"Drop all credit scoring.... It's all about the science."
or
"Project X sux, cause my RAC dropped 1 point!"
:-)
DownUnder we have a road law which basically states 'drive as close to the left as practicable'. This discourages driving next to the middle of the road lines, for the clear reason that the oncoming traffic ( on undivided highway ) maybe doing the same - hence risking an horrific head-on collision. Too far to the left puts the car onto the generally rough edges, less traction etc and loss of control et al. The middle of a lane is good, but to the left of the lane's midline is better. There is no exact specification for the distances involved because that could only sensibly depend on a multitiude of factors which a statute law could not predict, and in any case drivers couldn't precisely adhere, nor could the police accurately assess for either conformance or the breach.
The analogy here at E@H is clear. The administrators of the project have chosen a formulation which will achieve the science 'destination', with minimal risk, a modicum of speed, and not too much micro-management. Unsurprisingly this could be labelled as a 'middle of the lane approach'.
Since I've been here ( December 2005 ) I have witnessed a moderate amount of oscillation in the lane, significant progress none-the-less, a good deal of enjoyable enterprise and the odd squeal of the tyres.
While taking care not to over stretch a useful analogy, I'd say we are travelling very well! :-)
This is certainly true considering the sheer number of passengers, and the refreshingly disparate aims expressed! :-)
As you can deduce from my posts, I'm generally synthetic in my comments ie. emphasising the unities present here rather than highlight the distinctions. But I ought perhaps politely frown ( nothing personal ) at those who seek solely their own benefit by participation at E@H. This is an historic enterprise. And no, the argument that one's donation of time/resources enables a license to disrupt the essential aim of the project ( new and exciting science! ) regardless does not wash with me .....
There. I've vented my spleen, such as it is. :-)
Cheers, Mike.
( edit ) It occurs to me that I ought explicitly belabour the following points to forestall objections from those who haven't gleaned any of my prior posts:
- those who care only about the science must care about the concerns of those who don't. This way the credit can accumulate for the latter, and the science is done irrespective of their motivation.
- those who care only about the credit should consider the following. Suppose I have a project called IAC@H. This has all the usual BOINC structure, the WU's simply count the number of integers between successive multiples of one hundred. We start at 1 - 100 and continue on to infinity. Call it 'Integer Axiom Concordance At Home' as it verifies that each century interval has precisely one hundred integers ( which would test that multiple successive additions are equivalent to a single block subtraction ). As this is clearly lucidrous you could easily re-title as 'Inane Aimless Computation At Home'. Would any credit focussed participants be satisfied with credit from such a silly activity? If you would, then stop reading here, as I won't affect your view - ever. If you wouldn't, then why? Hopefully it is because the credit, by proxy, means something in the world outside of the computer! Here at E@H it is the ( eventual ) detection of gravity waves.
- the likely reality is that we are all at least a bit of each. I'll admit to earlier being real keen to get into the top 100 for RAC. I did that quite a while ago, briefly, but having done it I'm now indifferent. Mind you I could be bluffing you all here, as I see it's alot steeper slope there now. So I could be faking lack of concern ... :-)
( You have to crack about 4500 to sneak in now, I did it at just over 3000 - how far we come! )
So I'm actually talking about how we deal with each other in this collaborative enterprise! :-)
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
You are right. Lots of scientists examined the official application before that was issued, so they accepts the results of the official application. They didn't examined the modified apps, so they will not accept its results.
I understand that they want less speed and more confidence. But if by censuring Akos work (at least the unofficial one) they think they will be safe is wrong. If they want to be absolutely sure, then they have to quit using BOINC, because simply there is no way to be safe when the job is done by hundred of thousands of computers they have not control.
For example, any hacker with 10% of Akosf´s skills could change the optimized S5 and replace the text output so it mimics the stock one. In that case, there is simply no way to detect it on the database because the validator already accepts the results… How in the world there are going to distingue between stock applications results and the hacked ones??. So, in my opinion, if you use BOINC or any remote execution solution, you have to presume that some percent of WU will be executed by non standard application, and you have to be prepared for that, by means of a very strong validator. And if you do a validator well enough, you don’t care about any optimizations done, because you know that if it pass it, the calculation is as goods as it gets.
