Bible Pay

Read 417810 times

  • Rob Andrews
  • Administrator

    • 4097


    • 97
    • June 05, 2017, 08:09:04 PM
    • Patmos, Island Of
    more
So Ive been pondering this morning on what can I do to give PODC the highest credibility with the masses - in order to give PODC a run for its money against the heat mining big dogs, like bitcoin.  Some features that verifiably prove with 99% accuracy that the work being performed (in the cancer mining subsystem) is actually being performed in exchange for the BBP.

So far here is what I came up with:
- Each time a miner asks Rosetta for work, Rosetta awards the user with a Task (and that task contains multiple workunits)
- Each TaskID is assigned to a HostID (that is equivalent to one PC in your UserId)
- Each UserId has multiple HostIDs running multiple TaskIds which have multiple WorkUnits

However, a host always takes a certain amount of time to crunch a Task.

So if we choose random CPIDs in the Biblepay rosetta team, as reference machines first, then

If we had a table like this in the pool:
Id, CPID, UserId, TaskId, StartTime, EndTime, ValidatedByRosetta, Updated, Added, POL-Script-In, UTXOOut

Then we could write a report that runs in real time in the pool that shows at any given time what a given Users Hosts are working on IE 4 active tasks, average solution time per task, etc.

And if we start adding a POL script into the User, it adds integrity to the report, because it proves the user (is real and staking a certain number of coins per task-set).  This basically says a miner must stake X amount per day to prove they are cancer researching and the coins are attached to Y tasks.

The report would show live activity - so anyone saying How do I know this is really happening, if they were to monitor the activity of a random host, they would see the host finish actual Single work units, and ask for more. 

We could create a metric, for average solution time per task also, to prove that tasks take time across hundreds of researchers, therefore the system is not being falsified by a small group of hackers.  If some end up being solved quicker by a certain account we could then flag that account for potential fraud, but if they are not, we would know with relative certainty the systems integrity level.  Then we make a global metric called Integrity, and calculate it based on all these figures, and use that integrity level for PODC overall integrity.

Then we expose the reports to a new menu in the pool in the "accountability.biblepay.org" domain.

Would that help us?  Is it worth all the work?

It would take me about 50 hours to write all this for the pool.

Im thinking the pool will be less useful as a heat-mining pool when we transitiion to PODC, but, these reports could be added to the pool to increase the integrity of PODC for the World, globally, viewing this as "the new way" and the Light of cryptocurrencies as a whole.



« Last Edit: February 11, 2018, 10:15:43 AM by Rob A. »


  • Rob Andrews
  • Administrator

    • 4097


    • 97
    • June 05, 2017, 08:09:04 PM
    • Patmos, Island Of
    more
Rob: your response was for windows: i mean how set in linux that boinc manager/ only client

ouch:  we will be using genproclimit on our machines=1?  right?

xeon=24cores  ==  genproclimit=1    1core for BBP and 23cores for rosetta will be automaticly?

im trying understand how it working  linux vs windows

news interesting that i was hit superblock 7128?  5907 tBBP

ROB im using 2 machines for test 3dprotein=  HTPC+GAMER PC=  both 189 version: both i see on rosetta info:

when i getboincinfo on Ryzen= i see parameters,but when i getboincinfo on HTPC machine i see this




thanks for help guys

Two problems:

Your not running 1089 on that machine- got to run the latest software to test the features.
2) You should only run getboincinfo on the machine you associated the cpid with.


Then paste output.



  • Rob Andrews
  • Administrator

    • 4097


    • 97
    • June 05, 2017, 08:09:04 PM
    • Patmos, Island Of
    more
Some will say to my post, but wait, how do you actually trust rosetta with the solution validation?
We can construct an equation that addresses that:
1) The 'reference machine' must mirror the difficulty of the average solve time by more than one standard deviation (68% + )
2) No account may solve a task quicker per measured processor gigahertz than other users in the measured pool
3) Random results will be audited by someone at biblepay on a regular basis

In my opinion if these 3 things were performed, it would prove a certain level of reliability and integrity exists.



  • T-Mike
  • Sr. Member

    • 375


    • 2
    • February 06, 2018, 06:12:58 PM
    more
Some will say to my post, but wait, how do you actually trust rosetta with the solution validation?
We can construct an equation that addresses that:
1) The 'reference machine' must mirror the difficulty of the average solve time by more than one standard deviation (68% + )
2) No account may solve a task quicker per measured processor gigahertz than other users in the measured pool
How will you obtain the processor gigahertz? If you can obtain that from public data, I'm not sure it's correct at the moment as per the discussion I had with you regarding the cpu list.

