[Discussion] 4th Generation BOINC credit system

Message boards : Questions and problems : [Discussion] 4th Generation BOINC credit system
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · Next

AuthorMessage
SekeRob
Volunteer tester
Help desk expert

Send message
Joined: 25 Aug 06
Posts: 1545
Message 69226 - Posted: 28 Apr 2016, 8:54:09 UTC - in response to Message 69224.  

"This also means that projects need to supply an accurate estimate of how much FLOPs are needed to compute a workunit."

By and large impossible for most that use a non-deterministic algorithm. Credit_New (Gen 3) is wholly incapable to deal with variable runtimes within a single app... it's a permanent stream of irritation posts, when the momentary batch suddenly runs much faster than the previous, or much slower, let alone if a single batch contains molecules of varying complexity against which the simulations are run. The benchmark is rig-able, in fact a useless base with CPU's that substantially can change speed... run benchmark at max turbo, set skip benchmark in the config and lower the multiplier and you already have a device that is not doing the amount of work it says it is doing... Something needs to be devised that can measure true work against which a credit can be set per unit of work... something inside the science app API maybe that facilitates the recording of such. [WCG at one time prefixed a benchmark test front of all jobs that lasted a minute or so for example]
Coelum Non Animum Mutant, Qui Trans Mare Currunt
ID: 69226 · Report as offensive
SekeRob
Volunteer tester
Help desk expert

Send message
Joined: 25 Aug 06
Posts: 1545
Message 69230 - Posted: 28 Apr 2016, 11:10:06 UTC - in response to Message 69226.  

Not publishing host detail is default. More so, there's a config option to also suppress network detail be transmitted such as host name.
Coelum Non Animum Mutant, Qui Trans Mare Currunt
ID: 69230 · Report as offensive
Profile pschoefer

Send message
Joined: 5 Aug 06
Posts: 48
Germany
Message 69238 - Posted: 28 Apr 2016, 13:19:48 UTC - in response to Message 69155.  

Any publicity is good publicity. If a BOINC cheating expose article hit the front page of reddit, there would likely be an influx of new users joining the BOINC platform. Our community hit the front page of reddit for a couple hours and we managed to recruit approx 1000 users in a single day to our team.

Well, I know that kind of short-term influx of new users. It's not very sustainable and doesn't work forever. But that's another point for a 'clash of cultures' type of thread.

There's no point worrying about BOINC projects shutting down - more projects will likely be created in the future to fill their place, and the benefits of free computing power from volunteers likely exceeds inconveniences caused by cheaters especially as we iron out how to make cheating less effective in this thread.

We'll see. There are less active projects nowadays than there used to be a couple of years ago and less new projects are established. There's no such thing as infinite growth.

Gridcoin has not imposed new problems/issues onto the BOINC platform - the problems rising to the surface have existed for years now (with 0 discussion by the BOINC community nevermind the Gridcoin community). If it had not been Gridcoin, then CharityEngine or inter-team competitions would have exposed the same cheaters.

There is one major difference. Yes, the cheaters that claimed high rankings in the long-term statistics had their 15 minutes of fame. Punch in the faces of all honest crunchers who care about the stats. But most cases, their credit was zeroed some time after the cheating was revealed, so that fame is not sustainable. In short-term competitions, the cheaters lose their credibility; no one will take them serious in future competitions, if they are allowed to take part at all. Just like with doping in sports. On the contrary, according to one of your earlier posts, those who cheated in order to get more Gridcoins do retain their benefit (the same might apply for that CharityEngine thing, I don't know anything about that).

---

Just found this thread on cosmology@home that's relevant to this topic: http://www.cosmologyathome.org/forum_thread.php?id=7341#20601

As implied by an ATLAS@Home admin, the LHC-related projects may also have a new credit system soon, which is another step toward 'project-defined credit'.
ID: 69238 · Report as offensive
Profile Agentb
Help desk expert
Avatar

Send message
Joined: 30 May 15
Posts: 265
United Kingdom
Message 69386 - Posted: 2 May 2016, 22:23:51 UTC - in response to Message 69226.  

ChristianB wrote: What I want to bring up again is what should Credits represent?

This is a very important question.

