be compared or established. If we spend 4 hrs processing a wu then we have done that much work. If it wasn't worth it or the process aborted or gave an invalid result then how are we to control it? We can only proces what we are given. We have no control over this aspect, and what about all that wasted time while the validation/credit system stabilises with each new client, why does it process things past the deadline, etc etc?

This is a volunteer project, running on short resources at Berkley, and all the staff have other things that they have to get done at Berkley SSL also. If the main goal was to assign you credits you would have a complaint, but it is for the science behind it first.

Only right up to a point. But the only thing we get out of it is the recognition of our credits. However, it is a university research project which presumably is going to advantage the faculty, professors, students, (pre and post grad). So they should give us something for helping them.

[quite]
From our standpoint we do not see the whole process, just what Matt says and what is in the server status page. I am sure that as time permitts at Berkley SSL things will improve. Yes we do see a lot of other projects without seeing the problems like at SETI, but how many problems do they have that you do not see? It is just that SETI has many more people working on crunching the numbers, and letting more info out to the users that you see the delays that you do. I am sure that the other projects would have more problems trying to keep this many volunteers going.
[/quote]
Yes again. But the answer is that to keep ALL the stakeholders happy we need to be informed as well. We are not all sheep here merely for the benefit of the students and professors at Berkeley. And again as mentioned above there is a vast untapped pool of experience in all of us.

And I am doing both classic and boinc. I am running blinc on a pilot basis, but having been with classic since the beginning want to see what I end up with at the end, so am not jeopardising that. But when (if) I move fully to BOINC I want it to work at least as well as classic, which to date it doesn't seem to do. Since the credit system of boinc is so obscure then that is another reason for staying.

---

Just to add my point to this whole subject. When I first downloaded BOINC (v4.09), there were a lot of things that weren't quite right with it. In fact nearly half of the features implemented now didn't exist back then. Then I lived without them, but now I just couldn't live without the Update All button. I couldn't write some of my software until there were GUI RPCs, they didn't fully exist until v4.19.

Why has it taken so long?

But classic updates itself. People want to be able to get something for the efforts they put in not merely a delayed reward if it happens to be validated at a future time, and then with strange fractions of credits as well!

So, just wait and all the features you need may find their way into the software. The facts that Nick Cole has stated, are somewhat half true. The fact that the server load has got high is a fact, but the fact that it wasn't planned for is not really a fact, it happens to be the way things worked out.

I hope not. That is not the way to design a proper international sysytem catering for up to 5 million users. Dare I say QED?

If it weren't for BOINC, I don't think you easily run 6 or seven different projects on each PC. Each of the projects would EAT the CPU cycles of the others! BOINC was a big leap and it is still growing. It is much more stable than in about January this year and so therefore I think the chaps at Berkeley need a pat on the back for managing to design it to this standard already and to keep working on it to bring it even further forward to have the way forward for DC in the future.

Not a point to argue about, but not everybody wants to run several projects. So why install software to do something unnecessary? However, if only one project is run then overall is is not an issue but then comparisons with features and manageability become the overriding issue rather than whether or not as a user I can run additional projects. (All of which will share available cycles anyway).

I am not criticising just trying to play devil's advocate and debate the pros and cons.

---

SETI Classic has been way behind on validateing like BOINC is now for over 5 years according to Matt. But when you are given credit ahead of time in classic you did not see it.

So what has validation got to do with it? Fix that, don't introduce further validation backlogs and other problems. Credits are a multi-headed issue. The BOINC method is confusing, but is a counting mechanism. Classic is clear and immediately gives credit for the efforts put in. BOINC only gives credit if what you have been given to process does something valid or useful. This will inevitably underscore some people's efforts and boost others, but in a way that cannot be compared or established. If we spend 4 hrs processing a wu then we have done that much work. If it wasn't worth it or the process aborted or gave an invalid result then how are we to control it? We can only proces what we are given. We have no control over this aspect, and what about all that wasted time while the validation/credit system stabilises with each new client, why does it process things past the deadline, etc etc?

