Specifics of the farming algorithm

Steady state on a short term time frame. The 50% setpoint could fluctuate long term, and dynamic variations around this setpoint on shorter time scales. I wouldn’t say “unstable” but rather “adaptable”.

2 Likes


7 Likes

Probably solving the OOM issue, right?

1 Like

Nopes.
This is in farming branch, it is not yet included in the testnets.
This was just an issue among others that we are rooting out while testing these new features. (There will be more).

7 Likes

Sorry, just goofing. One of thing I remember from programming is infinite loops are very bad. :slight_smile:

There is a need for minting new coins. Part of what makes bitcoin successful is they promised max 21M coins and they delivered it, still do, will continue to. The promise is important.

MaidSafe promised 2^32 coins. People engaged with the project based on that. They also promised some loan repayment stuff and bnktothefuture shares and conversion stuff, and those promises also matter. Changing the total supply now is a terrible idea because it puts the security of the whole project at risk by breaking the promise.

This was the case in rfc0012, but since rfc0057 and this current design farmers are not paid when clients download. They are only paid when data is uploaded.

I can appreciate where you’re coming from, but as a general point I think we’re going to have some communication difficulties around farming if we are talking about different (old) things. And when the farming has and is changing so frequently discussion can become difficult unless we are very clear about what/which/when we are talking.

But what about every other farmer? They’d rather get the bigger reward from the dilution :slight_smile: You’d be protecting your initial rewards but missing out on a lot of potential future rewards. The most value comes from attracting the most people (network effects are more important than exclusivity effects).

Ah I see, I thought all nodes were paid at the same time when elders were changed. I see now that each vault is paid their reward when they relocate, so payout is naturally staggered in that way.

Yes, interesting point. Whether checks on data will be part of farming or not, it’s an interesting question. Maybe failed check has some economic consequence but is not part of farming per se, we will have to wait and see.

8 Likes

Thanks for correcting me! :dragon:

This does not seem a reasonable decision. Storj and Sia pay for both upload and download to the farmers. If you only pay for upload, why won’t people reset their Vaults when they fill up?

In fact, that’s good. This will allow you to be a farmer in several SAFE Networks at the same time. While waiting in line to enter in one SAFE Network, you farm in another. When you fill the space in one, you reset and start filling in the other

2 Likes

I agree, but it’s temporary for the sake of getting a testnet up and running.

3 Likes

Come to think of it, this is good for humanity because it will stimulate the creation of many networks. If there is only 1 dominant network this is 1 point of failure. If there are many who compete for the resources of farmers, it will lead to better results for humanity…

Only if you assume there will be no checks that nodes do have data (which is what a GET is, that is the quality it brings - for popular data), even though it has been said in this topic that that is to be implemented.

Paying on upload is not necessarily that temporary. It is in fact simpler than pay on download, and pay on download does not provide anything unique that motivates the added complexity. In fact, as described in a post above, it introduces problems with encouraging false gets, without guaranteeing permanence of all data.

7 Likes

My expectation is that given the choice between a reward that maintains or grows its value over time relative to a reward that will be significantly diluted over time, most farmers would choose the former.

Would you rather be paid in USD or Argente Pesos? (slightly extreme example perhaps).

Assuming the market has a certain demand for network resources & the network balances supply & demand, paying out a numerically higher reward to farmers than users pay the network to consume resources doesn’t increase the value of the reward to farmers, it only reduces the value of the network’s currency & therefore all previous farming rewards and holdings in that currency.

It seems like programmed in pure inflation with a single function of devaluing previous earnings, investment, and wealth. This may be worthwhile, but I can’t see a convincing reason why.

2 Likes

25 posts were merged into an existing topic: About the possible existence of multiple SAFE Networks

Pay on Put ensures a steady queue of new farmers lining up to join the network. It’s kind of like a UBI for farmers. Pay on Get with big rewards to vaults based on various relative performance criteria and chunk age is the only way to incentivise continuous quality of service improvements and perpetual data.

1 Like

There’s the carrot, and there’s the stick. What you describe is the carrot only. I think it’s incorrect to say that it would be the only way to ensure those properties. If the only way you can keep earning money is to not get kicked out, and you get kicked out by not having the chunks you should, then the exact same properties are achieved.

4 Likes

Yes, you need the stick too. And the fear of losing out mechanism (FOLO?) you describe is one layer of the incentive strategy. But that will not be enough to promote continual investment to improve quality and performance. Farmers are people, and there is a significant psychological benefit to being rewarded for meritous service rather than for just showing up and fear of rejection. Positive vs negative reward mechanisms etc. Seems obvious that network needs all these features, POP, POG, Audit, Stick.

3 Likes

I include some measurement of quality in the definition of “having the chunk”, and so that audit would be more than just about “showing up”. It can be deviced to put the same evolutionary pressure on quality of service.

But I agree that there are many possible models, and there are models that I think could be more useful, but they are not within same ease of reach, which is an important factor now.

4 Likes

I think you currently have a good approach… Need to start somewhere. Pay on Put has some nice features/incentives so use them. It will be easier to see a simple way forward for Pay on Get after you have a testnet with the Pay on Put mechanism fully functioning. Intuition tells me that that an ideal exists around weighting the total reward payouts somewhere around 10% PoP and 90% PoG. Of the 90% PoG you could then have payouts weighted based on node age, chunk age, xor distance, and latency with older and faster chunks closest to the original vault being worth the most. Might also consider giving some PoG for passing an audit. It then becomes a double incentive to do a good job as a farmer. Double carrot and double stick.

6 Likes

But it wouldn’t be a bigger reward if the relative value of Safecoin tanks. It would be like all these other coins that aren’t even worth what it costs to mine (or in this case, farm) them.

Agreed.

Once again, agreed. For this to not be the case would require that there are no economies of scale. The larger the network gets the faster, more efficient, and more secure it becomes. Those factors alone present formidable competitive advantage to rival networks.

So are you planning to fork the Network? Perhaps SAFE has found its Roger Ver come Craig Wright.

At best it would be like Bitcoin Diamond, Cash, SV and the litany of other Bitcoins that no one really trusts or uses.

5 Likes

If I understand correctly, I like it.

The idea of a reward being held back until relocation should keep vaults honest. It means:

  • They can’t just cut and run without losing their earnings. They have to stick at it until relocation time.
  • They must continue to prove they have what is requested (perhaps even spot checks, etc).
  • If they leave after relocation, then the network had invested the least effort in them.
4 Likes

I think we are talking a lot about farmers but little about the other pillars that are needed to make the network a success.

We need:

1.-Users. Who will only enter the network if they find some advantage in doing so. The benefits will come from access to interesting data and helpful applications. So we need…

2.Public data and Apps. We need the network to be filled as soon as possible with useful information and applications that give something that is not found elsewhere. At the beginning the new data must be very cheap and, a point that we must emphasize, we need developers of applications that find it beneficial to work for the network.
Finally we need…

3.-Farmers to store and manage this data, which will have to be compensated in a fair and proper way to make it profitable to continue doing it in time. As the above points increase (users, data and applications) we will need to increase the number of farmers which will also increase the security of the network which, by the way, will be the best way to avoid the existence of forks.

9 Likes