Firstly even the word "credit" especially when combined with a numerical value has an implicit monetary meaning. Terms such as credit, debit and debt imply a currency, and then all the problems of "money" follow (fraud, counterfieting etc). (Greshams's Law of bad money overtaking good money being one)

If asked, currently the current answer is "nothing much really. Especially not between projects". Some may say - that is a good thing, and long may it remain so!

If it is to measure "something" there is a cost of standardization (think of the cost to run the stock exchanges, exchange rates, standard measures and their enforcement).

Taking this question a step further.

"What one thing alone should a credit measure?"

For example
"what each project admin feels best at the time for their project"
"the average run times of 50 different hosts"
"the energy in Joules required to complete the task on <some measuring system>"
"0.0001 USD worth of processing at Amazon S2 on <some standard system>"

It can not be a measure of "all things" in fact it can't be a measure of two things.

Another related question is, "Does attempts to standardize cause more problems in that it become more like money, and so cheating become more commonplace?"

SekeRob wrote: By and large impossible for most that use a non-deterministic algorithm. Credit_New (Gen 3) is wholly incapable to deal with variable runtimes within a single app...

For applications which have non-deterministic algorithms, this can be really difficult, but I can forsee something like the result itself recording some application specific chain of counters (like checkpoints) which is included as part of the result. This could be validated at the server end, in the usual quorum manner, or other methods - as being a valid chain, and an adjustment made.
ID: 69386 · Report as offensive
mikey
Avatar

Send message
Joined: 6 May 16
Posts: 2
United States
Message 69486 - Posted: 6 May 2016, 11:07:25 UTC - in response to Message 69224.  


P.S.: I don't know anything about the specific reasons why people hide their hosts but they should have the possibility.

Regards
Christian


My suggestion would be to allow Team Founders to see this data anyway, keep the average person from seeing it, but if you join a Team then the Team Founder CAN see it, after all it's THEIR Team that's being affected by the cheating. People used to hop from team to team when team total credit went with them, when that stopped some of the team hopping stopped, but some people still hop from team to team for various reasons. If someone is cheating and viewing their pc's can help stop that and allowing Team Founders to see all Team Members pc's can stop it, then I'm all for it. But as I said if you have 50 pc's because you are running a computer lab or whatever and wish to keep them hidden from the general user, I'm okay with that too.
ID: 69486 · Report as offensive
nanoprobe

Send message
Joined: 9 Apr 12
Posts: 51
United States
Message 69494 - Posted: 6 May 2016, 18:39:12 UTC - in response to Message 69224.  


P.S.: I don't know anything about the specific reasons why people hide their hosts but they should have the possibility.


Regards
Christian

100% agree. I have very specific reasons that are important to me as to why I choose that option. One is that it helps cut down on the shenanigans that go on during challenges. I'm sure many of you know about the crap that happens during the annual large challenge with projects that require a "wingman." JMHO
ID: 69494 · Report as offensive
Steve Hawker*

Send message
Joined: 16 Dec 12
Posts: 14
United States
Message 69497 - Posted: 7 May 2016, 0:31:39 UTC

I truly do not care what BOINC credit measures. It matters not one jot.

What does matter and this is where CreditNew fails spectacularly is that two essentially identical WUs receive essentially identical credit and this applies to yesterday's WUs, today's WUs and tomorrow's WUs. All too often we see credits falling or plummeting while the work done remains constant. It's no wonder it was renamed CreditRandom.

Credits are supposed to recognize contributions and frankly WCG's simplistic core-hours beats CreditNew hands down. 1 core-hour is always 1 core-hour. However this system rewards clunkers and hyper-crunchers alike.

I'm all for letting the projects set a fixed value per WU. Highly similar WUs will get the same reward and faster machines will get higher RACs. Critics may complain that this destroys cross-project comparisons but when has that ever been realistic. Let the admins do what they feel is right for their projects. On projects that do this, users self-regulate anyway and it's common to see comments supporting a "fair" rate. Fears of runaway credit inflation are groundless.

If BOINC had never tried to be all things to all small furry creatures from Alpha Centauri, then the FLOPS vs ASIC BitcoinUtopia fiasco would never have occurred. BU could just say x credits per share and have done with it. Instead we get into a "my machine does more FLOPS so give me 2 gazillion points" style pissing contest.

