Is Safe Inflation real?

Absolutely. The algorithm’s job as I see it will simply be to facilitate a market between suppliers of resources and consumers of resources so the network maintains an optimal level of spare capacity.

That really shouldn’t need to be complicated or require human intervention.

I think RFC0057 is very different to how you expect things to work based on previous RFCs, e.g. farmers aren’t paid based on GETS, but on PUTS in their section, if I’m reading it correctly:

When a client pays to store or mutate data, the payment will be immediately be divided amongst farmers.

Though, as David has said, the current implementation is not expected to be perfect and there will be a lot of thinking & observation of test networks done before the final algorithm is designed.

If 1 SNT were received from a PUT, why couldn’t the rewards given out total 1 SNT (e.g. 0.85 to farmers, 0.05 to core devs, 0.1 to app devs)? Why would there need to be more received in store costs than given out in rewards?

I don’t think the need for a buffer is a fact.

What kind of event would cause everyone to stop putting data onto a successful and healthy global data network?

If there’s a drop in demand for storing data, the StoreCost & Farming Rewards need to drop. When this happens, it’ll trigger more uploading of data as it becomes cheaper. If you prevent the StoreCost & farming reward dropping when demand drops, you reduce the efficacy of the market that balances supply and demand of resources.

Even if short term dips in network activity became a thing, farmers would simply average the rewards out over time. I doubt they would cease farming because they had a bad hour if they can see their daily / weekly rewards are healthy.

Of course automating big uploads to take advantage of any short-term dips in StoreCost could likely be done, which would further reduce the likelihood of significant dips in demand for uploading to the network.

In my view, it just needs to be kept simple - demand and supply of resources should be kept in balance, and that doesn’t need a buffer. Hand-to-mouth is not a problem if there’s a constant, or regular flow of network usage.

All a buffer does is to smooth out the transient spikes in supply and demand. The larger the buffer/reserve capacity, the larger the damping effect to abrupt changes.

Would you rather drive a buggy with no suspension system down a bumpy road, or cruise over it at 4x the speed with air ride shock suppression?

This is a mechanical analogy to having no network reserve buffer and a purely human controlled market based pricing mechanism:

This is a mechanical analogy to having a simple/passive network reserve buffer with simple get reward and put costs adjusted by the network:

This is a mechanical analogy to having an intelligent/active (possibly AI driven) network pricing system:

1 Like

March last year. I have been preparing for the pandemic for 60 days. Food, water, masks, disinfectants, cash. The pandemic hit and closed everything in Bulgaria.

Uncertainty. Although I have been a farmer for 3 years in the Storj network, I have stopped all my farms, with the clear awareness that I will lose my rating in their network and when I start my farms again I will start from the beginning. I stopped everything without hesitation.

And it’s good that I did. The crypto market collapsed. Farming would not cover my electricity costs, and I was left without income… so we need a buffer. :dragon:

I guess I’m not expecting farmers to quit if they have a bad hour or two of rewards, especially given the importance of node aging in determining rewards, so the behaviour of farmers is an automatic buffer.

Unlike having a car with no suspension on a bumpy road, I really don’t think a short term blip in farming rewards would do any damage to farmers who have a healthy average reward, and I don’t think it’ll be a ‘bumpy road’ once the network has a good level of usage and users / automated uploaders who would want to take advantage of any short-term dips in StoreCost.

I’m sure test and Beta networks will help clarify the issues and give some insight to some behavioural factors.

Thanks Dimitar for forcing discussion of these issues. I entirely agree with those that say the data on the network is what backs SNT and gives it value. Also the businesses/apps, functionality, etc. First and foremost the farming algorithm must be optimized around this.

However, all that said, the fact that people will and have used the MAID token as a speculation brings an additional factor into the equation. BTC is literally a pure speculation. It is nothing but a ledger of transactions, but has an enormous market cap because people want a decentralized currency even if it is “nothing” in an intrinsic sense. What we do with the algorithms behind SNT need to take that aspect into account. If we neglect this we could miss out on growth or worse. I’m not going to claim I know the answer, but I think is still important to consider and model as part of the farming algorithm development.

