StoreCost for a sustainable network

If SAFE is to become the new internet, what should the cost of storage (storecost) be? This topic is for exploring the storecost algorithm.


There’s currently about 2.3 trillion USD combined value in Amazon (865B), Google (905B) and Facebook (565B).

SAFE would be trying to represent at least that amount of value with 4 billion safecoin.

That amount of value makes a price of approx 550 USD per safecoin.

Amazon storage (using Simple Storage Service) is around 2c per GB.

At that price level, one safecoin would pay for around 26K GB of storage on AWS.

That makes for a required storecost of 0.00002 safecoin per PUT.

The algorithm in rfc0057 (based on percentage of full nodes) allows a lowest storecost of 0.005 safecoin per PUT (source). RFC0057 was always intended to be changed and this analysis shows it will be necessary if SAFE becomes very popular.

Is this approach to storecost valid?

How do you think storecost should be determined?

20 Likes

I think this is an important point. My opinion is that it should be the lowest possible value that ordinary people are prepared to accept. I mean if the payments are to low then the resources dry up etc. (we all know about farming). However the macro stance is that this will not be about corporate profit, but individual choice. some may operate vaults at a strict financial loss, but do so for other reasons such as freedom, inclusivity of others and more. To me SAFE in many ways will show us the true human price storage costs.

18 Likes

What about time? 2c/GB/t … I’m assuming that AWS isn’t offering permanent storage. Without that info, I don’t see how we can make any valid comparisons.

11 Likes

Good point. The S3 storage price is 2c per GB per month. We could get more complex and make the comparison more detailed, but I was looking to start with for back-of-the-envelope / orders-of-magnitude type analysis.

11 Likes

Be worth trying to get figures on how many months folk pay for a particular data item perhaps? I am not sure that is available, but may help in that view.

Another thing that would help is trying or asking what cost there is to lost data from years gone by, scientific papers, experiments etc.

14 Likes

It always seems a challenge when asked how/when we would ever even get close to the full 4 billion plus float as it is thus making that 550 number low at least temporarily. That number based off of current supply would make it somewhere between 4-5k a coin which sounds pretty ridiculous. My greater point is won’t price be very dependent on how much supply is farmed, used, held ect. At any given time?

4 Likes

Eternal storage of information is highly underestimated in my opinion. How much am I willing to pay to keep my family’s photos forever for the next generations of my family? Hundreds of dollars per gigabyte for sure…

7 Likes

It feels to me that there are two distinct markets here, one for data that is being used temporarily but for an unknown duration, and a second for archival data. Two markets, two price-points.

AWS is a good indicator for the first market, but maybe not for the second? To use myself as an example, data that I want to store permanently goes onto an SSD, not the cloud. For some it might be a RAID storage system (with regular HD’s) - the latter being cheaper, but requiring more technical skill and maintenance. As we hope for Safe to be easy for grandma and not technical to use and as it will require no maintenance I feel that the SSD comparison is the best …

So, current SSD cost/GB that I just checked on ebay is now ~0.14c/GB (for a 1TB Samsung drive).

As Safe intends to serve the latter market, IMO, that’s a better metric. I didn’t include the cost of a computer, but I think using an SSD or the cloud would both require a computer/device we can exclude that from the calculation.

Yes. Even if only six months worth of cloud time we are at 12c/GB which is close to the cost of the SSD storage, but 6mo., versus lifetime of SSD (a decade or possibly more if only archival use and drive is good quality and stored well).

Edit: and in a decade when that SSD is replaced, the cost/GB will maybe be a tenth of that - thinking in terms of a Safe farmer … 12c/GB + 0.012c/GB + 0.0012c/GB infinite series … we are coming to a number around 14c/GB - real hard permanent storage cost. But this assumes we start the calculation today, and not in two years or so when Safe begins to take off … at which point starting cost may drop significantly.

Of course if the world hyperinflates, then all bets are off.

10 Likes

I think what you calculate is how much it would cost a farmer to store 1GB of information, not how much people would be willing to pay for it. In my opinion people will be willing to pay between x10 and x1000 :dragon:

4 Likes

I would have though that 2.3 trillion is not to be represented by the whole of the possible coins. Otherwise we limit it to that market size. There are a few other markets that are not insignificant to those you mention.

To me those markets represent more like 10% of what SAFE is all about. For example buying and selling over the internet represent a growing segment.

ALSO more importantly is that there will never be the whole 4 billion coins existing at once if the recycling algorithm is working effectively

AND more importantly using 4 billion as the number of coins is WRONG altogether. Recycling means that data purchasing will involve many times that 4 Billion as the coins are recycled and farmed to be used again for storage So over time it could be 20 or 50 or 100 billion coins used to store all the data people want.

