4 ms·

I assume the tech behind this is brittle in that Proof of Research "miners" are assumed to be trusted in that they don't just lie to the BOINC servers about wha

by nuclx 6y ago

I assume the tech behind this is brittle in that Proof of Research "miners" are assumed to be trusted in that they don't just lie to the BOINC servers about what work they actually performed.

roboticmind 6y ago
The BOINC projects run checks on the work a user submit. Usually it involves sending out the same work to two random different people and seeing if they match
- tphyahoo2 6y ago
  https://en.wikipedia.org/wiki/Sybil_attack https://en.wikipedia.org/wiki/Sybil_attack
  - roboticmind 6y ago
    If someone has the level of processing power to get a majority of a project's tasks, I think it would go very quickly noticed, and the project admin can just up their task validation higher making it much harder (e.g send to 4 random clients instead of 2). If they were to try to abort a bunch of the work, it would almost certainly get detected automatically by the project's spam protections and they would get less work sent to the point of no work if they do it too much. As well you'd have to pick only a single project which only accounts for 1/19th (currently) of Gridcoin research rewards. Seems easier to just do things legitimately if you have that level of power The main way you could really pull anything like that off is through a bugs or miss-configurations on where work is sent, which did happen once on a single project that messed up how work was sent, but was fixed rather quickly before they ever got any Gridcoin. As well, Gridcoin incentivizes people to look out for it since they are competing against any cheaters. If a project didn't respond to any attempts to do this, they can also be removed from the whitelisted projects on Gridcoin if needed For some comparison points about the level of power you'd need without any bugs, Rosetta@Home has 327,525 tasks currently running in place with an estimated ~370 terraflops of power[1]. Now Rosetta@Home is one of the more competitive projects due interest in their COVID-19 work, but overall there's still a lot of computing power in BOINC and from people earning Gridcoin for running BOINC. I believe there's more to the verification system, but my that's current understanding of it. [1] https://boinc.bakerlab.org/rosetta/server_status.php https://boinc.bakerlab.org/rosetta/server_status.php
    - ric2b 6y ago
      > If someone has the level of processing power to get a majority of a project's tasks But you don't need a lot of processing power to do a sibyl attack, if you're actually not running the tasks.
      - roboticmind 6y ago
        I also realized that I should have mention that I'm not sure you could always identify which tasks were the checks you were looking for For that scenario you mentioned, you'd not get any more tasks, as projects have limits on how many tasks they send out to people with tasks in progress. If you tried to spoof a bunch of hosts, they would see a very suspicious account that has a ton of tasks in progress but almost none completed. If you were spoofing users, I think the project would notice that half of their tasks are running way longer than normal all of a sudden or at least creeping that way if done slowly. Projects could make the tasks deadlines tighter and then make it harder for someone to try to hoard tasks. As well if you were running in virtual machines or something like that, they could tell that suddenly they have way more VMs "running" tasks and limit and even abort tasks just for those. If you try to spoof as a real machine, then you can look at IP addresses, etc. Having to do all that ends up taking more and more work try to do. If you have actual hardware running it, well that's fairy similar to having a large amount of processing power, and they could still limit the deadlines and all. You'd still have to compete against the other tens of thousands of users getting tasks on the project meaning that the project would have to see a massive increase in hosts and users on the project. As well, if you were trying to earn Gridcoin for all that and were spoofing users, you'd also have to be creating tens of thousands of beacons (tie BOINC accounts to Gridcoin rewards) which need for each user which cost 0.5 GRC each plus transaction fees, and those beacons take some time to verify which would give the projects a chance to stop it if they see it before you ever got any Gridcoin. You'd also have to stake on each account which would take a while and likely require a really really high balance to try to pull anything like that off and earn much Lastly some project have a literal either a very high number or some cases an infinite amount of tasks possible like some of the mathematical projects, so most of your tasks would not be the checks you'd be hoping to have
dimplebottom 6y ago
The economics of the coin address this. It's a fairly complex system. Lots of issues but solid progress due to some recent devs and contributors. Additionally contributions toward BOINC are not used to secure the ledger due to the security hole you point out but again look at the economics to learn more. Proof of Research is not a real thing.
- roboticmind 6y ago
  Ah yeah I should have mentioned that its PoS and that PoR is not really a thing anymore[1]. It's proof of stake that makes the actual blocks. The research rewards are just added on top of staking. Even all the project just suddenly shut down or something crazy like that, the network would keep on going just on Proof of Stake [1] It was around for some time technically alongside PoS, but this was removed over 4 years ago. Not sure how long it was a thing since I was just learning about Gridcoin when it was removed

by nuclx 6y ago

I assume the tech behind this is brittle in that Proof of Research "miners" are assumed to be trusted in that they don't just lie to the BOINC servers about what work they actually performed.

