The Safe Network's economics

Autonomi Network Token
1

Hi all, I’m a frequent lurker of these forums. I’ve been able to get all of the info I wanted just from reading posts and replies up until now.

I’ve been following Maidsafe since before the crowdsale and wrote a couple of blog posts on deadloch.com, I’ve got the basics down fairly well. What I can’t seem to get my head around is the flow of Safecoin and how this relates to general users and farmers.

Users

I’ve seen some discussion regarding how much free storage space to give to new users. I am in agreement with those who argue that new accounts should be given just enough space for basic functionality. If users want to store more data, they pay for it. This raises a couple of questions:

  • Is the storage of data the only thing people need to pay for? What if a user just wants to access a publicly shared file? Is passive use of the network (no data storage) free?
  • How is a price for storage established by the Safe Network? If a user wants to store data how does the network decide the rate they will pay per GB? I have seen some discussion on this as well, and it appears that the network knows how much storage is being supplied by farmers, and how much of this is used/unused. Using this information, network algorithms can adjust the price of storage to bring supply and demand into balance. Sounds nice, but how will this actually work? For example, what would constitute an oversupply of storage? 10 x more available storage than used storage? How aggressively would the price of storage be changed to reflect more/less storage available on the network?

Farmers

Farmers earn Safecoin by responding to GET requests on chunks that they are storing. Their income stream will be predominantly made up of newly minted Safecoins, but will also include a small amount of recycled Safecoins due to the expenditure of users discussed above. More questions:

  • How are new Safecoins minted? And by that I mean, how does the network decide how much safecoin to award to a vault that responds to a GET request? I’ve seen graphs depicting the intended distribution of Safecoin over time and it appears to follow a model similar to bitcoin, ie high volume of new minted coins intially that tapers off slowly over time. Does this mean that there will be a decided upon number of Safecoins minted per day? Per hour? Per minute? Will safecoin minting be completely independent of the number of GET requests in a period of time?
  • Where do Safecoin payments for storage go? In what manner are these payments earned by farmers?

Apologies for all the questions, but I have done my best to find solid answers to these points with no success. Hopefully someone can shed some light. I need to have some confidence that the network’s economics will be designed in a way that allows the network to flourish. This is very important given the assumptions I have made:

  • The tech will work
  • The tech will be more efficient than existing alternatives for at least a few use cases

If the economics aren’t right first pop (and even if they are!) we’ll see the code copied and new economic incentives tried out. An implementation will gain popularity eventually, great for humanity and the future of decentralized internet, not so great for Safecoin investors like myself!

5 Likes
2

There are several steps to the minting process as I understand it. So the first is that when a GET request is issued, the nodes with the appropriate caches compete to try and meet it. The node who successfully meets the GET request has a certain chance of a minting attempt (I’m not clear on whether this is a node rank based chance i.e. how many times has this node successfully met a GET request within a certain amount of time, or whether its a network based request, how many GET requests have been serviced within a certain amount of time, I suspect its the latter). @dallyshalla can you shed some light on this?

However at this point the process is still not over! You’ll notice that I said minting attempt, because once you have won the right, then the 7 close groups form, and create a minting consensus, and select a certain Safecoin address. (Safecoin is based on a coin-based approach rather than a wallet based approach like Bitcoin. In bitcoin the blockchain evaluates and registers transactions between wallets, whereas each individual safecoin address is controlled by a particular key pair). So if the safecoin address selected by the consensus group has already been minted, and is already controlled, then you receive nothing. This is essentially the difficulty rate.

The way that this difficulty can go down, is that when people spend Safecoin on PUT requests, the network reabsorbs those coins, so those addresses are now available to be minted again.

That is the best answer I have right now. I hope that others can clarify.

1 Like
3

I think its safe to say that there is a clear consensus that browsing the network will be free, both for publicly shared files, and (assuming you have the priv-keys) even private files.

1 Like
4

Ok, great to get a clarification on that. It does seems like the most logical approach.

5

@kirkion Can they pay for more storage with a straight swap. I bring a terabyte for the networks use, does that mean I can access a terabyte assuming some sort of parity target for conversion, including things like demonstrated up time?

6

These are all testnet 3 questions. I don’t think we have the hard data to really say.

7

You know @Warren I remember hearing that back when everyone was debating the free v. Paid storage models, but I haven’t heard anything about that in a while.

I believe so, but it hasn’t come up in so long.