In the end it may not matter much to the day to day use of the network as the reward in SNT could adjust to the value in fiat, but still could disenfranchise a lot of people if we massively devalue the fiat price by releasing too many SNT at once. This could also hurt farming as many will likely want to convert to fiat to finance their farms. On the other hand a “marketing fund” of SNT to release to early adopters to help grow the network initially is a very good idea (e.g. pay for Wikipedia upload).


Did the internet fall silent in Bulgaria when the pandemic hit?

Did the Storj network keep working as intended through that time? If so, you leaving might just be Storj’s network operating well and a drop in demand leading to a drop in supply?

If the Storj network had started pumping out more STORJ to hosts during that time, would it have kept you in the market despite that increase in supply pushing the price of STORJ down further? It would certainly make many people less likely to hold their STORJ if they knew the supply would be growing in times of reduced demand. I’m not sure that would be productive.

How the Safe network behaves in the event of a significant & sustained drop in demand is an important scenario to consider, but I don’t think issuing more tokens would be an effective solution to that issue - it would quickly be priced in by farmers and the market making it ineffective.

1 Like

Maybe much like the dynamics of real estate agents. The following refers to the U.S. housing market.

As a result of the last recession and elongated recovery, the active real estate agent population declined 35% from 2007 to 2014. In other words, 94,000 individuals dropped out of the working agent pool, unable or unwilling to survive in the harsher housing market environment brought on by the recession and slow-moving recovery.

The number of agents has since increased steadily each quarter, while annual sales volume has been nearly constant during this recovery.


No, our country is poor. We are far from the world’s dollar printer. While people in the United States were given free money, we were given free loans, but only if you had a job. So I turned off my farms to save money.

They do just that. The amount due is in dollars, but the payment is in storj tokens. If the price of the token drops, they send you more tokens.

The risk is that when the price goes down, you don’t know if you will be able to sell before the others. I pay my electricity bills in euros. If I sell storj tokens, I lose money not only from the low price, but also from the tax due. But if the company Storj had announced that they would pay me double or triple, I would have left my farms and risked losing cash.

Therefore, I believe that if there is a reserve and you know that you will receive more tokens, the decline will not stop, but its volatility will decrease.

1 Like

After launch, it will be very important, imo, to continue to study the dynamics of market-based SNT prices as they relate to actual storage costs on the network and be ready to make adjustments in the farming algorithm as needed.

1 Like

While you work with the all or none then you will not see the issue.

Its a case of say on one day there is X SNT spent to store per hour (network wide figures), down from a larger amount.
And there is a large event (say the first rocket to mars), and initially the gets are typical and are using the SNT as it comes in. So it might be a modest “Z” SNT per GET. And bam the rocket is in its final countdown and the rate of spending drops to X/100 as people drop everything and watch the launch on SAFE with GETS going up 10,000 times as people are getting the live stream. (Yes the launch is being PUT to Safe and almost the only thing for 3 hours).

So then its X/100 coming in per hour and 10,000 times the gets. IE 1 millionth the rewards.

Use imagination and see how the network living hand to mouth will see massive swings in farming rewards as the rate of PUTs compared to GETs swing up and down depending on events of the day and even day/night cycle hit high and low population regions.

An increase in GETs does not indicate an increase in PUTs occurring. They are quite independent for the most part. Only live streams will have a correspondence and can be inversely related. But for the most part there is not a direct relationship.

Thus living hand to mouth can be a dangerous situation in the confidence of the farmers. I’d say farmers who see one week getting “K” SNT per day and the next week getting “K/100” per day, then getting “K*2” the third week might decide its just too much of an unpredictable roller coaster ride to commit their resources too.

Its a disincentive

Oh and the Mars mission might cause problems for the first week, then at various times during the mission, especially when they are going to land and for a few weeks following.

If you read the quote from the document you referenced then it you’d see that the principles were not being changed but just the calculations and mostly to be simple for testnets. The principle is still that the cost of PUTs is logical disconnected from the rewards. Only the mathematical calculation of the two are similar equations. So similar that the temporary maths use one calculation as part of the other. IE the equation to get spend cost is doubled to get GET cost according to what you posted. But it does not mean there are 2 PUTs for every GET occurring.

1 Like

I see you refuse to consider a physical analog to the market. There is another simple analogy that mimicks the operation a properly implemented buffer/reserve/capacitor/flywheel/damper. Consider a comparison of any commodity price against its moving average over different time scales. The effect of the buffer is to cause an instantaneous price curve to look more like a moving average curve. The larger the buffer, the greater the smoothing effect (up to natural limits).