There've been several lenghty discussions about that. Everything then boils down to two things:

• it's only fair to get credit (what ever it is) for work which is good for the project. Eg. if the task is: calculate 2+2 and your answer is 4.5, would you expect a reward? Would you get anything but negative in primary school for that?
  
• it's quite fair to get the credit proportional to the amount of work you do. Not all WUs are the same and it's fair to give more credit for harder WUs. You do expect to pay more for a large 5-bedroom house than for a studio in a block of flats, don't you?
  

Both aspects were neglected in SETI classic.

IMHO, for a user that does his/her best and no cheating, the crediting in BOINC and classic will in average be just the same. If an average WU is worth for example 30 credits in BOINC world, an average user will get say 5 credits/day in classic or 150 cobble stones in BOINC ... in average.

---

Metod ...

Health-warning: Naked opinion about to be expressed.

To Nick Cole,
(Manager (of some sort))

At last some more informed debate. Firstly though we need to re-iterate ... So any critical comments relating to BOINC are based on our experience of it running SETI.

Fine, good management-speak disclaimer.

I'll include the direct non-abstract disclaimer that the following is my personal opinion and that I speak for noone else.

NOTE THAT YOU DO NOT SPEAK FOR ME AND YOU DO NOT MANAGE THIS PROJECT nor any other part of Boinc.

Adjust your attitude and approach accordingly please.

... but do not seem to tap into the pool of knowledge that is there trying to help them. That also is at the heart of all the comments.

You are VERY WRONG. But then I guess you have not looked.

As you haven't noticed, Boinc is a FOSS project (look it up!) and there are a number of Volunteer Developers. There's also direct communication with the project developers if you have something meaningful to contribute.

(LOTS of abstract higher-level general big project management abstractions and inapplicable waffle snipped. Or in brief: Obfuscation HOT AIR.)

I am not a developer, I am a systems engineer, I am an experienced project manager. ...

Yes, it is blatently obvious that you are not a developer!

As for your high level big project management techniques, there is one thing you've directly overlooked:

There are perhaps only two paid staff for this project.

Adding YOU into the mix would add a full 33% of distraction plus additional waste in keeping you humored!


Has it not occured to you that there are continuous meetings and reviews amongst those that are actually doing the work?

Also, a lot of that is open to view to the world and hence open to Peer Review. Matt and Rom very professionally add unofficial comment here on this forum which you've likely not noticed (nor understand).

... But when (if) I move fully to BOINC I want it to work at least as well as classic, which to date it doesn't seem to do. Since the credit system of boinc is so obscure then that is another reason for staying.

Sorry, Boinc credits are not obscure and it is all well explained. Sorry #2, the description takes perhaps three sentences rather than one.

Also note that the primary objective and motivation is to do useful Science, both in the data analysis (the WUs) and in developing the project (computer systems & protocols) itself.

Keeping you happy with 'credits' is just literally that: just to keep you happy.


ENOUGH OF THE "Its, broke, its crap, everyone else are idiots, but I'm not a developer so I can't do better" STUPIDNESS.

Try asking a few questions rather than dictating your fantasies that you know better.


Good luck!
Martin

[Edit]
Also note that throughout, WUs continue to be usefully processed uninterrupted. Also, in parallel, Boinc development also continues.

Or is it that you demand all your 'toys' "NOW, THIS VERY INSTANT!" ?
[/Edit]

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

Mr. Cole,

I have read through most of your 85 posts on the forums here. All of them have a negative view of Boinc/Seti as compared to Classic Seti. This would indicate an extreme dissatisfaction for Boinc/Seti.

Why are you still here? Why subject yourself to the cruelty and pain of using Boinc/Seti when it is obvious you don’t want to? Your first post was 176 days ago, why that’s nearly 6 months ago! Why are you still here after nearly 6 months of pain and suffering? Most people would not continue with something they were totally dissatisfied with more than a few days much less 6 months. There must be something keeping you here!

Could it be that Boinc/Seti is improving, bugs being worked out? Could it be that the credit system though not to your personal design does reward you with something? Could it be that the credit system is fair and balanced unlike Classic? Could it be that (gasp) maybe you really do like Boinc/Seti and like seeing the improvements over the last 6 months?