Mi final point is this: I won´t crunch any longer for EH, because they want the benefits of free distribuited computing without understanding the nature of Internet itself. It’s a wonderful world, with many incredible people like Akosf who will double your processing out of the blue… but with thousand of hackers who will make your job difficult. If you don´t understand that… don´t use it… Go back to your sterilized computer labs….
You are right. Lots of scientists examined the official application before that was issued, so they accepts the results of the official application. They didn't examined the modified apps, so they will not accept its results.
I understand that they want less speed and more confidence. But if by censuring Akos work (at least the unofficial one) they think they will be safe is wrong. If they want to be absolutely sure, then they have to quit using BOINC, because simply there is no way to be safe when the job is done by hundred of thousands of computers they have not control.
For example, any hacker with 10% of Akosf´s skills could change the optimized S5 and replace the text output so it mimics the stock one. In that case, there is simply no way to detect it on the database because the validator already accepts the results… How in the world there are going to distingue between stock applications results and the hacked ones??. So, in my opinion, if you use BOINC or any remote execution solution, you have to presume that some percent of WU will be executed by non standard application, and you have to be prepared for that, by means of a very strong validator. And if you do a validator well enough, you don’t care about any optimizations done, because you know that if it pass it, the calculation is as goods as it gets.
Mi final point is this: I won´t crunch any longer for EH, because they want the benefits of free distribuited computing without understanding the nature of Internet itself. It’s a wonderful world, with many incredible people like Akosf who will double your processing out of the blue… but with thousand of hackers who will make your job difficult. If you don´t understand that… don´t use it… Go back to your sterilized computer labs….
Excellent points.
The reality is that the signal processing would not be done in any reasonable time frame without some form of distributed computing. A glance at the server page shows some 60 TFLOPS ( potentially 90 ) available in this fashion. In addition to upfront cost, and the maintainence, and electricity and ...... this is out of the price range for the project's budget. This is why the volunteer behaviour of E@H participants is so valuable, and why considerable lengths are gone to to keep the users as content as possible.
I guess someone might ( even already ) be 'spoofing' the project. For the life of me I couldn't dream why ( what's in it for them? ), but aside from that how could that corrupt the science result unless it was widespread? The application and WU distribution, quorum and validation system would severely negate the impact. It's not just text response .... you'd have to fool around to the level of producing an apparently valid result file ( by it's nature, this contains numbers which could indicate a detection of some type to some confidence level ). In any case I would strongly suspect any 'positive' signal would be cross-checked 'in-house' rigourously before any scientist ( or group ) would lay their reputation on it before publishing. I'd always assumed my participation was in a screening ( catch if you can ) role.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
The only ones that can say if the science in reguard to changes to the cruncher app are valid, are those directly involved with the science (cruncher) apps from the project.
Seems those folks have spoken. Why beat a dead horse?
Seems those folks have spoken. Why beat a dead horse?
Could somebody please point me to where "those folks" have spoken on this subject (and I mean Bruce or Berndt). A couple of days ago, I posted this message here because a lot of people posting here were "jumping to conclusions" about what had happened. The frenzy has died down some but I still haven't seen anything I would accept as an "official" (although I could have missed it.) Instead, we have a lot of very good, unanswered questions. Bruce, et. al. could put this issue to rest by posting an "official" explanation. I think they "owe" it to Akos.
Seems those folks have spoken. Why beat a dead horse?
Could somebody please point me to where "those folks" have spoken on this subject (and I mean Bruce or Berndt). A couple of days ago, I posted this message here because a lot of people posting here were "jumping to conclusions" about what had happened. The frenzy has died down some but I still haven't seen anything I would accept as an "official" (although I could have missed it.) Instead, we have a lot of very good, unanswered questions. Bruce, et. al. could put this issue to rest by posting an "official" explanation. I think they "owe" it to Akos.