roboticmind 6y ago

The BOINC projects run checks on the work a user submit. Usually it involves sending out the same work to two random different people and seeing if they match

tphyahoo2 6y ago
https://en.wikipedia.org/wiki/Sybil_attack https://en.wikipedia.org/wiki/Sybil_attack
- roboticmind 6y ago
  If someone has the level of processing power to get a majority of a project's tasks, I think it would go very quickly noticed, and the project admin can just up their task validation higher making it much harder (e.g send to 4 random clients instead of 2). If they were to try to abort a bunch of the work, it would almost certainly get detected automatically by the project's spam protections and they would get less work sent to the point of no work if they do it too much. As well you'd have to pick only a single project which only accounts for 1/19th (currently) of Gridcoin research rewards. Seems easier to just do things legitimately if you have that level of power The main way you could really pull anything like that off is through a bugs or miss-configurations on where work is sent, which did happen once on a single project that messed up how work was sent, but was fixed rather quickly before they ever got any Gridcoin. As well, Gridcoin incentivizes people to look out for it since they are competing against any cheaters. If a project didn't respond to any attempts to do this, they can also be removed from the whitelisted projects on Gridcoin if needed For some comparison points about the level of power you'd need without any bugs, Rosetta@Home has 327,525 tasks currently running in place with an estimated ~370 terraflops of power[1]. Now Rosetta@Home is one of the more competitive projects due interest in their COVID-19 work, but overall there's still a lot of computing power in BOINC and from people earning Gridcoin for running BOINC. I believe there's more to the verification system, but my that's current understanding of it. [1] https://boinc.bakerlab.org/rosetta/server_status.php https://boinc.bakerlab.org/rosetta/server_status.php
  - ric2b 6y ago
    > If someone has the level of processing power to get a majority of a project's tasks But you don't need a lot of processing power to do a sibyl attack, if you're actually not running the tasks.
    - roboticmind 6y ago
      I also realized that I should have mention that I'm not sure you could always identify which tasks were the checks you were looking for For that scenario you mentioned, you'd not get any more tasks, as projects have limits on how many tasks they send out to people with tasks in progress. If you tried to spoof a bunch of hosts, they would see a very suspicious account that has a ton of tasks in progress but almost none completed. If you were spoofing users, I think the project would notice that half of their tasks are running way longer than normal all of a sudden or at least creeping that way if done slowly. Projects could make the tasks deadlines tighter and then make it harder for someone to try to hoard tasks. As well if you were running in virtual machines or something like that, they could tell that suddenly they have way more VMs "running" tasks and limit and even abort tasks just for those. If you try to spoof as a real machine, then you can look at IP addresses, etc. Having to do all that ends up taking more and more work try to do. If you have actual hardware running it, well that's fairy similar to having a large amount of processing power, and they could still limit the deadlines and all. You'd still have to compete against the other tens of thousands of users getting tasks on the project meaning that the project would have to see a massive increase in hosts and users on the project. As well, if you were trying to earn Gridcoin for all that and were spoofing users, you'd also have to be creating tens of thousands of beacons (tie BOINC accounts to Gridcoin rewards) which need for each user which cost 0.5 GRC each plus transaction fees, and those beacons take some time to verify which would give the projects a chance to stop it if they see it before you ever got any Gridcoin. You'd also have to stake on each account which would take a while and likely require a really really high balance to try to pull anything like that off and earn much Lastly some project have a literal either a very high number or some cases an infinite amount of tasks possible like some of the mathematical projects, so most of your tasks would not be the checks you'd be hoping to have

dimplebottom 6y ago

The economics of the coin address this. It's a fairly complex system. Lots of issues but solid progress due to some recent devs and contributors. Additionally contributions toward BOINC are not used to secure the ledger due to the security hole you point out but again look at the economics to learn more. Proof of Research is not a real thing.

roboticmind 6y ago
Ah yeah I should have mentioned that its PoS and that PoR is not really a thing anymore[1]. It's proof of stake that makes the actual blocks. The research rewards are just added on top of staking. Even all the project just suddenly shut down or something crazy like that, the network would keep on going just on Proof of Stake [1] It was around for some time technically alongside PoS, but this was removed over 4 years ago. Not sure how long it was a thing since I was just learning about Gridcoin when it was removed