If not then your actions, by staying here, are contrary to your stated opinions!

---

Boinc....Boinc....Boinc....Boinc....

Mr. Cole,

I have read through most of your 85 posts on the forums here. All of them have a negative view of Boinc/Seti as compared to Classic Seti. This would indicate an extreme dissatisfaction for Boinc/Seti.

Why are you still here? Why subject yourself to the cruelty and pain of using Boinc/Seti when it is obvious you don’t want to? Your first post was 176 days ago, why that’s nearly 6 months ago! Why are you still here after nearly 6 months of pain and suffering? Most people would not continue with something they were totally dissatisfied with more than a few days much less 6 months. There must be something keeping you here!

Could it be that Boinc/Seti is improving, bugs being worked out? Could it be that the credit system though not to your personal design does reward you with something? Could it be that the credit system is fair and balanced unlike Classic? Could it be that (gasp) maybe you really do like Boinc/Seti and like seeing the improvements over the last 6 months?

If not then your actions, by staying here, are contrary to your stated opinions!

Actually it is because we need to see some improvements. People who raise issues that need a resolution are the ones driving things forward. Being an uncritical apologist for something with some serious implementation flaws does not fix anything.

If the people who jump down everybody's throats because they had the temerity to suggest that something might not quite be perfect looked at the issues properly then there might be an even better driver to get things right. It is QUITE clear that there are major flaws in the practical implementation of BOINC/SETI. Just look at recent ones at the top of the list. All of them say much the same thing.

The problem is not really that the volunteer participants have to feel they need to keep defending the indefensible but that the project team seem to be so dismissive of those of us keeping their research project (and livelihoods remember) alive. If they responded and acknowledged the issues through the boards or by home page announcements or whatever then that would keep more people happy.

It seems that despite the problems (common to both implementations) that they ar ploughing on regardless. BOINC was supposed to solve what was wrong, yet in reality all it has done is create an additional burden on the team and diverted atteetion away from what needed doing.

We all want this project to succeed, but pretending things are right when they are not will not help. And keeping quiet about things is as bad as pretending nothing is wrong.

Fix the server side of things before fixing (and imposing) the client.

To answer some uyour questions, there is little evidence that things are being fixed. The Search for ETI is a wonderful task. The tools we are given need improving, modern-man is the toolmaker after-all. Actually the credit system is neither here nor there, it is just that the additional complexity and uncertainty is all part of the unnecessary central processing burden. The old system was no less unfair than the current but had the merit of being understood. If people stopped at looked at the wood as well as the trees they would see that the old system largely measured as much as the new. If someone has bigger, faster, or more powerful processors than someone else well, thats life. They will get through more (and higher electricity bills) than those with less. Why use convoluted maths merely to try and hide this?

The new core server structure should have been sorted out and load stressed first. Complicated client side management is less important than clearing all the outstanding queues and making the database, validation and other work properly in a fail resistant manner. That is where the effort should go. Leave the clients which are fundamentally simple till later and then decide whether the additional complexity (and cost) is worth it.

Ultimately the difficulty is that not enough people are sufficiently critical to drive improvements on!

The prophet always gets shouted down!

---

... (Lots of waffle :( )
Ultimately the difficulty is that not enough people are sufficiently critical to drive improvements on!

For myself, I consider that positive constructive enthusiasm works much better.

There's too many ignorant people around that are too keen to blithely proclaim "Its Broke". They then expect whatever they consider broken for their liking to then 'magically' be 'fixed' by someone else for them, and for free, and for no thanks and for negative encouragement.

Yes, very clever. Especially so when their idea of 'broken' is way off from reality!

The prophet always gets shouted down!

I wonder why that might be in your case...

You'd be the first there with the cat o' nine-tails to 'sort 'em out!

Shit, I'd soon sort you out as a manager!


(With the greatest undue respect,)
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

Ultimately the difficulty is that not enough people are sufficiently critical to drive improvements on!