Fixed rate, let the market decide.

I am utterly uninterested in any possible side effect this has for GridCoin. As GridCoin is a profit-seeking venture, the BOINC credit system has no responsibility, ethically or morally, to change its system(s) to suit the needs of GridCoin.

If calling it "credits" has financial implications, change it to "points".

If this all means a new release of BOINC server s/w, can you somehow make it a mandatory core function to provide a forum on every project. Thanks!
ID: 69497 · Report as offensive
Joe

Send message
Joined: 23 Dec 14
Posts: 2
United States
Message 69498 - Posted: 7 May 2016, 2:20:10 UTC - in response to Message 68956.  
Last modified: 7 May 2016, 2:21:16 UTC

Seems to me there is a need to start with basics.

If a particular Boinc project has a financial reward then it should develop whatever payout system that works for that particular project.

Expecting projects that were never conceived as for profit to consider adopting another credit system to accommodate for profit projects shows quite a bit of arrogance in my opinion.
ID: 69498 · Report as offensive
Profile Agentb
Help desk expert
Avatar

Send message
Joined: 30 May 15
Posts: 265
United Kingdom
Message 69499 - Posted: 7 May 2016, 8:36:41 UTC - in response to Message 69497.  

Steve Hawker* wrote:
Credits are supposed to recognize contributions and frankly WCG's simplistic core-hours beats CreditNew hands down. 1 core-hour is always 1 core-hour.

Are referring here to the "Total Run Time" statistic published on WCG?

I'm not sure i understand how WCG point system is "better" or "different" - I think they just using the number 7 to convert credits to points - see WCG FAQ or am i missing something?
ID: 69499 · Report as offensive
Profile Ageless
Volunteer moderator
Project administrator
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 12091
Netherlands
Message 69501 - Posted: 7 May 2016, 10:18:38 UTC - in response to Message 69224.  
Last modified: 7 May 2016, 10:19:37 UTC

Christian Beer wrote:
P.S.: I don't know anything about the specific reasons why people hide their hosts but they should have the possibility.

The original reason for this option was for hardware testers who due to their NDA were not allowed to show the hardware in combination with their account. Think about new GPUs being tested, or new CPUs. They'd still show up in the lists, but not against a specific person's account.

As to why people do it these days, probably ignorance as they don't believe other users cannot see their intimate information, or just because they can for a bit of privacy.
Jord
Only partially available.
Please do not private message me for tech support. Use the forums for that. Tech PMs will be ignored.
ID: 69501 · Report as offensive
Steve Hawker*

Send message
Joined: 16 Dec 12
Posts: 14
United States
Message 69502 - Posted: 7 May 2016, 10:23:52 UTC - in response to Message 69499.  

Steve Hawker* wrote:
Credits are supposed to recognize contributions and frankly WCG's simplistic core-hours beats CreditNew hands down. 1 core-hour is always 1 core-hour.

Are referring here to the "Total Run Time" statistic published on WCG?

I'm not sure i understand how WCG point system is "better" or "different" - I think they just using the number 7 to convert credits to points - see WCG FAQ or am i missing something?


My point is that on WCG, crunching for one hour gets you one hour added to your total. This was true 10 years ago, is true now, will be true in 2026. Hours are the currency at WCG not BOINC credits

WCG's conversion from their points to BOINC credits is 7:1 and I don't have a problem with that because it's ALWAYS 7:1.

CreditNew does not pay the same credit for the same WU last month as it does today, nor next month.

It is that lack of consistency from using CreditNew that damages a project's credibility. It is WCG's consistency that makes it better than CreditNew.

Did I explain it better this time?
ID: 69502 · Report as offensive
Profile Agentb
Help desk expert
Avatar

Send message
Joined: 30 May 15
Posts: 265
United Kingdom
Message 69503 - Posted: 7 May 2016, 14:09:11 UTC - in response to Message 69502.  

Steve Hawker* wrote:
Credits are supposed to recognize contributions and frankly WCG's simplistic core-hours beats CreditNew hands down. 1 core-hour is always 1 core-hour.

Are referring here to the "Total Run Time" statistic published on WCG?

