Low credit for such a risky project

KPX
KPX
Joined: 23 May 05
Posts: 8
Credit: 5931956
RAC: 0
Topic 84782

Just 500 credit for 20+ hours of crunching is not much, given the risks associated with these units, with testing where, as you say, "Don't expect ANYTHING to work here". Perhaps we deserve a bonus for crunching for such an uncertain project, wouldn't you think?

Bernd Machenschalk
Bernd Machenschalk
Administrator
Joined: 15 Oct 04
Posts: 155
Credit: 6218130
RAC: 0

Low credit for such a risky project

No.

Also from the front page: "Most of the time Albert@Home will have no work to run at all, and when it has, the applcations are experimental, might be unstable, unreliable and may even damage your computer. Validation might be unreliable and we may cancel workunits without prior notice. If you care about credit, this project is certainly not the right one for you."

Also see my post here. Albert@home doesn't even export stats.

BM

BM

KPX
KPX
Joined: 23 May 05
Posts: 8
Credit: 5931956
RAC: 0

Hm. Credit is the only

Message 78753 in response to message 78752

Hm. Credit is the only currency we crunchers get for crunching. Strange that you feel giving credit would cost you something... All u get for credit is more crunchers. Is that bad for this project?
And why not to export stats? All projects test stuff from time to time. So what if servers need to be reset, units canceled, etc., it can happen in other projects as well. I crunch many alpha and beta projects. And they still try to treat crunchers somewhat normally...
But from your post I'm getting the feeling that discussion is not appropriate here, perhaps even asking questions like this. Well, couldn't help it. If things don't feel right, they don't feel right.
Otherwise, good luck in testing and developing the project. And this I mean sincerely.

pragmatic prancing periodic problem child, left
pragmatic pranc...
Joined: 26 Jan 05
Posts: 153
Credit: 70000
RAC: 0

Albert@Home isn't an alpha or

Message 78754 in response to message 78753

Albert@Home isn't an alpha or beta project, it's merely a testing ground for new applications for Einstein@Home. It's no use to attach your whole farm here, there's only need for so much testing.

Once the application and its back-end (feeder, assimilator, validator, scheduler) is deemed to work without too many problems, it can go to the main project, Einstein, where you can rack up all the credit in the world for it.

zombie67 [MM]
zombie67 [MM]
Joined: 10 Oct 06
Posts: 73
Credit: 30924459
RAC: 0

Several projects have

Message 78755 in response to message 78752

Several projects have alternate development sites (RALPH, SETI Beta, etc), and they *all* issue credits. They export them too. No reason to *not* award credits for work done.

As for how much/little to award, it doesn't matter, so long as the credit remain internally consistent. Cross-project parity is impossible anyway.

Dublin, California
Team: SETI.USA

Bernd Machenschalk
Bernd Machenschalk
Administrator
Joined: 15 Oct 04
Posts: 155
Credit: 6218130
RAC: 0

RE: Albert@Home isn't an

Message 78756 in response to message 78755

Quote:
Albert@Home isn't an alpha or beta project, it's merely a testing ground for new applications for Einstein@Home.

[nitpick] Not precisely. We can and would test new applications (and application versions) either privately or on the main project. What we intend to test on Albert is server side code that can't be tested on the main project without risking serious damage. The OpenCL support of BOINC requires server side modifications, too, that's why we test the OpenCL apps here. [/nitpick]

BTW: In what sense is Albert@Home different from an alpha test project, in particular related to credit?

Quote:
Several projects have alternate development sites (RALPH, SETI Beta, etc), and they *all* issue credits. They export them too. No reason to *not* award credits for work done.

I asked David Anderson about how they handle credit "on Alpha test projects". His answer was "I ignore them, and I don't run the script that exports statistics.". Thus I decided to do the same here. Did I get wrong information?

The point for me is that having to care about credit generates an amount of work (thinking, maintenance) that I don't want to commit to for a project where I may want to test something completely different. If workunits turn out to don't work for an application I want to cancel these. If for some malfunction of me or a daemon (or just for comparison) I decide to restore an old copy of the database, I will do this. If I just want to watch how one credit system compares to a different one, I'd like to run both systems in parallel or one after the other. In any of these cases (and many more) I don't want to spend time thinking about and taking measurements against people losing their valuable credits, and won't apologize for anything that goes wrong. Something going wrong is the normal case on test projects, it's the reason why these are there.

I think I didn't miss any opportunity (front page, announcements, forum posts) to point out that I don't care about credit over here, and don't want to have to; and that if you do, you shouldn't run Albert@Home but sign up for the main project.