The prophet always gets shouted down!

Ultimately the project is going to do what the project is going to do.

We are not the project.

Criticism is good up to a point, but I seriously doubt that we can affect things as much as some think.

---

The prophet always gets shouted down!

Hmm... Nick, are you calling yourself a prophet? Anyone who must claim to be a prophet is obviously someone not to trust. Someone who must bring the attention to himself is someone who is only interested in the attention but has no validity for himself.

---

Firstly though we need to re-iterate that people are not against BOINC itself.

After stating this, you then proceed to only state that you don't like BOINC. So, if you are not against BOINC itself, what are you against?

What also needs to be remembered is that yes this is posted on the SETI BB so actually while we DO all realise that BOINC can do more than one project, the fact is that it is SETI that is probably of most interest. So any critical comments relating to BOINC are based on our experience of it running SETI.

Well, I too have run both versions of SETI@Home and my enjoyment is undiminished. As I stated in my other post, literlly everything that you could do, wanted to do, or did with SETII@Home Classic is doable with BOINC. A point which you continue to ignore. Exactly how is the BOINC experience worse, other than it is different?

Not everything that the project team need to do has to be publicised, but it is VERY good practice to keep all the stakeholders informed.

You and I are not stakeholders. We are volunteers, just as before. Unlike as before where we had no input on how Classic worked, we have a little chance to input now, if you have something to input. That something from you we are still waiting for. Don't like the benchmarks, like me? Well, we can make our input. I have developed a couple of concepts, found a new and perhaps better benchmark program, both of which I made known to Dr. Anderson. Will he and the developers choose one or the other, or something completely different? I don't know. I don't own that part, and the point you are still missing is that neither do you.

When there is no visual evidence through information that the project team even read the BBs or take any notice of the critics who are motivated by a desire to improve things after all, then it generates concern.

Matt posts here, so does Rom. Not as often as we would like, and obviously you have chosen to ignore the fact that they do post. Other project personnel on other projects also post, we have the mailing lists to which they and others post. And, yes, we also have volunteer developers that read and answer here ... and also write code.

All the people who provide their time and machinery for nothing are stakeholders and without them the concept would not work.

Needed? Sure. Stakeholders, Ok, I can grant that ... but being a stakeholder carries with it no special magic or requirement that the desires be met. If BOIC does not do it for you ... vote ... vote in the only effective way you have ... donate elsewhere ...

I am all the other committed classic operators are NOT against a new system, we just want one that is better than we have at present and when anybody actually looks unemotively at the evidence it is far from better.

You keep harping on the "evidence" and have yet to show any. I always look at evidence unemotively. It comes with autism ... free ... yet, you show no evidence ... a statement that "...it is far from better." is not evidence, it is opinion. And an unsupported one at that.

If of course somebody wants to run several different projects then BOINC is the only practical solution. But that is probably not what a lot of the current classic users are interested in, as evidenced by the sheer weight of numbers.

Wow! Is this a jump in the dark. How do you know they don't have any other interest. A small unfficial study sent out some e-mails to current classic users and the results were divided pretty equally into those that did not know about BOINC, those that did not care too much one way or another, and those that were about to drop the whole DC thing.

So, where is the evidence? Exactly how do you know that this is the case. I can make just as wild of an assertion. Well, no, actually I can back it up with facts. Fact is:

* there are a lot of BOINC participants.
* Many of them have come from the classic project
* Many have had problems, but, have come to appreciate the strngths of BOINC while tolerating its short-comings.
* and if I want up to date numbers of exactly how many people are doing what, well, there are 3-4 statistic sites that will tell me more than was ever possible with Classic. (*I* like the organization of Willy's site, forgive me, you other providers, it is not that you are lacking, I just find his site easier - or maybe it was just that I found it first and can't break the habit - so, everyone, it is a PERFERENCE not an annointment of Willy)
* Matt has stated several times, and he is the guy that does this, that BOINC is far easier to manage, has fewer back-end problems, and is just all around better.
* We have so much more information about what is going on, a much more involved community (and an easier mechanism to post with) that what SEEMS to be more common problems is simply that we are AWARE of more problems. Again, the evidence is from Matt, and independently for the old technical news logs ...

