StoreCost for a sustainable network

The company will have at least 2 types of data, the operating data and the backups.

If they use the SAFE network for operating data and for some of their backups then all they need is to have carefully chosen backups off line. (some offline storage heavily rewritten).

2 Likes

The logic was for any market with a current and given max supply, if the asset is traded by supply/demand.

The example with Fiat was just an easy way to illustrate the logic.

If you buy a glas of water from a pool with a closed loop, it is only important what the current level of water is or the max level of water for the pool, how fast the water flows through the pool in a closed loop should not be of importance. The example also relies on that after you drink the water also give it back to the pool so it don’t dry up after some time. :slightly_smiling_face:

3 Likes

Except its inverted for one. Its the amount of water NOT in the pool that is important. The amount in the pool is what doesn’t count towards @mav’s analysis above and that was my focus.

Its not getting a glass of water, but the rate the water is returned for the work done. Work being data stored. Its not like the water is the product either.

Its not a static state that is giving value. Its the amount during set periods that is used for working out a value. IE rates.

NOTE: This is for @mav’s question above. The safecoin PUT cost

The speculation value is different and I think that is what you are saying

1 Like

It don’t make logic sense to me. I will try to take your approach for a spin a few more times in my head when I’am more well rested. :slightly_smiling_face: But my logic in above comment should hold true for any asset. :slightly_smiling_face:

The rate returned can only possibly give a higher price, if it for some reason is a delay in the recycling process.

The turn over rate could give a higher price for storage because it gives high demand for storage but that should also be compensated by more farmers which would lower the price again. But I will come back when more well rested, mayb we talk about different things. :slightly_smiling_face:

2 Likes

Having not read or thought this through…:thinking:

Perhaps there are opportunities to not be bound by unknowns?..

Whatif a network was live that only assured short-term at a cost competitive. Some cutover from that v2 follows with bells on and a long term commitment… based on a sense of confidence and pricing… perhaps now because of assurance and long-term store, at higher one off cost. Benefit being small stable steps, less risk, opportunity for bug fix. Safecoin perhaps single real continuity through those.

Trying to do everything at once is fine but more stress… offering a simple store solution alone is powerful competition for old style without privacy. If the rate of return for nodes is good then that assures stability for providers and users, more than jump in dark… which initially like any new market service bound to pricing, might be volatile.

2 Likes

A few random thoughts:

The reward system and PUT cost needs to be economically viable whether SAFE is $0.01/coin or $10,000/coin. I think only a supply/demand feedback mechanism will work for this given what will likely be a wild ride for SAFE coin, and some kind of damping may be required to keep things from getting crazy. On the other hand, if we try to mandate a price I think we are doomed for failure.

We need to be careful about assuming people offering up “free” resources to keep cost down. We don’t know how much churn there may be. Anyone who has run OmniCore will realize that’s not something you can do on a machine you want to do other things on that are intensive. And at some point we may outgrow that pool of resources and end up relying more on professional farms.

We need to consider GETS. These need to be payed for by the initial PUTS unless there is another mechanism in place. The actual storage space is only one piece of the overall cost. There is the internet connection, power, environmental management, maybe local backup, etc. to consider. And if not local backup when a vault drops offline there is another PUT that occurs that is essentially free when that data gets re-copied by the network itself. Probably a very small overhead cost, but these things add up so I just caution about imposing too low a price as it seems we could discourage network growth. And, obviously, too high a price will kill growth too. So somehow a free-market approach has to find the sweet spot I think.

I guess this a roundabout way of saying the price should be whatever it needs to be. There was another thread on this, and several of us offered we would be willing to pay in the neighborhood of $1 to $10/Gb for what the SAFE network is offering in way of a storage solution at today’s prices.

4 Likes

I was getting drawn into trying to understand the OP calculation and network behaviour, but isn’t StoreCost really going to be determined by:

  • the perceived value of a SafeGB in fiat

  • the maximum PUT cost in Safecoin, which will be SafeGB in fiat divided by the exchange rate of fiat per Safecoin

[Aside: For the network to be viable, the maximum farming reward needed to secure an additional GB storage must remain near or below the maximum PUT cost per SafeGB in fiat (over a sustained interval, as short term fluctuations would not cause the network to fail immediately). The actual farming reward will be as low as possible, while enough to ensure it is near or slightly above the PUT cost per SafeGB in fiat.]