Of course the pricing is on a decreasing curve too. So future pricing will be even lower yet the value of safecoin (storage and utility) will not be dropping overall.

Thus whatever the today safecoin cost of a PUT the future safecoin price very likely be lower. And so we would need to allow for even lower PUT cost

And of course using spare resources means that the actual costs to the farmer for farming is a lot lower than for say Amazon.

This is extremely hard to quantify for what we need to know.

Someone keeping their file storage for 6 months might have just taken up another “sale” elsewhere and moved their data. Or they buy another SD card for their phone and move it to there. Or they buy a NAS at home.

So in fact all these cases are what we want for SAFE perpetual data.

If we could get the figures they really would be meaningless other than what they spent on one portion of their total data storage needs.

But we can use the data usage figures of a reducing usage over time. But this varies between type of data too.

My gut feeling is that using 6 to 12 months of storage as the comparison point would be a good starting point.

11 Likes

imho, we here can never hope to centrally plan an efficient price. That is a function of free markets, to provide price signals that guide participants to allocate resources efficiently.

I am not an expert on SAFE network yet, but I would hope that there will be mechanisms for farmers providing resources to set a price and “customers” to accept, counter-offer, or go elsewhere.


5 Likes

You have not specified which “price” you are meaning.

safecoin Put price is the network price for a PUT and not controlled by markets.

Safecoin Fiat price is decided by the markets

1 Like

Yes, good points, I think pricing will need to be flexible enough to become very cheap in safecoin per PUT.

Maybe half way between might work, not a strict algorithm but not a pure market, something that can flexibly adapt the storecost within some hard boundaries?

Not sure what that would look like though.


I think it makes sense for the availability (or scarcity) of spare space to be a factor in storecost, whether that be introduced by algorithm or market mechanism I think it’s inevitable to be part of it.

Do you think network size should be considered as well? A larger network should result in cheaper storage? Or is that already implied since a larger network should have more spare storage than a small network so storecost is already going to be cheaper? Not sure on this, but I think network size doesn’t need to be an explicit factor in storecost.

1 Like

@mav, I think that one major issue on the actual safecoin PUT price calcs you have done is that recycling means that using 4 billion is incorrect because over the years it will be a lot higher.

1 Like

I realise this is an order-of-magnitude type calculation but I think you’re setting the $ value of current cloud storage a little high, since Facebook and Google make the majority of their money from advertising rather than purely storage and Amazon provides many other cloud services, plus retail. It always depends on how you cut and dice these things of course, but this analyst firm puts the value of global cloud storage at $207.05 billion by 2026.

I wonder how flexible we can be with the PUT price in the early days and whether fine tuning will be possible after launch?

4 Likes

Recycling means that data purchasing will involve many times that 4 Billion as the coins are recycled and farmed to be used again for storage So over time it could be 20 or 50 or 100 billion coins used to store all the data people want.

The only important numbers for calculating price can never be anything other than current future supply or max supply 4.5 billion, never above 4.5. It do not matter if you recycle 10 billion or 100 trillion, only the coins in play will be used to calculate price. If you would for example make a deal with the national bank that you would put 100 billion in paper money on the barbecue and then the national bank would print 100 billion new money and give to you, then you would not have affected the money supply or value, it would be the same as before your bbq.

I don’t know if you meant what I interpreted you meant.

2 Likes

I was pointing out that when purchasing data resource then using 4 billion does not give any reality to the PUT price calculation.

You describe the fiat side of things. At no time was I trying to say how this information would affect the fiat cost of a PUT.

I pointed it out to prompt @mav to reconsider that side of things.

If I was to try and say what the result of what I said was, then I’d say that the rate of uploading needs to be accounted for and not just what 4 billion coins in fiat would do to the safecoin PUT price.

Safecoin is not a static coin like blockchains are and we cannot say “Oh the market is worth $X so each coin is worth $X/4billion in purchasing power.”

It is more like What is the turn over of that market in a set period and what is the average amount of coin used for data uploading during that period and then its more like $Turnover/Ave#coins-used-for-uploading

If we talked of another market then need to consider the number of coins used in the set period for that market

3 Likes

To be realistic: Not eternal but limited by the life of the SAFE Network. The SAFE Network might start to irreversibly lose participants one day and reach a point where (all) data is lost. That’s a risk anyone should take into account when storing data on the network. Personally I do not think this is a realistic scenario within the foreseeable future, but when talking about many generations you can never be sure.

6 Likes

This is a good point. If a new company came to me and said ‘look, we may be charging ten times more than Amazon but our storage is forever’ I’d certainly be looking for some sort of guarantee. That’s a big trust hurdle to overcome.

5 Likes

At least for the less critical data then one would expect they have time to copy their data off the network before it gets critical. One would expect there would be indications of the network reducing in size.

3 Likes