So whatever processing application we use has to be better or cause less problems than what we presently use.

Comparisons with valid work units, repetition and so on do not stand up to independent scrutiny.

It does. And you have never explained how BOINC does not work better. Until you do, this is once again, an unsupported allegation with no basis in reality. So, once again, we would love to debate you, explain to you, discuss with you the issues. But we still have no idea what you think the issues are, until you tell us, well, your statements to this point still amount to "I don't like BOINC", which is interesting, but hardly relevant.

Out of the vast hundreds of thousands of contributors there is a very strong likelihood that the collective experience and knowledge of operating, designing and installing large computer systems, major projects and so on which massively exceeds the experience of the small (and not to denigrate them at all) project team. They are doing a job and the only ones who are in a position to do it, but do not seem to tap into the pool of knowledge that is there trying to help them. That also is at the heart of all the comments.

Actually, again, you ignore the facts. Matt listens, Rom listens, JM VII listens, heck I have even seen Dr. Anderson listening ... that does not mean that I, or anyone else for that matter, "wins" the discussion point. But, fact is, a lot of what went into BOINC came from them listening to us. We complained in classic about cheaters, and well, I suppose it can still be done, but, it sure is harder.

Comments highlighting flaws and problems are not negative. Distributed computing for the greater good is not negative, but it has to work properly. That is all we are asking.

This is true. And as soon as you highlight a flaw we will sure appreciate it. BOINC Stinks is not a description of a flaw, but an opinion. And if BOINC is not working properly, how is work getting done. That is the true definition of working. Is it flawless? No. Never said that. Likely never will say that. But, I have been contributing work for nearly two years now. And it is all valid work. Even now when it is supposidly completely inoperative from your biased viewpoint, well, I am still doing work, returning it, haveing it validated and creditied ... more than likely, so are you ...

Address the problems don't ignore them. And the problems are NOT caused by those people brave enough to highlight them.

If you bothered to read anything posted here, in technical news, or the replies directly to you, they are addressing the problems.

The point is that BOINC (as is classic) is a complex system. It has clients, servers and communications. While there were functional issues with classic command line client, it worked extremely well, and with the gui add-ons superlatively. Fully controllable, proper queuing, statistic generating, etc. the communications were not a major problem generally other than we have all experienced over the last year or so and affecting BOINC as well (anecdotally far more so in fact).

BOINC works just as well, has the same add-ons, and some others too ... you just won't admit it.

The real problems lie in the server infrastructure. While there were issues with validation, this has absolutely nothing to do with the client software that we run. And a solution to this problem was systemically extremely easy to resolve, even without knowledge of how they have structured it. However instead of concentrating on fixing this, a new client and multi-project capability was built. While some of the server side issues were resolved ... and now there are a massive volume of complaints and issues around the new client as well!

You are correct, that some of the server-side issues could have been corrected independently with no need to release a new client. However, the point you choose to ignore is that even for SETI@Home there is not going to be just one client much longer. UCB has 3-4 projects, all under SETI@Home that are going to have their own client side applications. The classic client not only could not deal with that, your admiration for all the add-on tools ignores the issue that they should not have been so necessary. Though I do like logging, it is not a core issue and as in classic, actually there are now about 3 different ways to log data now; anyway, data logging may be added in the future. Just as the statistics tab was added by, you guessed it, a non-UCB, volunteer, developer.

So, even without projects outside of the UCB SETI@Home project, there is a need for a new client, and client framework. Thus, BOINC ...

This is where the design review needs to have been kicked off, and it should have happened sooner rather than later. None of this is against distributed computing or BOINC per se, but given the range of issues it is still very much at beta stage.

In my opinion, so is the initial release of every Microsoft OS. SP1 is the point where it is really a production release. Though, as an OS/2 user for a decade, I have to tell you that XP Pro is still not as robust and functional as OS/2 Warp SP4 was a decade ago.

