Yes, the constant searching for hot data by way of churn events was something that came to mind as a potential side effect, but the disincentive to churn due to loss of farming rewards would seem adequate for preventing this. Another concern might be from farmers getting “bored” since their vaults are full but no longer generating revenue because the data is not being accessed enough, so they would just get rid of them and leave the network permanently/indefinitely. As others have pointed out, it would seem that transaction fees (for data, not safecoin) could keep them interested. Thinking out loud for a moment… is it possible to split the total farming reward between the farmer’s vault and all other nodes that pass a data chunk to the requesting client, or is this essentially how it’s done now? It would seem that depending on the storage vs. transaction reward ratio it would help keep the vault managers/farmers motivated.
EDIT: Searching the forum I found some great discussion that was related to this topic in :
Their loss should be someone else’s gain. It will potentially create demand for more storage space, which could increase the farming reward. Obviously, we are talking small impacts by individuals, but the network as a whole will react in such a way.
“Another concern might be from farmers getting “bored” since their vaults are full but no longer generating revenue because the data is not being accessed enough, so they would just get rid of them and leave the network permanently/indefinitely.”
A worse scenario is the farmer constantly seeking a more active vault by rebooting his device after a prolonged period of inactivity in an effort to improve his returns. I know there will be a price to pay for this strategy (starting all over as an “infant”) but if his full vault is producing nothing, or close to nothing, as far as farming rewards go, he would not be losing anything by starting over. If enough farmers thought this way it could create an abnormally large volume of churn which probably would not be a good thing for the network.
What is needed to mitigate the waste of a full, non-performing vault is a “pass-through” fee. If a “fetch” goes through a node on the way to the desired data, that node (farmer) will share in the rewards with the destination farmer. This would be like an automatic toll booth along the highway. Not sure of the technical challenges of such a system but it sure makes the network more equitable for farmers.
This is not really feasible but is possible (like getting hit by lightening 100 times is possible). Data is distributed evenly across the address range so you will have parts of popular/unpopular data all in the same ratio as any other vault. At least that should be the case, it is all probabilities, but this probability is extremely high, otherwise the secure part would have to be removed from the hashing algorithm (so no SHA just HA )
Thanks for clarifying. Remarkable how often the Theory of Probability enters the picture. Btw, imho ToP is, over your lifetime, the most useful and rewarding subject anyone could study in the formative years (high school/college) no matter the specialized field. It could also be the least appreciated.
If all farmers left their device running 24/7 and had exactly the same hardware and Internet specs, they would receive approximately the same reward?
And since the assumptions above will not exist in real life:
The largest increase in farming rewards above the average farming rewards will not be expected to exceed 20% (or the percentage the network settles on). Will this percentage be dynamic or will it have to be fixed at some point?
This is indicative right now, but when the algorithm becomes more concrete we will know more. I suspect the farming rewards will change over time, I fully expect these parameters though to be handled eventually by some deep learning tools. that is far away but they are better than humans at that kind of thing. Right now it will be static though IMO and possibly altered via updates as the network grows and we know more. I don’t expect too much movement though, until archive nodes come into play, then the extra resource will be used and rewarded for that as opposed to just a bigger peer in a group. The extra resource needs to be worth something to the network and archive or cross group data will be a good reason to increase rewards for the larger vaults.
To maintain incentive for those vaults to remain on, a sort of “heartbeat lottery” at a lower earn rate could be implemented to reward nodes for uptime.
It also comes at a loss to the network if too many nodes begin doing this. Increased churn, incomplete groups etc in addition to that already caused by other factors. Better to lessen the reasons one would restart their vault if it can be helped IMO.
It’s a chart plotting the development of hypothetical outflows and inflows of economic value into the SAFE ecosystem over time.
The all factors combined represent the sum of all costs (storage, bandwidth, CPU, labor…).
The virtual baseline represents “lifting” of the zero cost level caused by non-economic forces, like altruism, merit seeking, etc… This construct captures our intuition that people are willing to cary some economic costs without expectation of economic payback. The point in time where the all factors curve meets this baseline is the moment, where for all practical purposes all costs are zero. I called it NIRVANA.
The SAFE scenario represents the balance of the virtual SAFE network account, witch must not be below the all factors curve, if the economy is to be stable. It is calculated as a difference between the inflows and outflows of national currency into the economy. It is the national currency and not Safecoin, because all costs are also in the national currency. The points on the time line labeled S1 - S3 represent different scenarios of how the account balance might evolve. We do not know which one is going to be realized. Please note - any variants of this curve “terminating” to the right of the S2 time, would mean, that the SAFE network economy has surplus (is stable).