By the way @Deadloch I’m just a regular member of the forum, who tends to be up later than most. I wanted to make sure you got a response tonight. But I’m just regurgitating what I’ve read on the forum and discussed with people on the Slack channel.

1 Like
8

Thanks man, I really appreciate your feedback. Maidsafe tends to keep me up at night as well :smiley:

1 Like
9

@Deadloch Above is an older post explaining the difference between account creation vs activation. Pricing for PUT requests (storage), is still unknown. We talked about a Network Utilization model, which you already described above. Once we analyze TestNet3 performance, we can get numbers to answer your questions.

@Warren this was the POR Tokens, which is currently disabled. IMO, direct swap is not enforceable.

The OP triggered a concern because it relates to Safecoin economics. It’s early, but I won’t be able to sleep until I put it out there. This might be a bit long…

Large mining pools are leaving Bitcoin because the fiat price is too low to afford the cost of their operation. People are predictable when you understand their incentives.

The following is PURE speculation…

STAGE 1
SAFE Network Beta launch draws a huge influx of farmers, resulting in massive storage availability. Based on the Network Utilization Model, I predict Safecoin PUT costs will drop dramatically, until capacity fills up. Since there are only a handful of APPS, there won’t be enough demand to keep pace with growing storage capacity. Deduplication will also factor in and further cause the Safecoin PUT costs to drop.

STAGE 2
Due to the incredibly cheap cost of SAFE storage, for life, whale investors may seek to capitalize and buy up all the space then resell it. This means they would acquire Safecoin, resulting in the fiat price rise, buy up the storage space and wait. Because the Network Utilization Model only counts what is “used” a 20% increase in the Safecoin cost would trigger some to cash in.

Safecoin Example
(Assume the launch price is 1Gb = 1SC)

Network Price = 10Gb @ 12 Safecoins
Investor Sold = 10Gb @ 11 Safecoins
Investor Bought = 10Gb @ 10 Safecoins

It’s cheaper to buy from the investor than the Network.

Fiat Translation
(Assume 1SC = $0.05)

Network Price = 10Gb @ $0.60
Investor Sold = 10Gb @ $0.55
Investor Bought = 10Gb @ $0.50

The average Hard Drive is around $0.05 per GB, which is how I priced my example. Now this is where it gets interesting… Imagine if the fiat value of Safecoin also went up. This would lead to a compounding effect in profitability. Let’s say the value of Safecoin went from $0.05 to $0.10.

Fiat Continued
(Assume 1SC = $0.10 BUT the Investor bought storage when 1SC = $0.05)

Network Price = 10Gb @ $1.20
Investor Sold = 10Gb @ $1.10
Investor Bought = 10Gb @ $0.50

The investor doubled his fiat money on the fiat side, AND from the Safecoin side. It makes sense the fiat rises along with Safecoin because they are acquired for the same reason.

STAGE 3
With such a golden egg opportunity, there will be more farmers/investors coming in from mass adoption. I didn’t factor in killer APPS yet. But it points in the same direction. At this stage, we may get some serious consumer demand. The SAFE Network capacity starts to fill up with unique data. Which means, Safecoin PUT costs rise, until equilibrium is temporarily reached. So far everything is great! Consumers are happy, farmers are happy, investors are happy.

STAGE 4
Over the next 20 years, the SAFE Network reduces Safecoin payout. This means farmers, don’t get paid as much as they did before. If the fiat value of Safecoin drops below cost of operation, we will have the same mass exodus effect as Bitcoin. I hope this will not be the case, and the key factor is bandwith/energy costs.

If the farmer is faced with increasing cost of production with decreasing Safecoin payout, they will leave. This event may not happen for awhile. But it’s something that I think we should address in the future. My solution is to maintain the payout rate. As less coins are created, Safecoin PUT costs increase, taking over what the Network was subsidizing.

Won’t that mean PUT storage gets more expensive? Yes, that draws more farmers with even bigger storage and better bandwith. It should be self-sustaining. Or we can try the GET request model. Just food for thought.

1 Like
10

How does an investor resell SAFE storage space to other people? I thought transfer of storage space can only happen through the network mechanisms, which absorbs the SafeCoins rather than transferring them to space owner.

Anyway, if possible, I’d much prefer a pay-on-put system that disables people from hoarding space when the price is cheap. Example:

When you have no storage credit, and you put a file of 1MB on it, you automatically pay 1 SafeCoin (or whatever the smallest unit will be) for let’s say 1 GB credit. 1 MB is subtracted for the current PUT, you have a credit of 999 MB of storage space left. Next time you PUT a file, it will be subtracted from your storage credit and you don’t pay SafeCoin. Only when you run out of storage credit for a full PUT are you charged SafeCoin again. It won’t be possible to get more storage credit above what you get for 1 SafeCoin.