The monetary figures in Safecoin can’t easily be calculated because they depend on the Safecoin fiat exchange rate which is damped, and perhaps bounded, by the characteristics of the network, but determined by multiple unpredictable factors.

The best guide we have will I think be other cryptocurrency projects that have been around for a long period, such as bitcoin and Ether. I think those are probably good examples because SAFE Network is similar in breaking new ground (bitcoin) and creating new applications and services through additional features and desirable characteristics (Ethereum).

So when the network launches and Maidsafecoin is converted to Safecoin there will be quite a wide range of possible values while people try to determine the speculative value of the coin over different time periods. I’d expect the coin to be in demand, with fiat price rising but very volatile initially, and then rising in fits and spurts over the early years as demand grows, applications expand, reputation matures etc. Eventually it will remain within a less volatile range and the range may rise or fall according to external changes. That is all just speculation and guesswork though, so don’t trust it!

But putting figures on any of that will be unreliable guesswork IMO, which carries over directly to any Safecoin denominated figures that depend on estimates of things denominated in fiat (such as the cost of cloud storage).

3 Likes

What about having different price models for different data types?

It would be nice to have some situation that you have to pay for your GETs, so Safecoin is backed by data (GETs).
But you don’t want this model for getting public data like websites.

So having different models seems valid for me.
For example:

Public data (also NRS names)
High PUT price, free GETs.
Higher put price makes sense because public data will consume more resources than private data.
But the data should be available for everybody, so free GETs. If pay-the-producer model is applied, the PUT price might be even higer.

Private data
Low PUT price, low(?) GET price.
Uses less resources, so lower price. The GET price will compensate for the lower PUT price. This is a cheap way for backing up data, so it can be very competitive to other cloud services.
Most persons will have their private data local anyway, so most of the time they don’t have to pay.
They will need SAFE when they NEED their backups and are willing to pay then. In this way Safecoin is really backed by (private) data.

Or forget the data types all together and let the uploader decide what price model he wants.
High PUT price / Free GET price.
LOW PUT price / Non-free GET price.

4 Likes

I imagine having my whole /home directory on SAFE. And eventually maybe even booting the kernel from SAFE.

4 Likes

While I agree with your point, I just wanted to add that even backup data will not sit there idly. Somehow the network needs to assure that the data is really stored by the vaults. I think that this will be evident during churn, but I don’t think we have info on how the network will look after this in the ‘steady state’. If the network doesn’t look after this, the users should check from time to time that the data is there by GETting it (which is free to do anyway, so why not!). Your point that data will be accessed unevenly definately stands though.

2 Likes

Before mounting SAFE you will need the kernel, so I don’t know how you could do this.

But the cloud drives I know all keep copies local on your HDD, they only GET data when the data on the server is newer. Otherwise they have to load everything in memory each time. Or are some services working differently? I am not that familiar with it

1 Like

I guess you’re right. I’m just speculating. My point is, if storage is cheap enough, I’d like to use SAFE very actively, just like any drive.

2 Likes

Well - but you could have a read only memory with the code for getting the boot data off safe and therefore a virus safe hardware… You could get a virus but with the next image startup it would be gone again (as long as there is no session data recovered…)

1 Like

You would need Pxe boot or similar, maybe modded. You could call it SAFE PXE.

6 Likes

Yes, something like that. I haven’t studied on it much, but I figured I could just tell the Bios to go look somewhere else beside my local drive. Backing up the whole system, not just documents and stuff, into “the cloud” would be nice.

2 Likes

Good point. I guess it’s hard to know the amounts, but the variables for how many coins will be used for storing data seem to be

  • percent of max coins that have been rewarded
  • percent of rewarded coins used for storage vs non-storage
  • how many times each coin is recycled

So there may be more than 4B coins being used to represent total spent on storage (eg might be 10B if coins are reused an average of 100 times), or it may be less than 4B coins (eg 1M coins if most are used for non-storage purposes). Kinda hard to know isn’t it.

I still feel the original calculation is an ok approximation for the possible range of storecost prices.