I think you are saying "Yes" to this question. If so, then the slower your system is, the more core-hours per task.

This metric is not a good measure of contribution - if that is what "Credit" is supposed to be.

If a host can complete two tasks in one day, should its "reward" be the same as a host which can only do one task in a day?

This metric is good as it is simply defined, but it does mean the most you get is 24 hours per day, irrespective of tasks completed. This will encourage running silly VMs for example.

As an aside, if anyone wants to calculate their host's core-hours - all the data is the job_log_<projectID>.txt file in the boinc data directory. This is very useful when comparing apps for example. (which app should i run, is it more efficient after an upgrade etc)

I'm not sure i understand how WCG point system is "better" or "different" - I think they just using the number 7 to convert credits to points - see WCG FAQ or am i missing something?


My point is that on WCG, crunching for one hour gets you one hour added to your total.

This was true 10 years ago, is true now, will be true in 2026. Hours are the currency at WCG not BOINC credits

WCG's conversion from their points to BOINC credits is 7:1 and I don't have a problem with that because it's ALWAYS 7:1.

CreditNew does not pay the same credit for the same WU last month as it does today, nor next month.

It is that lack of consistency from using CreditNew that damages a project's credibility. It is WCG's consistency that makes it better than CreditNew.

Did I explain it better this time?


Not quite there yet, i'm still not seeing something, why is "WCG Points" system better than BOINC NewCredit system? Multiplying NewCredit by 7 doesn't not make it better or worse.

What i notice is different on WCG forums (i am not a regular reader or visitor) is no display of RAC, Points nor Totalcredit - only a) number of forum posts and b) badges. These badges seem to be given based on core-hours.

I personally don't like badges, and especially when being generated, downloaded and displayed repeatedly in every post - but that's just me and some folks do like them.

I do like to see the "number of posts", as that does often show the personal level of contribution.

Maybe we just need to ignore badges and credits, and worry about more important stuff!
ID: 69503 · Report as offensive
Profile CM

Send message
Joined: 13 Aug 15
Posts: 63
Message 69504 - Posted: 7 May 2016, 14:38:34 UTC - in response to Message 69486.  
Last modified: 7 May 2016, 15:06:21 UTC

Quoted from mikey (message 69486):

P.S.: I don't know anything about the specific reasons why people hide their hosts but they should have the possibility.

Regards
Christian


My suggestion would be to allow Team Founders to see this data anyway, keep the average person from seeing it, but if you join a Team then the Team Founder CAN see it, after all it's THEIR Team that's being affected by the cheating. People used to hop from team to team when team total credit went with them, when that stopped some of the team hopping stopped, but some people still hop from team to team for various reasons. If someone is cheating and viewing their pc's can help stop that and allowing Team Founders to see all Team Members pc's can stop it, then I'm all for it. But as I said if you have 50 pc's because you are running a computer lab or whatever and wish to keep them hidden from the general user, I'm okay with that too.

It's not just an individual team that is affected by cheating though, it affects the BOINC project (cheaters returning invalid wu results & 'wingmanning' the same invalid results, or skewing performance statistics) and it affects all other BOINC users (team competitions are fun, but not if users/teams cheat their way to #1 place & individual leaderboards are difficult to trust if cheating is widespread).

Even if host details were available for team founders to go over, it's quite difficult to determine if cheating is actually occuring.. cpu count mismatches are easy to spot, but benchmarks could have just glitched or be high due to overclocking..

Another issue with this is that a cheater could make their own team, or the cheater could be the team founder - both scenarios would prevent host details from being inspected.

Perhaps a privacy option could strip the cpid from public records (host.gz file), but expose the host details? Users could report to admins suspicious host details.

Quoted from nanoprobe (message 69494):

P.S.: I don't know anything about the specific reasons why people hide their hosts but they should have the possibility.


Regards
Christian

100% agree. I have very specific reasons that are important to me as to why I choose that option. One is that it helps cut down on the shenanigans that go on during challenges. I'm sure many of you know about the crap that happens during the annual large challenge with projects that require a "wingman." JMHO


What do you think about obfuscating work unit identification? Only the project admins would know which wu's match, preventing the wingman cheating scenario?

Quoted from Joe (message 69498):
Seems to me there is a need to start with basics.