The person who made the optimized applications has stated to stop using them, several places, but here is one of the main ones.
Since he is in the loop with the developers, and has stopped creating, stopped supplying, and explains everyone should stop using them, I believe this is an official word. I am betting that the developers told him to stop.
Interesting that nobody required Bruce's blessing to install untested, unofficial, possibly buggy science apps on all their machines five days ago... but now when we're asked to remove those apps suddenly people are saying "not until Bruce says I have to!" LoL. It's quite funny to watch actually.
. . . but now when we're asked to remove those apps suddenly people are saying "not until Bruce says I have to!"
Not at all my point. All I am saying is that Project Developers should feel some obligation to speak out on matters like this. After all, without our computing time, the project goes nowhere. I, for one, would just like to know what's going on. The posts on this thread, though often entertaining, shouldn't be confused with information.
RE: Agreed! I don't know
)
DownUnder we have a road law which basically states 'drive as close to the left as practicable'. This discourages driving next to the middle of the road lines, for the clear reason that the oncoming traffic ( on undivided highway ) maybe doing the same - hence risking an horrific head-on collision. Too far to the left puts the car onto the generally rough edges, less traction etc and loss of control et al. The middle of a lane is good, but to the left of the lane's midline is better. There is no exact specification for the distances involved because that could only sensibly depend on a multitiude of factors which a statute law could not predict, and in any case drivers couldn't precisely adhere, nor could the police accurately assess for either conformance or the breach.
The analogy here at E@H is clear. The administrators of the project have chosen a formulation which will achieve the science 'destination', with minimal risk, a modicum of speed, and not too much micro-management. Unsurprisingly this could be labelled as a 'middle of the lane approach'.
Since I've been here ( December 2005 ) I have witnessed a moderate amount of oscillation in the lane, significant progress none-the-less, a good deal of enjoyable enterprise and the odd squeal of the tyres.
While taking care not to over stretch a useful analogy, I'd say we are travelling very well! :-)
This is certainly true considering the sheer number of passengers, and the refreshingly disparate aims expressed! :-)
As you can deduce from my posts, I'm generally synthetic in my comments ie. emphasising the unities present here rather than highlight the distinctions. But I ought perhaps politely frown ( nothing personal ) at those who seek solely their own benefit by participation at E@H. This is an historic enterprise. And no, the argument that one's donation of time/resources enables a license to disrupt the essential aim of the project ( new and exciting science! ) regardless does not wash with me .....
There. I've vented my spleen, such as it is. :-)
Cheers, Mike.
( edit ) It occurs to me that I ought explicitly belabour the following points to forestall objections from those who haven't gleaned any of my prior posts:
- those who care only about the science must care about the concerns of those who don't. This way the credit can accumulate for the latter, and the science is done irrespective of their motivation.
- those who care only about the credit should consider the following. Suppose I have a project called IAC@H. This has all the usual BOINC structure, the WU's simply count the number of integers between successive multiples of one hundred. We start at 1 - 100 and continue on to infinity. Call it 'Integer Axiom Concordance At Home' as it verifies that each century interval has precisely one hundred integers ( which would test that multiple successive additions are equivalent to a single block subtraction ). As this is clearly lucidrous you could easily re-title as 'Inane Aimless Computation At Home'. Would any credit focussed participants be satisfied with credit from such a silly activity? If you would, then stop reading here, as I won't affect your view - ever. If you wouldn't, then why? Hopefully it is because the credit, by proxy, means something in the world outside of the computer! Here at E@H it is the ( eventual ) detection of gravity waves.
- the likely reality is that we are all at least a bit of each. I'll admit to earlier being real keen to get into the top 100 for RAC. I did that quite a while ago, briefly, but having done it I'm now indifferent. Mind you I could be bluffing you all here, as I see it's alot steeper slope there now. So I could be faking lack of concern ... :-)
( You have to crack about 4500 to sneak in now, I did it at just over 3000 - how far we come! )
So I'm actually talking about how we deal with each other in this collaborative enterprise! :-)
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: You are right. Lots of
)
I understand that they want less speed and more confidence. But if by censuring Akos work (at least the unofficial one) they think they will be safe is wrong. If they want to be absolutely sure, then they have to quit using BOINC, because simply there is no way to be safe when the job is done by hundred of thousands of computers they have not control.