The usage fees scenario captures the proposal to introduce fees into the SAFE economy, to keep it stable regardless of the all factors curve shape. The green line representing this account balance evolution is lifted above the main all factors curve for clarity. It should sit right on top of it.
I don’t see why the All Factors Considered line is tending to the baseline. Why would this be the case, given data continues to grow at an exponential rate?
the all factors combined are the costs external to (imposed onto) the SAFE economy. The consensus form the discussion is, that these are falling. The shape of the curve might not be hyperbole in the long run or for a particular factor (like electricity, or cost of CPU) , but right at this moment in history we seem to agree at least, that the slope of the curve is negative.
Hmm (no time now) I think a problem is that the x axis is time…
We basically need to compare storage costs per data chunk in relation to the probability it there is a farming reward for it… And then the control function of safecoin as second line… Then one can see if it is a stabilizing control circuit or a destabilizing… But since there are other factors that play into this it might not be that easy…
My goal was first to see whether the graphics qualitatively captures the reality as it is seen by the people who participated on this discussion and if not, what would need to be modified. I believe a number of inferences could be made once we could agree that it is more-less ok.
To answer your question - no. Actually I believe people will run the vaults even if they get no economic reward for it.
Maybe I am just not understanding the conclusion you are trying to draw. IMO, the cost of storage matters little, as long as it is proportional to the cost of PUTs. As long as the network remains viable and data continues to be stored, these should remain proportional.
What the chart above suggests is, that the SAFE scenario is undetermined - the SAFE economy could run a deficit, surplus, or anything in between. @foreverjoyful started this thread when he expressed a concern, that the economy might run deficit, due to the “pay once” feature. I believe he has a point, because businesses similar to SAFE network (when it comes to cost structure) like ISPs, or cloud companies, charge recurring fees for providing services. This is the normal in the industry today. Therefore it should be the duty of the defenders of the “pay once” concept to prove its economic viability. This discussion thread did not deliver such a proof and I believe it could not, because there are just too many factors to consider.
What can be done then, considering the economy might run deficit?
The network could implement some form of usage fees mechanism (the green curve) - just to be sure we make it to “nirvana”, at which point the fees can be canceled,
The network could treat the “pay once” feature as a “loss leader” and cover the potential deficit from other sources,
A radical option would be to not request PUT payments and not pay farmers for farming. This option would be consistent with the belief implied in the statements of some of the disputers, that costs like storage, bandwidth, electricity, etc… (red curve) will in a couple of years be so low, that people will be motivated to keep the network going even without farming income (the red curve will dip below the virtual baseline). After all, the ISPs and cloud services today are not paying us to keep the internet going - we are the payers. We do it because the utility we gain from the network is greater that the costs we have to incur.
Lastly, there is the “default” option - hold fingers crossed, that the “pay once” feature will somehow work. Because that’s really all we can do when we lack a believable proof that the SAFE scenario is going to be stable.
It is only pay once per data item, but there is an endless and growing stream of new data items… we have just added more here!
Dwelling on data ceasing just seems a bit pointless to me. As long as the network is viable, it will be used to store data. If or when it isn’t people will leave it irrespective of the payment model.
Also the pay once is only for immutable data. This will be a large section of the storage yes but less so of the economic input/output. Things like mail, forums, blogs which typically are small amounts of data but each cost to store and cheap to retrieve. Also things like 75-90% of blogs and nearly all (e)mails are read rarely and could include many of the specialised forums too
MDs cost to update/modify/add/delete fields will also be a factor.
So focusing on immutable data is short sighted and misses the mark somewhat for the economics.
The pay once feature obviously means that the charge will be covering the future accesses. Since for the typical file (obviously not all) this access reduces over time and so if we get the algo right there is no reason that it would not work. Just like lifetime memberships work for clubs etc so should lifetime storage work. Just need to charge the correct amount to cover the future (reducing) accesses.
Also to charge rental on persistent data means it is no longer persistent since it can disappear. So to achieve the goal of persistent data we need to have it without a method to delete it on certain conditions (eg no pay rental). Also there is no reason that a “temp” data type (non-persistent immutable data) cannot be introduced to enable removal of data unless someone “refreshes” it by rewriting that chunk. This rewriting of the chunk could be called rental by those fixated on rental/recurring fees, but it would be much wider concept.
)