If a particular Boinc project has a financial reward then it should develop whatever payout system that works for that particular project.

Expecting projects that were never conceived as for profit to consider adopting another credit system to accommodate for profit projects shows quite a bit of arrogance in my opinion.

Current non-profit based BOINC projects are highly unlikely to begin rewarding their volunteer researchers (maybe GPUGRID/SETI/LHC/WCG since they're backed by big companies), since the major benefit of BOINC is that the volunteer computing resources are free.

If a BOINC project did decide to start rewarding their users, they'd run into the issues being raised in this topic.

Gridcoin isn't a BOINC project, it's an external system that distributes rewards on behalf of BOINC projects. Fair enough BOINC projects were never initially intended to reward users anything other than credit/achievements/badges, but that doesn't mean we shouldn't do it (Ripple, Charity Engine and inner/inter-team competitions expose the same issues).

The GRC (DPOR) reward mechanism isn't in scope here (it accurately rewards BOINC RAC), what's in scope here is a theoretical next-gen BOINC credit system that can overcome issues currently affecting the 3rd gen system.

I think it would have been far more arrogant of me to have just begun campaigning for immediate removal of affected projects from our whitelist instead of initiating this topic in the first place. Remaining ignorant to the shortcomings of the current system is not a realistic option.

Quoted from Steve Hawker* (Message 69497)

I am utterly uninterested in any possible side effect this has for GridCoin. As GridCoin is a profit-seeking venture, the BOINC credit system has no responsibility, ethically or morally, to change its system(s) to suit the needs of GridCoin.

What you suggest wouldn't have any negative side effects for Gridcoin - users are already rewarded on an individual project level.

You're right, the BOINC credit system doesn't have any responsibility to change its system(s) to suit the needs of Gridcoin; it does however have a responsibility to continue improving/developing its system(s) to address the needs of BOINC users/projects as a whole.
ID: 69504 · Report as offensive
Steve Hawker*

Send message
Joined: 16 Dec 12
Posts: 14
United States
Message 69508 - Posted: 7 May 2016, 17:03:23 UTC - in response to Message 69503.  

Not quite there yet, i'm still not seeing something, why is "WCG Points" system better than BOINC NewCredit system? Multiplying NewCredit by 7 doesn't not make it better or worse.

What i notice is different on WCG forums (i am not a regular reader or visitor) is no display of RAC, Points nor Totalcredit - only a) number of forum posts and b) badges. These badges seem to be given based on core-hours.

Please ignore the WCG points, they are irrelevant to my point.

I am comparing two systems: WCG's hours and CreditNew's credits. My complaint with CreditNew is that it is inconsistent and I think that is well established and recognised. As I said, and you just alluded to, the WCG currency is hours. So I'm saying that WCG system is 100% consistent in that hours crunched are always the hours awarded, whereas you are unable to predict how much credit you'll be getting with CreditNew. Imagine if your employer chose CreditNew as your wage system...

Your concern with comparing slow machines with fast machines is valid if your focus is on earning credits and not hours. I have already said that this is a problem but only if you're focused on earning credits.

Converting WCG points to BOINC credits is a red herring. It exists purely so that crunchers can include WCG in the BOINC system. I don't care if it's 7:1 or 70:1. What I care about is that it was 7:1 yesterday, 7:1 today and will be 7:1 tomorrow. If the conversion used a CreditNew calculation, you would have no idea what you'd be getting in the future, based on what you're getting now.

I'm not concerned at all with forunms or badges at the moment.

Maybe we just need to ignore badges and credits, and worry about more important stuff!

Well you can reduce any endeavour in that way. But while we're crunching, a LOT of people want the credit system. Many are concerned with pure credit totals, some like to achieve milestones, some like to collect badges while others like to accrue hours. Whatever your concern, it all makes crunching more interesting and keeps people involved.
ID: 69508 · Report as offensive
noderaser
Avatar

Send message
Joined: 2 Jan 14
Posts: 259
United States
Message 69512 - Posted: 8 May 2016, 5:07:34 UTC - in response to Message 69501.  