This decision isn't final, though. When certain development currently going on here has settled a bit and we need a larger community for testing something else, I may find that worth the work to commit to proper "credit management". But not right now. I certainly have all my hands full with a lot of other things.

BM

BM

ChertseyAl
ChertseyAl
Joined: 23 Nov 05
Posts: 6
Credit: 110625
RAC: 0

That's all fair enough, and

Message 78757 in response to message 78756

That's all fair enough, and you've certainly made it clear in enough places regarding credit :)

But just a couple of points:

The hardcore milestone collectors will crunch *any* project that's available to them (that includes me BTW, 110 projects and rising). I took the risk, got some credit, quite happy despite losing loads of pending credit a while ago.

Your user base was inherited from Einstein, so you automatically got a whole bunch of people who were bound to be interested. Maybe if you'd started out as 'invite code only' ... But I can see that would have taken more time/effort than this approach.

As for turning off stats export, how about either leaving it running until a clearly defined date, or run it just once more at some time in the future. Maybe set the cutoff time to the dealine of the WUs (2 weeks IIRC?). And announce it in big red capitals on the front page ;) Perhaps you could email it to everyone, but I don't know how much effort that would take.

Personally, I'm not bothered whatever happens. But I can understand the angst of others :)

Cheers,

Al.

Bernd Machenschalk
Bernd Machenschalk
Administrator
Joined: 15 Oct 04
Posts: 155
Credit: 6218130
RAC: 0

RE: Your user base was

Message 78758 in response to message 78757

Quote:
Your user base was inherited from Einstein,

The user table was copied from Einstein. But no host that was attached to Einstein was automatically attached to Albert.

Quote:
As for turning off stats export, how about either leaving it running until a clearly defined date

That's certainly worth a thought. I'm still pretty busy with other things (download server), but I'll try to arrange for that.

BM

BM

zombie67 [MM]
zombie67 [MM]
Joined: 10 Oct 06
Posts: 73
Credit: 30924459
RAC: 0

RE: RE: Several projects

Message 78759 in response to message 78756

Quote:
Quote:
Several projects have alternate development sites (RALPH, SETI Beta, etc), and they *all* issue credits. They export them too. No reason to *not* award credits for work done.

I asked David Anderson about how they handle credit "on Alpha test projects". His answer was "I ignore them, and I don't run the script that exports statistics.". Thus I decided to do the same here. Did I get wrong information?

Yes, DA was not accurate.

DA has one "test BOINC clients" project, aka BOINC alpha, does not do real work (just UPPERCASE), and *does not* export stats.

On the other hand, his SETI Beta project, which tests new apps and sub-projects, and does real work, *does* exports stats.

And RALPH, which is the alpha project for Rosetta@home, which tests new apps, and does real work, *does* exports stats.

Amelegrid Testgrid (an active test project for Amelegrid), also exports stats.

ABC@home also had a test project, and exported stats.

Malaria Control had a test project, and they exported stats.

Cels@home also had a test project, and they exported stats.

Dublin, California
Team: SETI.USA

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 10
Credit: 9256493
RAC: 0

I'm rather boggled by the

I'm rather boggled by the queries here. Which part of "If you care about credit, this project is certainly not the right one for you." ( .... etc ) was unclear ? :-)

So hence quite explicitly the 'user contract' for Albert@Home is exactly that - don't ever assume credit. No surprise. As Bernd says we don't/won't have the capability/interest/manpower to account for the user credit concerns of those at Albert.

So if the question now reverts to "let's change the definition of Albert@Home" then E@H loses the primary purpose for Albert existing at all - a simple playpen for technical testing un-encumbered by the more complex milieu that applies to E@H.

What is or is not done elsewhere ( +/- Mr Anderson ) is a non-question, and argument by plurality is a non-sequitir also. We aren't elsewhere, we're here. :-)

So to be excruciatingly precise : you ought not be surprised/offended for the refusal to alter the machinations of this project !

This reminds me of a sign in a local shop "don't ask for credit as a refusal often offends" ..... :-)

Cheers, Mike.

"I have made this letter longer than usual, because I lack the time to make it short." - Blaise Pascal

TRuEQ & TuVaLu
TRuEQ & TuVaLu
Joined: 11 Sep 06
Posts: 75
Credit: 615315
RAC: 0

As I have ream in other

As I have ream in other projects forums...
The boinc system is based on credits so we crunch for credits.

So my sugestion is. Use a fixed credit system. And export stats if possible.

If you want just a few people helping out: a small credit for each task.
If you feel the need for more people, just increase the credit and put a notice in for instance in www.boincstats.com and more people will join.
And let people know what kind of hardware you are developing the application for.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.