3) Random results will be audited by someone at biblepay on a regular basis
I'm interested in doing the auditing.

In my opinion if these 3 things were performed, it would prove a certain level of reliability and integrity exists.


  • klondike
  • Full Member

    • 157


    • -10
    • October 10, 2017, 09:00:24 AM
    more
ROB

machine 1 = ryzen upgrade wallet from 187 to 189 via git pull master origin
machine 2 = htpc   newest installation wallet 189 from web

both of these machines is on same chain

on rosetta web i see machine 1+2 ...

getboincinfo must be working on both machine in debug window?

and next question:  we sending real power to rosetta now in testnet mode?

when i have next machine: i can to add to rosetta with same email+password via wallet?

thanks
« Last Edit: February 11, 2018, 12:11:34 PM by klondike »


  • Rob Andrews
  • Administrator

    • 4097


    • 97
    • June 05, 2017, 08:09:04 PM
    • Patmos, Island Of
    more
ROB

machine 1 = ryzen upgrade wallet from 187 to 189 via git pull master origin
machine 2 = htpc   newest installation wallet 189 from web

both of these machines is on same chain

on rosetta web i see machine 1+2 ...

getboincinfo must be working on both machine in debug window?

and next question:  we sending real power to rosetta now in testnet mode?

when i have next machine: i can to add to rosetta with same email+password via wallet?

thanks
exec getboincinfo should only work on the controller wallet.  Pick one controller wallet and then stick to it.

For machine # 2,3,4,5,6,7 you should be adding Boinc only, and when you add Rosetta to Boinc, you use the existing same email & pass you used for machine #1.  This way you have 7 machines connected to 1 CPID (like the instructions in the wiki say).

Btw, the boinc GUI (its called Boinc Manager in linux) is the same as (look & feel) of windows boinc manager.  I recommend using that in linux.


  • Rob Andrews
  • Administrator

    • 4097


    • 97
    • June 05, 2017, 08:09:04 PM
    • Patmos, Island Of
    more
Some will say to my post, but wait, how do you actually trust rosetta with the solution validation?
We can construct an equation that addresses that:
1) The 'reference machine' must mirror the difficulty of the average solve time by more than one standard deviation (68% + )
2) No account may solve a task quicker per measured processor gigahertz than other users in the measured pool
How will you obtain the processor gigahertz? If you can obtain that from public data, I'm not sure it's correct at the moment as per the discussion I had with you regarding the cpu list.

3) Random results will be audited by someone at biblepay on a regular basis
I'm interested in doing the auditing.

In my opinion if these 3 things were performed, it would prove a certain level of reliability and integrity exists.


Ive got a long history of boinc on other projects from back in the day and I assert that whatever we were looking at on that web report is a Rosetta Reporting problem (or report duration issue).  And thats pretty common, because the people that run the Server side webrpc software dont really care about it.

But I assert that the boinc side, the side that measures the CPU gigahertz and cobbles per core, is Very exact.  We just need to be pulling those figures:  Host information from Boinc, cobblestones per second from boinc, cores per motherboard from boinc, etc. 

Of course you have a point, that in a public project there is no way to trust that info.  But there are some things we can do.  We can ask the auditors - like you - to run a piece of biblepay software that gets it from the horses mouth and hashes it, and basically we say - this certifies the scales from the auditors point of view, and, I think we *can* trust boinc to a certain degree, because out of 100 hosts, its a true statement that the supermajority are most likely Not running hacked versions of boinc, and that compiled client is very complex, so on average, we can trust a cross section of boinc CPU info.

So I think we do have something to work with.  I would like to help to a certain degree, to where we put out enough software to raise the integrity level to above 90. 

Im going to start tonight on writing an SQL injection routine that allows the pool to report on certain things per host per work unit.

We can talk about designing something inside biblepay that helps certify the scales also.




  • klondike
  • Full Member

    • 157


    • -10
    • October 10, 2017, 09:00:24 AM
    more
exec getboincinfo should only work on the controller wallet.  Pick one controller wallet and then stick to it.

For machine # 2,3,4,5,6,7 you should be adding Boinc only, and when you add Rosetta to Boinc, you use the existing same email & pass you used for machine #1.  This way you have 7 machines connected to 1 CPID (like the instructions in the wiki say).