But beta or not. It is working well enough for us to do science. And if you don't want to do BOINC right now. Cool, no one is forcing you to ... do classic till it closes ...

Unfortunately the user population is being taken for granted, and while there may well be useful add-ons, documentation (which I have read and leaves more questions than are answered), volunteer support, and so on there is NOTHING on this web site that lists or links to any of it.

What questions? I don't recall seeing any. Lacks in the Wiki? Sure, tons. It is as good and as complete as I can make it, and health permitting I try to work on it every day. When able, my "normal" work day is about 12 hours doing little but adding, correcting, etc. the content.

I, and others, answer questions here, I answer questions sent to me directly, and I add material submitted, correct mistakes pointed out to the extent of my time. There are also nearly 20 other people with access to the Wiki and even on the days where I do little or nothing, well, I don't think I have seen a day go by without at least one change.

If there is a lack, where is it? What question did we leave unanswered? You make these sweeping generalizations and, as I stated before in this post, and in my other post, exactly what is wrong? Saying the documentation is leaving questions unanswered is easy, it is just as easy to say what question you have that was not answered.

And, yes, it is unofficial. I do have reprint from UCB by Dr. Andersons permission directly, and by the Open Source rights as indicated. Heck, I even have a matrix which links the UCB pages to mine, and on all of mine should be a link to the official UCB page. All I do is clean up the "sloppy" (in PAUL"S OPINION) use of termonology. Correct spelling mistakes and other outright errors. As i get time I will be adding material there, I just have not gotten to it yet ...

I spend some time when I feel motivated enough answering these points in the vain attempt that somebody at project level will do something, other than rely on volunteer responses which do nothing to address the problems as the volunteers are also not in a position to do anything. In fact ignoring us volunteers has been a fairly consistent theme. Back in 1999 I was in a position to supply 500 (then good) PCs available 24hr per day, unused for 16hrs to participate in the project. However I needed to discuss security and a number of other matters, did anybody reply? No.

Exactly what problem? Other than Nick Cole is unhappy, what is/are the problem/s? And, as stated by myself and others, they are working on it.

I am not a developer, I am a systems engineer, I am an experienced project manager. I do not criticise the code, but the structure and how it is put together and crucially how it all works together. And it is the system as a whole that needs to be reviewed, not necessarily small aspects of it. Loading, network performance, UPS (even), database structures are all part of the system and it is visibly obvious that these are the areas where there are problems, and to repeat yes, it is SETI, not BOINC.

Well, I too am a systems engineer. And the other point I have raised is that BOINC is tolerant in the presence of error. And it is. Is the validator number climbing? So what? So is the national deficit. The fact of the matter is that this is a number. They are working on what seems to have been identified as a new problem that to this point they were unaware of its existance. Regardless of the quality of the design, choke points happen. And they usually happen in places that are not expected. There is no point in expending resources trying to find them in advance, but to handle them when the crop up ... this is a good and practical practice and it was what was done. Limited budget, as a project manager what would you have done differently? "Gee boss I need to spend 100K on a potential bottle neck ...", yeah, you would approve of that one in a shot ...

If you had bothered to read any of the posts by Matt, or the analysis of the problems in the technical news (and it is obvious to me you cannot have done so in light of the last statements) they have been doing end-to-end reviews of the system to find the bottleneck. Now we are trying changes and in the mean time continuing to march on.

Yes there are more features in the new gui than the old vanilla client. But, less features and manageability than with the add-ons. The add-ons have been around for several years predating the BOINC gui which should by definition be better all round, which it isn't. It is extremely easy to use the Setistash and others) gui to stop processing or transferring wus. GUIs can also generate text files. I haven't written any text files to run 4 instances of classic. Since we are talking about a recent system then it is entirely fair to compare the new with what it is intended to replace.

That is what evolution is about. (Or are the pro-boincers closet creationists?)

I don't get this at all. With the stock BOINC GUI I get more features but less managability? With BOINC I can do what you say you need an add-on for with Classic. Sorry, but a contridiction. I can queue work, I can turn on and off the uploads and downloads all without any other programs. How is Classic better when you need setistash to do what comes in the stock BOINC Client?