I think you’re right about the perceptions in fiat being important for storecost. Not sure how this affects the design of storecost algorithm since it won’t know about fiat. Do you think fiat (exchange) price will lead or lag storecost (network) price?

Yeah would be cool to see a chart of bitcoin tx fees in fiat and in btc (kinda their equivalent of storecost). Not exactly comparable but might give an indication of what to expect?

I wonder if we’ll be able to measure this on Fleming?! That would be a cool stat to gather.

1 Like

I think it will be a see saw - either could lead - but my intuition is that for quite some time fiat price of Safecoin will dominate going up or down more quickly, followed by and damped by PUT cost in fiat. I would guess the side which dominates will be the side on which there is most turnover.

But who knows! I haven’t thought deeply about it.

2 Likes

What about the fiat price will both lead and lag storecost price?

I Imagine the following, a farmer sees that Safecoin is traded on exchange for 50$/coin, to cover cost and a profit margin let’s say the famer want’s 5 cents per GB, the farmer sets 0.005 Safecoin as price. I Imagine that the farmer also have some kind of app that auto adjust the asked price if exchange price goes up/down. Maybe some json or similar file that the app can access to adjust asked store cost price. Ps (I don’t know about technical limitations or possibilities)

The exchange price will depend on current/max supply of Safecoins, so I have a hard time understanding how turnover/recycle rate would affect things as price will be set on the exchange and farmers adjust amount of coins to the price of the exchange.

I still can’t wrap my head around how the recycle rate would affect things. If someone can expand deeper on reasoning and give examples that would be interesting.

1 Like

Incidentally, Storj just published their prices, so now there’s more of a like for like comparison, except they don’t store in perpetuity of course.

  • Static storage is charged at 10 per terabyte per month, or .01 per gigabyte. Storage is calculated in “terabyte months,” which is a prorated calculation of how much storage is used throughout the month, broken down by hour. Storing one terabyte for 30 days, two terabytes of data for 15 days, or three terabytes for 10 days, would each be the equivalent of one terabyte month.
  • Download bandwidth is charged at 45 per terabyte, or .045 per gigabyte, of bandwidth consumed. Storj does not charge for upload bandwidth, so uploading files to the network doesn’t incur any cost. Download bandwidth measures the transfer of data from a Storage Node Operator (or Host) to an Uplink. We’re also offering discounts if you decide to leave the network, so customers don’t feel locked in. That’s how confident we are that you will love Tardigrade.

https://storj.io/blog/2019/11/announcing-pioneer-2-and-tardigrade.io-pricing/

8 Likes

I would agree that initially that somewhere around this is reasonable. But in fact my initial thought is that it would be less than 4 billion since the initial amount is around 700 million (~15%) and the markets are valuing the coin based on what is available but being mindful that it is expected to increase to somewhere between 3 and 4 billion in over a decade.

I also would expect later on that at any one time there is less than a billion ear-marked for storage upload. In effect I would be using 1 billion as an approximate number of coins to be used for uploading at almost any point in time. I am not saying that the billion will be used in that period though, just there are that many earmarked and obviously thinking of a year or two after launch.

Obviously the billion figure is not meant to be absolutely accurate but a reasonable guesstimate as to the number of coins that at any one time people have reserved for data uploading and not for buying/selling/trading. To others - one person having 1000 coins may say 500 is for uploading and take 12 months to do it and later on think that 250 more could be used. And another similar person may only use 250. The figure I suggest is that amount people consider is kept for uploading at a specific point in time.

Then the price is determined by the all the coins and based on the utility (buy/sell, resource purchasing, speculation, ???)


if reused 100 times, 10B suggests only an average of 100 million coins are reused.


@mav in summery I would think that the long term figure used would be better if it was 1 billion coins as this seems reasonable as a first estimate of the coins at any one time that is ear-marked for uploading. Initially (1st year) maybe 250 or 333 million coins might be a better figure.

We may in future though realise that the utility for buy/sell over the network is much higher and the figure is less than 1 billion. OR the opposite and be much higher.

But then we need to consider that if there is massive usage of storage then the coins for uploading might be always a lot less than 1 billion, but that needs to be seen.

2 Likes