Steve Hawker* wrote:
My point is that on WCG, crunching for one hour gets you one hour added to your total. This was true 10 years ago, is true now, will be true in 2026. Hours are the currency at WCG not BOINC credits

However, newer machines with multiple cores will generate more computing hours per clock hour. Computing hours are how WCG's badges are awarded, but credit is still a factor, especially when comparing it to other BOINC projects. While there are a few who argue that everyone's contribution should be rewarded the same, it seems the majority of those who have voiced their opinion would rather see higher credit numbers be awarded to those who get more work done for the project.

Ageless wrote:
The original reason for this option was for hardware testers who due to their NDA were not allowed to show the hardware in combination with their account. Think about new GPUs being tested, or new CPUs. They'd still show up in the lists, but not against a specific person's account.

This can still be a legitimate reason as I know at least one person who works at Intel and is using "company" hardware under development for BOINC, although they still have their hosts visible to the public.

ID: 69512 · Report as offensive
Ju Li Ling

Send message
Joined: 8 May 16
Posts: 3
Message 69513 - Posted: 8 May 2016, 8:43:56 UTC

I'm sorry. I have to laugh at this discussion. Why volunteer our time crunching projects with the Boinc software when one can volunteer and get paid too! LOL I see two choices here. Fix the credit system so all projects pay the same and extend the reward system of getting paid to all teams......or.....ban Gridcoin from Boinc teams. I heard the taxman is curious about this and is looking into Gridcoin paying people and also looking into pulling the non-profit status from Boinc in general. Hope you have fun with the red tape thanks to a team like Gridcoin.
ID: 69513 · Report as offensive
Profile CM

Send message
Joined: 13 Aug 15
Posts: 63
Message 69516 - Posted: 8 May 2016, 12:27:37 UTC - in response to Message 69513.  
Last modified: 8 May 2016, 12:30:02 UTC

I'm sorry. I have to laugh at this discussion. Why volunteer our time crunching projects with the Boinc software when one can volunteer and get paid too! LOL I see two choices here. Fix the credit system so all projects pay the same and extend the reward system of getting paid to all teams......or.....ban Gridcoin from Boinc teams. I heard the taxman is curious about this and is looking into Gridcoin paying people and also looking into pulling the non-profit status from Boinc in general. Hope you have fun with the red tape thanks to a team like Gridcoin.

I don't think anyone is proposing that all projects should reward equal quantities of credit; inter-project credit comparison is a flawed comparison. Inner-project credit comparison is what's important.

Ban Gridcoin? Going to ban charityengine and any sponsored BOINC competitions too? How would you enforce this across 30+ projects?

Got a source on your tax claim? Generally it is the responsibility of end-users to pay tax on crypto earnings according to the tax laws implemented in their geographical area - tax mostly doesn't apply until you spend it (VAT/CGT) or when you convert it from crypto to fiat.

Does BOINC operate under a non-profit company? I don't see any FAQ stating this, only the fact that the software is distributed with the Lesser General Public License (LGPL)? Besides, it's not BOINC distributing earnings it's a 3rd party. Though if you've got proof of this please do share.

Are you equally opposed to commercial BOINC projects making a profit from volunteer computation? Bitcoin Utopia is a commercial project, are you proposing to ban BU too?
ID: 69516 · Report as offensive
Profile Ageless
Volunteer moderator
Project administrator
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 12091
Netherlands
Message 69517 - Posted: 8 May 2016, 15:11:40 UTC - in response to Message 69513.  

Fix the credit system so all projects pay the same...

With BOINC being open source and anyone being able to change things in it as he or she likes, there's no way that any credit system can be enforced on all projects. As long as any of them thinks it can pay several million credit for 3 minutes worth of work, it's able to do so.

It's up to the users of such a project to tell them if they like that or not.
Jord
Only partially available.
Please do not private message me for tech support. Use the forums for that. Tech PMs will be ignored.
ID: 69517 · Report as offensive
Ju Li Ling

Send message
Joined: 8 May 16
Posts: 3
Message 69518 - Posted: 8 May 2016, 15:38:29 UTC - in response to Message 69516.  

