Initial price for upload

In this topic a question arose what will be the initial price for uploading to the Safe network - my understanding is that it will be set by the team. Is this true?

Currently, the price per GB in the Filecoin network is 20 times cheaper than in the Amazon S3. If we are going to set a starting price then how many times cheaper do we want it to be than Amazon S3? 10 times? 100 times? 1000 times?


This is not an effective approach to pricing strategy. One must layout the relative benefits provided by different solutions in order to understand a price that makes sense. It’s not just about being “cheap”. In fact, sometimes having a price that is much cheaper than the marketed value of the product can be damaging.

Otherwise put, if you tell someone you’re going to sell them a (vehicle commensurate to a) Bentley for $10K they are going to believe you’re lying about the quality of the car. Similarly, if you tell people you’re going to store their data forever for 1000x less than AWS, the price will make people doubt the quality of the product.

Effective pricing strategy is rooted in a firm understanding of behavioral economics and human psychology.


I reckon if a price were to be set (which will probably be implicit rather than explicit, see below) a decent starting point would be to aim for roughly matching S3 pricing. Amazon has a sustainable (profitable) business at that price so why reinvent the wheel. Maybe slightly cheaper than S3 so we can get some marketing advantage by the efficiency and cost effectiveness, but not 10x cheaper, maybe 2x at most.

Here’s what I imagine might happen and why the initial price won’t be ‘set’ by anyone.

  • MaidSafe start a testnet internally, hoping maybe it will become ‘the one’.
  • They send the ICO etc safecoins from the section wallet(s) into a separate wallet so they can pay those stakeholders when the time comes.
  • MaidSafe is able to obtain some safecoins using the standard burn (or sign) process for the maidsafecoin they already own.
  • At this point there’s nodes with some storage space but very little data on the network (probably only the safecoin data). The safecoin algorithm will calculate some storecost for this state depending on the specific algorithm. I don’t know what this storecost will be (in safecoin), or what the exchange rate will be (to give storecost in dollars and thus comparable with S3).
  • ↑ Maybe this is the point in time where your question is important? ↑ MaidSafe could look at the size of the initial data on the network, then work out how much spare storage is needed to give the desired starting price, and supply only that much space.
  • MaidSafe use some of their safecoins and do some tests with uploading and downloading (they will also be farming some new safecoins from the test activity itself).
  • This test activity will also cause space to be consumed so the storecost will be algorithmically adjusted.
  • Maybe some more nodes will be joined to the network further affecting storecost.
  • More tests of uploading / downloading / joining. More algorithmic calculations for a new storecost.
  • The testnet is working reliably and is now made open to the community members.
  • ↑ Is this the point where your question is important? ↑ MaidSafe could upload a certain amount of test data and supply a certain amount of storage which would result in a predetermined storecost (in safecoin), or they could adjust it based on the exchange rate at the time for a predetermined storecost (in dollars).
  • The community does a bunch of uploading and consumes space on the nodes. Storecost will probably become expensive (in safecoin terms) fairly quickly to keep uploads at a manageable rate.
  • The community will also start a bunch of nodes, making more space available. But this is slower than the uploading so new space does not compensate for the high demand of uploading, and storecost overall continues to rise (in safecoin terms).
  • The traders on exchanges respond to the news of the new testnet very positively, “it might be The One”, and a speculative bubble starts. Storecost quickly becomes even more expensive (in dollar terms).
  • Because of the double-whammy high storecost in both safecoin and dollar terms, people rapidly reduce their rate of uploading; it’s just too expensive and makes no sense to upload at those prices.
  • The need for new nodes is greatly reduced because uploads are greatly reduced. Almost all new farmers are being put in a queue because the current farmers can already handle the very few expensive new uploads arriving to the network without needing new resources.
  • The network is seen as being too expensive to use and is claimed to be dying. Speculators start selling safecoin and the exchange rate plummets, adding to the impression of a dying network [1].
  • Eventually the dollar price for uploading becomes attractive again because of the exchange rate crash. Uploaders slowly begin to increase their traffic (causing adjustments to storecost).
  • Slowly the existing nodes become insufficient to store the new data so more farmers are required and the queue of new farmers are slowly bought into the fold (causing adjustments to storecost).
  • The network grinds onward. Storecost fluctuates, both in safecoin terms and dollar terms, but less extreme over time. Speculators come and go. Nodes come and go. Developer interest waxes and wanes. The extrapolation scenario is starting to become tentative but as time goes on we learn more and the network becomes stronger.

[1] Maybe there’ll be a way to make storecost cheaper in dollar terms besides having the exchange rate drop? One way to achieve this is to use the high speculative price to generate interest in farming and bring on more resources so the storecost is cheap in safecoin terms (but is reasonable in dollar terms due to the high exchange rate). Sounds maybe ok but this strategy of taking on new resources might introduce a risk those new resources may depart when the speculative price is not so speculative any more.


Pay once; available for ever, is different and how the value of that is resolved not obvious to me.

I wonder there are two risks…
One risk of under pricing… no point under selling something valuable.
Another risk of harmonics in price similar to pump and dump volatility. Pricing high perhaps helps dampen the risks.

Why not price equal to Amazon et al and note it’s a one off payment etc etc.

Defining relative to others perhaps an error???


Not as simple to answer. It should work like this. (remember the network sets the price).