So what has validation got to do with it? Fix that, don't introduce further validation backlogs and other problems. Credits are a multi-headed issue. The BOINC method is confusing, but is a counting mechanism. Classic is clear and immediately gives credit for the efforts put in. BOINC only gives credit if what you have been given to process does something valid or useful. This will inevitably underscore some people's efforts and boost others, but in a way that cannot be compared or established. If we spend 4 hrs processing a wu then we have done that much work. If it wasn't worth it or the process aborted or gave an invalid result then how are we to control it? We can only proces what we are given. We have no control over this aspect, and what about all that wasted time while the validation/credit system stabilises with each new client, why does it process things past the deadline, etc etc?

Actually, I was going to pare this down, but it is so, um, well, no, I won't say it. Seriously, you need to review science principles and how valid science is done. We do expiriments, we redo them, then we do them again. In all cases we check as much as we can and validate as much as we can. Why? Because it is not science if it is not valid. The classic side validation took place behind the scenes, and if I uderstood Matts post we are going to be validating classic work a year after classic closes.

So, in classic, I turn in a bogus result that I faked in 10 seconds of time and it is as valuable as the one you took 10 hours to produce. My, you strech my concept of fairness ...

The result you calculate, in some cases can have a detected error, and if so, is aborted with a "Client Error" outcome. In other cases, the numbers make "sense" to the client program, but don't match reality. In this case, comparison with other calculated results are used to reject the error. In another thread there is a client that has 1,000 results (supposidly - I did not check) all of which are invalid. Sorry, I don't think that person deserves much of anything. I tell you what, send me 1,000 and I will super tune your car for you so you get 100 miles to the gallon ...

But the only thing we get out of it is the recognition of our credits. However, it is a university research project which presumably is going to advantage the faculty, professors, students, (pre and post grad). So they should give us something for helping them.

They do. The opportunity to volunteer ... or not ... nothing else was promised.... another point you seem to miss ...

But the answer is that to keep ALL the stakeholders happy we need to be informed as well. We are not all sheep here merely for the benefit of the students and professors at Berkeley. And again as mentioned above there is a vast untapped pool of experience in all of us.

Why do they need to keep you happy? That is not what the project is about. The project is about doing science. Everything else is secondary. Including holding your, or even my hand.

But when (if) I move fully to BOINC I want it to work at least as well as classic, which to date it doesn't seem to do. Since the credit system of boinc is so obscure then that is another reason for staying.

In my opinion, if you were less wrapped up in a biased opinion you would find BOINC does much better than Classic. Heck, after a long time I finally took the plunge (in part so I could write it up) and started to run optimized client applications. Now I do more science per unit time. Can't do that with classic ...

Ok, just so you know. At this time, I am firmly of the opinion that you have no specific complaints. You have none at all. Repeated requests for specifics have been ignored. Pointing out places where your point are refuted also seem to be ignored. So, bear this in mind all of this, in my last two posts was not intended to try to change what is apparently a closed mind with no specifics, no data, and nothing other than unsupported opinions; but, rather was for the other people who might read your post and fail to understand the total lack of content.

I can, and do, debate the merits of issues with BOINC, the last in a long series was the one about multi-queueing - a specific design choice that may be visited in the future, or sooner if someone writes code to implement it, but a choice none the less. If, and when, you do have someting specific by all means, let us discuss that problem.

In all honesty, I am embarrassed for you for the way that you have presented your opinion. Devoid of content and specifics it comes across as, um, no, I won't go there either. Seriously, when you do want to get serious, come back and we can talk. If you think I have missed a "fact" in any of your posts, well, list them out ... I looked, but found nary a one ...

---

Actually it is because we need to see some improvements. People who raise issues that need a resolution are the ones driving things forward. Being an uncritical apologist for something with some serious implementation flaws does not fix anything.

True. We need to see improvements. And for those of us with open eyes, we can see some. Not every day, but over time. This second line is the most common cop-out in these tirades. If you can't win with facts, label anyone that does not agree with you as "an uncritical apologist". Quite far from the truth. Worse, it is a demonstration that you have nothing to say.

