Maybe this has already been answered or discussed but I couldn’t find any thread about it.
What happens if I farm and fill up say 2 TB of disk space. Then I have earned a certain amount of safecoins. And then (hypothetically 
) I shut down the farm, delete all the data on the disk and register a new farm and start over with 2 TB free disk space again. Is that possible?
No, this is not how things work. If you connect to the network as a Vault and your drive get’s filled with 100GB. You didn’t earned anything. Only when someone requests a file (does a GET) and your Vault serves some chunks out of your Vault, a farming attempt is done. So you only get paid for delivering chunks out of your Vaults, not for filling your drive.
Ah, clever. Thanks, I’m still learning how it all fits together.
Also, to start you are less trusted. So while off/on vaults can still earn, you’ll do better the longer you can stay online.
So, if I understand it correctly, cheating is prevented by: In order to maximize my farming (given constant bandwidth) I need to have as many chunks stored as possible and keep my farming online all the time (temporary disconnects are ok since if my farm has a high ranking it will quickly regain it). The more chunks I have stored the more chance of my farm getting GET requests and thereby larger chance of earning safecoins.
The details are not finalised, so maybe, but the aim is to reward good farmers - in a way that maximises the number of small vaults using existing resources (amateurs & ordinary users) - and penalises behaviour that doesn’t help the network.
@Anders Have a read of this thread, while from May this year it does outline the change to non-persistent vaults being the preferred operation of the network for security reasons.
The only advantage now that a persistent vault has over a non-persistent vault is that it has more availability to serve chunks (earn) and doesn’t need refilling each period since it is never turned off. Also ranking has changed to account for the move from persistent to non-persistent vaults as the preferred vault operation.
To answer your question, its not cheating but the preferred model, and as @happybeing said the vault only earns when serving up requested chunks successfully and your rank now depends on successful GETs rather than time turned on
Ok, so the network will be more dynamic and be able to handle a lot of churn. That’s good, but I wonder how it will affect the farming reward. Will a farm with 5 TB of chunks stored be rewarded equally to a dynamic farm on a smartphone with only 100 MB of chunks? Surely more chunks stored should result in more farming reward, or else there is an incentive to provide as little disk space as possible.
Chunks are scattered around the network so in theory you buy tickets for a lottery. As in the lottery with more tickets you have more chances to win, but you don’t change the overall odds, so it’s not a guaranteed win. Technically a mobile with 100MB could farm the same or even more than a 5TB farm that was unlucky to get filled with spam.
Yes, I read somewhere that the farming reward is random but evenly distributed. So in practice since each farmer has so many chunks there should be a fairly even chance per chunk on average. And old unused chunks are sent to archive nodes I believe. The archive nodes will have to have a different farming reward since their chunks get much fewer GETs yet they store tons of data.
@Artiscience is right.
If two vaults with similar bandwidth turn on together then statistically it will not matter if one is 1GB and the other 100GB, they will store chunks at approx the same rate. The difference will be when the 1GB fills up.
Both will also (statistically on average) receive farming rewards at approx the same rate until the larger vault has more chunks stored.
Your example of mobile phone has some obvious differences in that the phone will fill after approx 100 chunks stored and the chance of GET rewards will be smaller than the 5GB that will have more chunks stored.
But as @Artiscience said certain cases could see the 100MB vault, with 90 chunks receive more GET requests on its contents than the 5GB vault with say 1000 chunks. But that is obviously highly unlikely
That’s what I thought. It’s the amount of chunks available that counts. So it’s only when the larger farm fills up that it will receive more GETs and larger chance of a reward. So the incentive is to keep the chunks stored since deleting them and starting over will reduce the farming reward. That will prevent cheating and make the farmers keep their chunks.
Old unused chunks should automatically be archived otherwise farmers may want to delete old data manually (that receives few GETs).
I think this would also be protected against - the closed group monitors the health of the vault periodically, including checking that it actually does have the chunks it’s supposed to have (this is proof of resource). This is done by asking all the vaults holding a particular chunk to do a calculation involving the chunk. The closed group can compare the answers, and make sure they agree. If not, the odd one out will be deemed to not have the chunk, and be penalised.
The rate would be the same throughout. In absolute numbers, they wouldn’t.
The difference in absolute farming earnings would simply reflect the difference in aggregate vault size on each respective client.