11

If there is one thing I learned, people will find a way when there is money to be made.

My first thought would be to buy blocks of storage with generic accounts. For example: create 100 accounts and buy 1Tb of storage each. Then just sell the login and pw to the new user. People sell max level characters on MMORPG.

Or create your own middleman software/server to facilitate it. They log into your server, buy storage, and your server sends it through to one of your generic accounts.

I think pay-on-put is very ideal. But according to @dirvine it will be done in blocks, so like 10Gb block or something like that.

12

Ok, so it appears that nothing is set in stone as of yet. I’m glad that a network utilization model is being discussed because to me this is the most feasible option.

I agree with you there, direct swap does not seem like it would work, even though it sounds reasonable, there’s no way to guarantee how your node will act in the future. Make the swap, upload your data for life, stop farming. To easy to game.

I completely agree with you there. We are going to see a massive farming push from day one, and data storage costs will be driven to essentially zero (given a network utilization model).

Well thought out. I believe the network will be quite capable in sustaining itself on PUT payment alone in the future.

Why can’t pay-on-put be implemented?

13

I’m sure the devs have a few actual algorithms laid out, but nothing is certain yet. It will be matter of experimentation during test net 3 and BETA.

One important consideration is that the network should aim to maintain a certain amount of free space in case major parts of the network go offline causing big churn events, either due to a natural disaster or a major attack to destroy the reliability of the SAFE network by causing data loss that way.

However, for the sake of efficiency and cost-effectiveness you want to make as much use of the available space as possible, so a balance will need to be found. My wild guess would be something between 50% and 80%.

I’m pretty sure it can be implemented, and probably will be. I’m not 100% clear on what the current plan is. It may actually be exactly this.

1 Like
14

cool, cool, I guess the reason why I couldn’t find the answers is because they haven’t been decided upon yet!

Agree

15

It is pretty much pay on put. You charge up your space account with safecoin and will be told how much space is left based on current prices (so if you leave it and store nothing your space will likely decrease). So each store will incur a cost at that time on the network. The network knows the cost of all puts in real time. I assume space will decrease in cost very fast and there will be an opportunity for your client to pay frequently rather than in larger chunks, thereby keeping your cost per Mb as low as possible.

The network aims for 20% over supply and will flatten out farming over that amount, as farmers reduce or space required increases then the network increases farming rate to get back to 20% above average.

6 Likes
16

I’m so glad to hear this part. “Use it or Lose it” approach. This will discourage hoarding.

1 Like
17

So what about 100 million user? The network still knows cost at all time? And that’s cost in Safecoins I guess? That’s like another mechanism I have to learn than :wink:
(*dreaming about writing a post on “All the economic layers for Safecoin”)

18

Yes, the cost is calculated based on the used:free space ratio of the close group, with additional requests from other (random?) groups for this ratio. Standard (and proven) statistical methods are then used to make sure there is a minimum deviation from the network average in the PUT price. So it’s not like the entire network will be polled, but there will be enough measurements to come extremely close to the factual network average (which is practically impossible to know) using math. This way there are no scaling problems at all.

3 Likes
19

@dirvine Can you point us to information on this?

Also,

Is it true that there are two stages of probability as @kirkion suggests. I was imagining the probability was only as to whether the Safecoin address was un-minted (second stage above), which provides a simple an efficient way of making farming harder as Safecoin are used up.

The other area of probability I’m aware of is related to rank - performance of a node translate into rank, which in turn increases its chance of hosting a given chunk.

So I am aware of two areas. Are there more, such as @Kirkion suggests?

20

Another way to look at it is that the data on the network is stored randomly, but statistically very evenly, over the entire address space. Vaults are created in a similar fashion, i.e., randomly but evenly across the address space. Therefore, if all vaults had an abundance of disk space, all vaults would have about the same amount of data stored on them at any moment.

So, even without doing any polling of other groups, a group of 32 vaults would represent an acceptable reflection of the state of the whole network. I guess an anomaly could occur in a group from time to time, but the churn of the network will prevent even that from lasting too long or causing a problem.

Even if at the moment when my vault attempts to farm a coin, I might be a bit more lucky or less lucky than I ought to be, it still evens out in the overall scheme of things.

1 Like