Once agian you state "some serious implementation flaws" with no indication of what they might be. How is this to help move things forward?

If the people who jump down everybody's throats because they had the temerity to suggest that something might not quite be perfect looked at the issues properly then there might be an even better driver to get things right. It is QUITE clear that there are major flaws in the practical implementation of BOINC/SETI. Just look at recent ones at the top of the list. All of them say much the same thing.

We are not jumping down your throat because you are complaining. We are stating that you have not made a case. Nor have you identified one issue. What exactly has you bothered? What does not work the way you expect? And why is that problem preventing science from being done?

Just look at WHAT at the top of WHAT list? Where is this list? What problems are one it? Good lord man, what is broken?

The problem is not really that the volunteer participants have to feel they need to keep defending the indefensible but that the project team seem to be so dismissive of those of us keeping their research project (and livelihoods remember) alive. If they responded and acknowledged the issues through the boards or by home page announcements or whatever then that would keep more people happy.

Do you know why there is a technical news page? A status page? Um, well, you are not going to like this ... we asked for them ... and we got them ... how is that being ignored. Exactly which problem is being ignored?

It seems that despite the problems (common to both implementations) that they ar ploughing on regardless. BOINC was supposed to solve what was wrong, yet in reality all it has done is create an additional burden on the team and diverted atteetion away from what needed doing.

Heck read my other post.

We all want this project to succeed, but pretending things are right when they are not will not help. And keeping quiet about things is as bad as pretending nothing is wrong.

None of us are pretending that it is right when it is wrong. But did you consider that you might be doing the converse? Assuming everything is wrong because you don't know what is right?

Fix the server side of things before fixing (and imposing) the client.

See my other post. Again, both sides needed enough changes that it was simpler and better to throw away the current and start clean. Granted it has been a longer road ... but, how exactly is the system NOT WORKING NOW?

To answer some uyour questions, there is little evidence that things are being fixed.

I subscribe to the CVS mailing list. Since the last time I was healthy enough to read it down there have been 938 changes to the baseline. If you subscribe, you too can see that there are changes being made. If you care to open your eyes.

The old system was no less unfair than the current but had the merit of being understood.

This is nice and insulting. You imply that no one understands the new credit system. Yes, it is a more complex system. But, also, vastly more fair. Return bogus science, no credit. And, if the explanation in the Wiki is confusing ... point at where and we can fix that ...

If people stopped at looked at the wood as well as the trees they would see that the old system largely measured as much as the new.

From my side of the keyboard I can make the same accusation. I ask for specifics and mostly what you do is call people names and continue the litany of "BOINC is bad".

The new core server structure should have been sorted out and load stressed first. Complicated client side management is less important than clearing all the outstanding queues and making the database, validation and other work properly in a fail resistant manner. That is where the effort should go. Leave the clients which are fundamentally simple till later and then decide whether the additional complexity (and cost) is worth it.

Must be nice to have unlimited budgets. Oh, and time. The point of this, again, incase you missed it, is to do science. If we do a little science everyday, then that was a successful day. Today, like yesterday, and for a long time prior, well science is being done.

Ultimately the difficulty is that not enough people are sufficiently critical to drive improvements on!

Good grief Charlie Brown! You can be critical of the system. Heck, you can be critical and specific. So far, all you are is critical. Of the system, and of all of us who keep asking "What exactly do you think is broken in BOINC?"

When are you going to answer that question?

The prophet always gets shouted down!

The only shouting we are doing, in a vain attempt to get a response from you that is not just opinion or unhappy feelings, but factual, concrete specifics.

Your inability to get beyond "BOINC is broke" is what drives us up the wall. Until you answer that, it is all just noise. So, start from the beginning. Exactly what is broken? Simple one liners. A simple list. What is broken ... here I will start:

1. Benchmarks are inconsistent and inaccurate indicators of performance.
2. Nick Cole cannot say what he things is wrong with BOINC.

Make your list below ...

---