Btw, the boinc GUI (its called Boinc Manager in linux) is the same as (look & feel) of windows boinc manager.  I recommend using that in linux.

after update to 189 i did this on RYZEN 1700 UBUNTU 16.04:

a) install boinc

sudo apt-get install boinc-client

b) create Rosetta account with email, password, username

boinccmd --create_account https://boinc.bakerlab.org/rosetta/ EMAIL PWD USRNAME

c) Check that it worked==It will return your account key that you need to copy

boinccmd --lookup_account https://boinc.bakerlab.org/rosetta/ EMAIL PWD

d) attach machine to Rosetta account to begin working on solving computations

boinccmd --project_attach https://boinc.bakerlab.org/rosetta/ ACCOUNT_KEY

e) Now associate your BBP testnet wallet with Rosetta. Open wallet in testnet. It costs a couple of tBBP (since we are in testnet) to associate your BBP wallet to Rosetta.

exec associate rosetta_email_address rosetta_password


and on RYZEN this working excellent= exec getboincinfo i have all informations.....


and i reading this


Quote
4. So you do the steps above and it works in linux. Note that you only need 1 machine with BBP wallet (running in testnet), then boinc on all machines you want to add to Rosetta mining (to count towards daily superblock).

1st RYZEN is main Linux BBP wallet genproclimit= 15 (getboincinfo receiving in debug windows)

example
2,3,4,5 will be windows wallets ..... enough in DC bookmark in wallet add  email+password and i have to do exec assocation email+pass in debug win too? 

i have mess from all this setuping :( and dont  have taste last days with BBP something to do

we need helps  Rob,Togo,Znfall,virus,westwarmoth etc

« Last Edit: February 11, 2018, 01:54:23 PM by klondike »


  • Rob Andrews
  • Administrator

    • 4097


    • 97
    • June 05, 2017, 08:09:04 PM
    • Patmos, Island Of
    more
after update to 189 i did this on RYZEN 1700 UBUNTU 16.04:

a) install boinc

sudo apt-get install boinc-client

b) create Rosetta account with email, password, username

boinccmd --create_account https://boinc.bakerlab.org/rosetta/ EMAIL PWD USRNAME

c) Check that it worked==It will return your account key that you need to copy

boinccmd --lookup_account https://boinc.bakerlab.org/rosetta/ EMAIL PWD

d) attach machine to Rosetta account to begin working on solving computations

boinccmd --project_attach https://boinc.bakerlab.org/rosetta/ ACCOUNT_KEY

e) Now associate your BBP testnet wallet with Rosetta. Open wallet in testnet. It costs a couple of tBBP (since we are in testnet) to associate your BBP wallet to Rosetta.

exec associate rosetta_email_address rosetta_password


and on RYZEN this working excellent= exec getboincinfo i have all informations.....


and i reading this


1st RYZEN is main Linux BBP wallet genproclimit= 15 (getboincinfo receiving in debug windows)

example
2,3,4,5 will be win wallets ..... enough in DC bookmark in wallet add  email+password and i have to do exec assocation email+pass in debug win too? 

i have mess from all this setuping :( and dont  have taste last days with BBP something do

we need helps  Rob,Togo,Znfall,virus,westwarmoth etc



I just want to point out a few things so you dont fly blind:


1)  Log In To Web Rosetta from here:
https://boinc.bakerlab.org/rosetta/home.php

Use your Rosetta boinc credentials.

Search the page for "Cross-project ID:"  Write the value down in notepad.
That is your CPID and it will never change.


2)  You said you associated (using exec associate) a couple days ago already.  You only do this once per lifetime of researcher.
So that wallet was already associated.  If you did it again today, with Ryzen, then you are now re-associated with Ryzen, dont change it.
(Biblepay will only Honor the Last Burn Tx in the chain for you, so now that CPID in #1 is associated with the ryzen)


3)  For machines 2,3,4,5,6,7,8,9 etc, you only install bonc, and use the credentials in step #1.  This will cause Multiple Machines to crunch for rosetta on ONE cpid.  All rewards will go to the Ryzen.  You do Not want to run the 'exec associate' command again on any of those machines, or install bbp on them.


4) You can lower genproclimit =1  on the ryzen if you want,  as Im pushing low competition on the heat side in the future.

Thanks,
Rob



  • T-Mike
  • Sr. Member

    • 375


    • 2
    • February 06, 2018, 06:12:58 PM
    more

Ive got a long history of boinc on other projects from back in the day and I assert that whatever we were looking at on that web report is a Rosetta Reporting problem (or report duration issue).  And thats pretty common, because the people that run the Server side webrpc software dont really care about it.