For example, any hacker with 10% of Akosf´s skills could change the optimized S5 and replace the text output so it mimics the stock one. In that case, there is simply no way to detect it on the database because the validator already accepts the results… How in the world there are going to distingue between stock applications results and the hacked ones??. So, in my opinion, if you use BOINC or any remote execution solution, you have to presume that some percent of WU will be executed by non standard application, and you have to be prepared for that, by means of a very strong validator. And if you do a validator well enough, you don’t care about any optimizations done, because you know that if it pass it, the calculation is as goods as it gets.
Mi final point is this: I won´t crunch any longer for EH, because they want the benefits of free distribuited computing without understanding the nature of Internet itself. It’s a wonderful world, with many incredible people like Akosf who will double your processing out of the blue… but with thousand of hackers who will make your job difficult. If you don´t understand that… don´t use it… Go back to your sterilized computer labs….
RE: RE: You are right.
)
Excellent points.
The reality is that the signal processing would not be done in any reasonable time frame without some form of distributed computing. A glance at the server page shows some 60 TFLOPS ( potentially 90 ) available in this fashion. In addition to upfront cost, and the maintainence, and electricity and ...... this is out of the price range for the project's budget. This is why the volunteer behaviour of E@H participants is so valuable, and why considerable lengths are gone to to keep the users as content as possible.
I guess someone might ( even already ) be 'spoofing' the project. For the life of me I couldn't dream why ( what's in it for them? ), but aside from that how could that corrupt the science result unless it was widespread? The application and WU distribution, quorum and validation system would severely negate the impact. It's not just text response .... you'd have to fool around to the level of producing an apparently valid result file ( by it's nature, this contains numbers which could indicate a detection of some type to some confidence level ). In any case I would strongly suspect any 'positive' signal would be cross-checked 'in-house' rigourously before any scientist ( or group ) would lay their reputation on it before publishing. I'd always assumed my participation was in a screening ( catch if you can ) role.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
The only ones that can say if
)
The only ones that can say if the science in reguard to changes to the cruncher app are valid, are those directly involved with the science (cruncher) apps from the project.
Seems those folks have spoken. Why beat a dead horse?
RE: Seems those folks have
)
Could somebody please point me to where "those folks" have spoken on this subject (and I mean Bruce or Berndt). A couple of days ago, I posted this message here because a lot of people posting here were "jumping to conclusions" about what had happened. The frenzy has died down some but I still haven't seen anything I would accept as an "official" (although I could have missed it.) Instead, we have a lot of very good, unanswered questions. Bruce, et. al. could put this issue to rest by posting an "official" explanation. I think they "owe" it to Akos.
RE: Seems those folks have
)
They fight back less than live ones.
Try the Pizza@Home project, good crunching.
RE: RE: Seems those folks
)
The person who made the optimized applications has stated to stop using them, several places, but here is one of the main ones.
Since he is in the loop with the developers, and has stopped creating, stopped supplying, and explains everyone should stop using them, I believe this is an official word. I am betting that the developers told him to stop.
Interesting that nobody
)
Interesting that nobody required Bruce's blessing to install untested, unofficial, possibly buggy science apps on all their machines five days ago... but now when we're asked to remove those apps suddenly people are saying "not until Bruce says I have to!" LoL. It's quite funny to watch actually.
RE: . . . but now when
)
Not at all my point. All I am saying is that Project Developers should feel some obligation to speak out on matters like this. After all, without our computing time, the project goes nowhere. I, for one, would just like to know what's going on. The posts on this thread, though often entertaining, shouldn't be confused with information.
And so long you use
)
And so long you use unofficially clients....? I don't know what to say.
Akosf and the project leader asked us to stop using the opt clients.