The network buffer should have negligible effect on the external fiat market (the road). All it does is smooth out and stabilize the pricea (PUT cost and GET reward) in terms of SNT within the network so that the SNT price to store a 1MB chunk has low volatility. The fiat road will be what it will be.

Farmers will make their decision to leave based on the fiat price of SNT and the number of SNT they are receiving in relation to energy and hardware fiat costs. One can’t state in blanket terms that doubling the number of SNT in human hands will cause the fiat price for SNT to get cut in half. It’s not that simple due to the multiple levels of indirect coupling and user demand.

1 Like

Then, as the rate of puts drops by 5% near the rocket launch, a bunch of low-priority network uploads / backups waiting for such a dip kick in, meaning the dip is not that significant and hardly noticed by farmers. If the dip continued beyond low-priority queued tasks, then it’s likely a more sustained drop in demand, and it’s important to let farmers know through the pricing mechanism that they shouldn’t add more capacity for now.

If there were massive daily swings in rewards, I agree that some smoothing would be helpful. If however, swings are sub-daily (as I expect), I don’t think there would be any real need. We won’t know for sure until tests are up and running with a good number of participants.

I guess we really will need to see how the network acts once it’s up and running in tests ahead of launch to see what kind of scenarios are the most likely.

Agreed on this - there is no reason why the two must be closely, even though both are likely to be correlated to the number of active users on the network at any one time. I wouldn’t be surprised if during the evening there are relatively more GETS than PUTS due to people watching stuff etc.

1 Like

I understand the analogy, just don’t have a high expectation that the smoothing will be required due to the factors I mentioned (fluctuations likely being short term, farmer behaviour, and likely automated uploads that are triggered by dips).

If such a buffer only smoothed out bumps like a moving average, then I agree it wouldn’t be harmful, as price signals would still get through pretty quickly & ‘noise’ would be reduced. But, if you tried to apply that over weeks / months / years, it may be more problematic in terms of possibly distorting the market.

The typical moving average ranges used by traders (50d na, 100d ma, 200d ma) as shown in the image above represent a reasonable amount of smoothing that would not offend their psyche. However, the use of a reserve buffer also allows for network growth optimization. This means not just stabilizing the SNT prices of a chunk, but dynamically altering them to maximize network growth. This aspect is more crucial just after launch and decreases in importance as the network grows to internet scale.

1 Like

what you saying? There will be more then 450 milions token? I didnt know that. I think its like bitcoin, 21 mil and stop?

Maid is only 10% of all coins that will be available on live network.


what happened with farmaring after all coins will be in network?

Coins get recycled.
As coins are spent on storage they are returned back to the network so can be re issued as farming rewards.


In my mind I have a solution similar to free float trading, a client sets a max/minimum limit, then clients ask elders what price farmers want for put/gets, then random buys data within limits, for what data is needed within nearest 30-60 seconds.

Random automatic trade, pre data storing (reserving space), would happen, let’s say 25MB from 5 random farmers within price limits. Client would automatic buy a small amount (like games allocate gpu vram memory to cover the nearest 30-60 seconds of game play) to cover put/get demand for nearest 30-60 seconds or so. This event will repeat when client gets close running out of bought reserved space. If price for storing change out of limits, then client will be prompted to raise/lower limits.

A farmer who decides to steal value of let’s say 25-50MB worth of data will get punished by removed age that will cost more to recover then to play fair.

Would it be able to work like this then no store buffer would be needed, no farming rewards needed, just free-floating market put/get trading.
In my mind this is a noble solution even if I don’t know if it is possible to achieve.

You know that evolution is a process of selecting the useful, right?

There are many reasons for something to appear, but there is only one reason for it to remain.

If fiat money was not useful to society as a whole, there would be no power in the universe to keep them in existence for so long. They would be removed and replaced again with gold, but they aren’t so Fiat money are useful and that’s why they are still here.

The centralized printer is the cancer. I explained why - it allows all the losses of the Too big to fail corporations to be transferred to society.

The decentralized printer of cryptocurrencies is the treatment for this cancer - yes, it appears because of human greed and speculation, but it will remain because it is useful to society.