But I assert that the boinc side, the side that measures the CPU gigahertz and cobbles per core, is Very exact.  We just need to be pulling those figures:  Host information from Boinc, cobblestones per second from boinc, cores per motherboard from boinc, etc. 

Of course you have a point, that in a public project there is no way to trust that info.  But there are some things we can do.  We can ask the auditors - like you - to run a piece of biblepay software that gets it from the horses mouth and hashes it, and basically we say - this certifies the scales from the auditors point of view, and, I think we *can* trust boinc to a certain degree, because out of 100 hosts, its a true statement that the supermajority are most likely Not running hacked versions of boinc, and that compiled client is very complex, so on average, we can trust a cross section of boinc CPU info.

So I think we do have something to work with.  I would like to help to a certain degree, to where we put out enough software to raise the integrity level to above 90. 

Im going to start tonight on writing an SQL injection routine that allows the pool to report on certain things per host per work unit.

We can talk about designing something inside biblepay that helps certify the scales also.

I'm having a hard time processing all this information to see the overall solution, I'll have to think about it through out the week. Let's keep working on this.


  • Rob Andrews
  • Administrator

    • 4097


    • 97
    • June 05, 2017, 08:09:04 PM
    • Patmos, Island Of
    more
I'm having a hard time processing all this information to see the overall solution, I'll have to think about it through out the week. Let's keep working on this.
Indeed, lets come up with a solution.  Because heres basically where we are:  The public trusts DC enough to value Golem, FAHCoin, Mediccoin and Gridcoin around 20 mil each.  Id like us to be a blessing to 10,000 orphans a month, so we need to be worth 200-300 million.  If we can come up with a solution together for this problem we could then hire 100 devs working around the clock and then be up in the top 10 (with the Billion $ coins) as this issue is seriously going to give the status quo, the heat miners a real option to jump ship and become a PODC miner.

One thing you can take a look at is all the logs that boinc emits (search the drive for client_state.xml).  You can scrape info out of the XML file that gives what boinc thinks about your machine.  Go ahead and see if you like what it says about you - cores - ghz - motherboard info etc.

Tonight Ill work on importing a researchers host record and tasks into the pool sql server.  Maybe we can start with a report that shows task durations per CPID.

But as we work on this problem, think of ways we can provably show a hosts speed and work activity per task per reward.

Casinos prove to us they stole peoples money with their RTG gaming reports, they show the dice are not loaded and prove the numbers were random
 and their mathematical edges obtained by actual measured outcomes.  Sounds to me we should be able to prove the PODC system was not fraudulently hacked over a certain number of measured outcomes.


  • klondike
  • Full Member

    • 157


    • -10
    • October 10, 2017, 09:00:24 AM
    more
HOLY HEAVEN
so will be enough 1 wallet on ryzen and all other my machines enough to install only boinc and add to group?   :o

« Last Edit: February 11, 2018, 02:13:25 PM by klondike »


  • Rob Andrews
  • Administrator

    • 4097


    • 97
    • June 05, 2017, 08:09:04 PM
    • Patmos, Island Of
    more
so will be enough 1 wallet on ryzen and all others my machines enough install only boinc and add to group?   :o

Yes, exactly.  You can have 1000 PCs mining against one cpid, and 10 tablets and 5 phones on the same cpid. You can still drill into the devices from the BoincStats and the Team info, so you lose no visibility on RAC per device. 

BBP calculates the Magnitude from the CPID total RAC per day.


  • znffal
  • Full Member

    • 129


    • 7
    • October 02, 2017, 04:01:47 PM
    more
HOLY HEAVEN
so will be enough 1 wallet on ryzen and all other my machines enough to install only boinc and add to group?   :o

Yes, 1 wallet on Ryzen.

On other machines just do

1) sudo apt-get install boinc-client

2) boinccmd --project_attach https://boinc.bakerlab.org/rosetta/ ACCOUNT_KEY


Then all machines are on same Rosetta account (same CPID) and you get all payments to 1 BBP wallet on your Ryzen


  • T-Mike
  • Sr. Member

    • 375


    • 2
    • February 06, 2018, 06:12:58 PM
    more
HOLY HEAVEN
so will be enough 1 wallet on ryzen and all other my machines enough to install only boinc and add to group?   :o

You can add many Rosetta accounts to the same wallet. On the bottom it will show all the CPIDs once you restart the GUI.