As things were explained to me, the IRS really isn't interested in Boinc itself or the projects. They want to know if taxes are being paid. They want IP addys. If someone is using company equipment to turn a personal profit by earning a taxable currency, they really don't care. What they want to know is if the companies are bothering to claim it on their businesses taxes. If not, they are going to get a hefty fine. Chances are, someone's gonna get fired if they are lucky. If not, they are looking at jail time for fraud, theft, and tax evasion. They have a copy of gridcoin's white list and will be digging up everyone's IP who is connected with the white list. The projects themselves can expect a visit from the taxman just to get the info they seek. As for me, I could care less if someone makes a profit or not. I just expect a cash reward for turning this in. Teams like gridcoin shouldn't exist. They ruin the fun for everyone.
ID: 69518 · Report as offensive
LennStar

Send message
Joined: 8 May 16
Posts: 7
Germany
Message 69521 - Posted: 8 May 2016, 22:31:55 UTC - in response to Message 69518.  

As things were explained to me, the IRS really isn't interested in Boinc itself or the projects. They want to know if taxes are being paid. They want IP addys. If someone is using company equipment to turn a personal profit by earning a taxable currency, they really don't care. What they want to know is if the companies are bothering to claim it on their businesses taxes. If not, they are going to get a hefty fine. Chances are, someone's gonna get fired if they are lucky. If not, they are looking at jail time for fraud, theft, and tax evasion. They have a copy of gridcoin's white list and will be digging up everyone's IP who is connected with the white list. The projects themselves can expect a visit from the taxman just to get the info they seek. As for me, I could care less if someone makes a profit or not. I just expect a cash reward for turning this in. Teams like gridcoin shouldn't exist. They ruin the fun for everyone.

1) Why should anyone from taxes ask anything from the projects? They have nothing to do with Gridcoin.
Gridcoin pulls it data from the projects - like e.g. Boincstats. Everyone else can do that too - faster then asking something.
And because Gridcoin (network) is pulling, the projects also dont know the IPs, just to mention it. Also IP does not mean you know who it is.

2) I dont think US (you are referring to them, right?) Tax authorities have any say in projects not in the US, which are btw. the majority.

3) Also I cant see how Gridcoin can "ruin the fun". Your BOINC does not work in any other way because Gridcoin exist. If you think Team Gridcoin should not be in any point competition - ok, maybe. Just ignore that team then. That is up to the one hosting the competition.
The only reason for your behavior here I can see is your fear of losing teammembers.
That is understandable but - here we are at topic - current BOINC system does not safely allow to have Gridcoin run without team requirement (and it would produce too much stress on the project servers if Gridcoin network would ask all data from every user on every project).
If that is changed the team requirement would surely be lifted.

Gridcoin has brought many people to BOINC and long-time BOINCers (like me) have increased their output in the hope of getting at least part of the energy costs paid (or in the future even make a profit, who knows?).

If you are interested in advancing BOINC, please provide useful feedback, if you just want to steam please GTFO.


-------------

From time to time, I noticed a sudden jump in credit that in reality does not seem justifiable from a user that had a static average credit. ...
I would suggest that the current option to hide your host be removed

Actually that very easy. And easy to explain. I too have 10-fold difference in the last year.
That suggestion seems to be good, it is not really a privacy matter but can help identify cheaters.
People who mentioned problems please briefly explain it.


I do not like the notion of measuring HDD usage, RAM etc.
1) it adds complexity
2) can all be gamed
3) is not really important as long as you dont use BOINC as a peer-to-peer storage system. (Whihc would be a nice form for a different (commercial) Gridcoin use, but not for science itself)
4) is very hard to predict on mid-term range and long-term. Do you want to give more credits for expensive and fast SSDs then conventional HHDs for example? What about CPUs? With optimisation I could get my APU computer down from 112W to less then 80W (or roughly 25%) - for just 12% less speed.


What is important is the work done. WCGs measurement in core hours just means it gives you more badges for running 2 Rasperrys then one intel i5 for doing 1/10 of the work.
So time alone is not a good measurement.
The only possibility I can see is a mandatory benchmark (can be as short as 10 seconds) every hour or so as a basis for calculations.

-.--
btw: Is there a list which projects run more on FPU and which more on integer, since someone mentioned that? I have been wondering for years now - since I got an AMD module APU.
ID: 69521 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · Next

Message boards : Questions and problems : [Discussion] 4th Generation BOINC credit system

Copyright © 2017 University of California. Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.