At genesis, we have a price of 1 safecoin per Put (for example, it does not really matter). As the network grows (as we dd nodes) the price gets super cheap, then as resources are required it goes back up again and so on (super high level).

So you could say we set the price of the first put, but it’s not setting the price per se’ the network will rapidly change that.


Thank you for the answer David! I’m sorry I didn’t clarify better.

The initial price is clearly irrelevant. What matters is how much the network will pay extra through inflation - this became clear in the topic I am quoting - filecoin pays extra and subsidizes with inflation, what will be our initial inflation? Obviously, there must be one to reach 4.3 billion safecoins…

The Safe network has two prices for storing data - what the user pays to the network and what the network pays to the farmer. My question is what the network will pays the farmer initially.

1 Like

Ah I see. We have no final decision there yet, but likely it will pay out something like 2X the reward or similar. We want all coins in circulation quickly, but we cannot yet define clearly “quickly”. When they are ll out there then they are in the hands of people and not a network wallet, so even bad sections cannot mint new coins. etc.


This is something I’ve always wondered and have failed to address. When I first started reading here I just assumed Safe would adjust its fluctuating token cost based on dollars. Maybe that’s completely wrong but the general population will look at it the same way and want to compare apples to apples.

That’s why I would think, oh maybe there will be some network oracle that has a price feed that informs the store cost. It is a solution but likely one that causes more problems than it solves?


I still live in a world where there will literally never be all coins in the circulations. I thought we will get in 3-7 years in the sustainable point when there will be around 3.7B (85%) coins in circulation. Have I missed something important?


That sounds like quite fast to me :slight_smile:
With x2 farming reward it will take several years I guess.

Anyway here were put lot of thoughts: RFC 57: Safecoin Revised

1 Like

This depends on how quickly the Network reacts to shifts in Safecoin market price as well as Network supply and demand.

Per David’s response:

While the mechanism that will deliver this high degree of sensitivity is TBD, the target for quick responsiveness seems clear.

Yes, the common POV shared on this forum over the years has been that it would take years before a volume bordering on max supply hits the market. I think a rapid inflation rate would be a dangerous proposition.


In my head I still imagine some kind of orderbook auction type with free floating for farmers to join or leave the network. A free-float market where there is no need for an algorithm. Where a user for example sets a high/low cost limit for storage/GB and then the client random buys storage, 100MB each time, within choosen range, until either low/high limit reached and then the client would be asked to choose new limits.

To set a low/high limit would allow for price fluctuations, the data to be scattered across various nodes and to reject evil actors from setting price to 0 for evil purposes of gathering data or destroying the network.

Don’t know if a free floating solution is only a dream or would possible. The hypothetical model is very basic and excludes probably alot of important scenarios, for example but excluded, not only choosing price but also reliability. latency and such.

Sia works on this principle. I was a farmer there for 1-2 years, I even give my storage for free. Unfortunately, their software was so buggy that it constantly broke when there was a power outage (just then they were building a subway next to my apartment and the power went out several times a month)


I’ve been thinking about this. Even if all money were minted, wouldn’t a bad section always have something to steal? For instance, it could steal the section wallet since the section disburses payments to participant nodes only after they relocate to another section. Is the idea that if the bad section does that then it’ll be caught because those who didn’t get paid would report it? Wouldn’t it be possible to likewise catch a section that misuses newly minted coins because it wouldn’t be able to pay nodes that did work?


We have an option where with DBC for instance a client pays directly the nodes the data will be stored on. It changes a few things, Adults IP will be exposed etc. but also the network only transmits the payments, they could be transmitted via a usb drive etc.

This stuff is really a focus for me right now, have the most secure network we can but give it no authority to mutate any (humans) data or data that can harm humans.

It’s all just me and a few others pondering right now, but all this is well within reach if DBC works well. The only gap left is the section wallets for the initial issue of coins and I am positive we can come up with some neat solution there to complete the circle.


Very interesting indeed. Thanks for sharing @dirvine

1 Like

PS: with that design change, you’d have to find another mechanism to ensure the nodes do keep and serve data entrusted to them. I really liked the ingenious simplicity of paying the node only after it has fulfilled its duties to the section and relocated to another. I’m sure you and the team have considerations like these well under control though. More power to your elbows


My current thinking is pay as you go. So a node who takes and runs is doing so for a few chunks, if he stays he gets paid more. If he leaves then others will get paid for all the new chunks. I recon it’s fair and balanced. So no wait penalty and we can show nodes wealth increase gradually and not in steps. So from a UX it might lead to nodes being more sticky.

As I say all thoughts right now, but with the tools we have right now these things are all possible. I see much more as we release. Just in case anyone reading thinks this slows launch, it does not, it’s typical last minute Engineering tweaks as we launch. I know some folk get scared when new angles appear, but this is absolutely to be expected. It’s using tools we know are proven to try a few improvements. In saying that DBC is not there, but if it was these considerations become real.


You guys have made the right technical call over and over. So I’m sure you’re having these considerations for very good reason. Thanks for the reassurance nevertheless, and I think having an imperfect version out to build upon would be a tremendous starting point.

Probably my bias talking and I stand to be corrected, but I suspect there’ll need to be some mechanism whether wait to get paid or deposit that can be forfeited otherwise nodes may not behave when it comes to serving back the data especially in the long term and with old/unpopular data.

EDIT: thinking more about it, it could work if it’s very difficult to get into position to start earning in the first place. It that case, the node might not want to go through the hassle. So that could serve as the penalty.