Early demand, and effect on storage costs, after launch

Features
1

That is incorrect. The VERY FIRST coin ever in the live system will only buy 1MB, but @dirvine has said that cost per MB reduces each coin after that till the network balances out. The reduction is exponential so the balance point is reached in only a handful of coins.

The very first 500 MAID could even buy more than 500GB, maybe less and initial balance point could be around one coin per GB, but we will not know exactly till the system is live. BUT not 1 MAID per MB

EDIT: and the above is incorrect according to the RFC which shows the cost of first PUT is approximately 0
And subsequent to that it remains at approximately zero and farming rate is approximately zero until the network starts to fill up. So 500 MAID will buy you a whole lot of storage and make 500MB as insignificant.

2 Likes
2

If that is true the the farmers will be screwed. Especially the early ones.

3

Some people might want to. If I want to backup my data to SAFE I might buy enough to allow me that and spend it all. Not everyone will, of course.
In your theory the capacity will be cheapest early on, so anyone looking to use it should do so earlier rather than later.

My Pi 2 B with 8TB HDD can replace literally hundreds of such vaults. If my setup won’t be economical, yours will be economically catastrophic.

4

Read the whole bit rather than scanning. That is for the network at startup.

1 Like
5

So when a few coin will buy enough storage to max out the upload bandwidth for most they are going to spend their hundreds/thousands of coins for storage that cannot be used on day one.

The theory is the RFC, so if you call the RFC theory then its theory, but better theory than your made up figure of 1MB cost 1 coin.

Also from memory you can only buy storage when its required, you cannot pre-purchase more than one coin worth. The idea is to have the client buy the resource as required. You set the number of coins it can do this for. Thus if your upload is 21 GB per day then thats all

6

I got my “theory” of 1 coin = 1 MB from a MaidSafe white paper.

I can upload quite a lot per second, there are thousands of people who can upload at more than 10MB/s.

Edit: I saw your comment about the RFC. I’ll take a look, but I won’t post future comments here because it’s off topic.

1 Like
7

Yea the whole paper said First Put will cost 1 coin. RFC changed that

EDIT: did the whitepaper put a figure on even the 1st PUT?

EDIT2: Yes off topic If I cannot find a topic to put these then I also will not comment more here

9

If the almost free price of capacity will last only 2 hours, why are you even mentioning it? It’s pointless.

When you upload the first 1MB the network already “starts to fill up”. But insignificantly. And that should only impact the farming rate, it can’t impact a single user’s cost because his Safecoin is already in his possession.

10

Because it negates your 500 coin for 500 MB on day one.

The very low price increases slowly because it takes time for data to fill up the vaults. The price is determined by the number of clients and the farming rate. It is not and probably ever get close to 1 coin per MB cost. If it ever gets close to that then there is extremely little free space

11

So if I have 1000 MAID and convert them to Safecoin, why would I care about the network farming rate? I just keep uploading until I’m done.
If the network farming rate changes fast enough to matter to a single upload, it will be a joke.

12

Well the network worries about it, even if you don’t and its the network that uses the FR to calc the put cost which affects you.

Obviously, but no one suggested that :smile:

But it will change quickly for the first few uploads but less so from there. Seeing as the last test network will be the live network then no one will notice when network is live.

NOTE I said LAST test. No one knows its the last test network until its announced that the network is going live.

13

So will people upload for free to the last test network?

14

Well we will see. That is what is suggested. But they will not know if their data will survive until its announced its live

Test network will see a new account given a coins worth of uploads. Then people have to earn test safecoin by farming. Then when its live that version of testsafecoin becomes safecoin and the token->safecoin will commence.

But that is only a possible plan put forth during the last sprint by David

As this going off topic for this thread, maybe we had better start a new thread or stop :smile:

15

Maybe @dirvine can jump in after he had a good sleep and some coffee ;-). We’re having all these debates about how much it will cost to PUT data and how long it takes for the network to balance out. Is the balancing indeed exponential? How fast will the network adjust when the price of Safecoin rises 100% in 3 days and all other factors (supply/demand) stay the same? Will the network “pick up” the higher prices in 24H or even in 4H??

1 Like
16

The network will not care about the price of safecoin, the farmers will though and they will either add a resource or remove it depending on how they “value” it economically. Many farmers will not care (like Skype users did not) and many will focus on economics IMHO.

The network will respond immediately to any resource added or removed from the network. Well immediately on removal of a resource and gradually on addition of a resource (at first churn event) so almost immediately.

So a farmer may get 1 SC per hour and be OK then way too many farmers are offering service and it will be 1 SC per 10 hours (say). Alternatively if farmers are reducing then it will be a SC per 30 mins then 15 mins etc. so the network only balances out resources against demand and no more. The price does not matter really, but indirectly it affects the farmers in so much that it is the farmer who can say “this SC payment is too little or more than enough etc.”

It is worth remembering we are trying to make farming at minimum shared “unused” resource so the economic cost of farming on a commodity computer (your computer you use for working) should be as close to zero as possible, then any payment may seem like a plus.

This is why the space should be the cheapest possible price to purchase. Even with the crowdsale % this will be fine, it’s not a 10% tax as the crowdsale folks will sell SC eventually and they become part of the recycle machine. They should merge back into the ecosystem over time. They definitely should be rewarded as without them this is all for nought, same with MaidSafe original investors. Worth noting MaidSafe original investors does not mean MaidSafe the company or staff, it gets nothing from these returns, it’s like the crowdsale folks they get the return for putting up the capital.

The tax part if you like will be app dev / PtP, code dev rewards and these are test values at the moment. Maybe none will work properly and get removed or maybe the method of reward will change over time, but the intent surely has to be good. Reward actual provable goods and services. So really I feel if we can get a system that at least works in the digital world first then we can see if it can extend to a wider catchment area.

For instance say we get the core code to it’s absolute minimum number of types etc. so close to perfect (ha) then core dev payments may near vanish as there is less maintenance required (fast written (just works) code is messy and needs lots of maintenance, i.e. Google replace 50% code per month, we can do better).

Then maybe core dev payments will look more like app dev and pay a % for new features that go to zero over time for each new feature (like compute) (they cannot stay at a flat rate forever or we are in copyright type market which has failed).

Early Saturday morning (well for me, last night was a late one) so please forgive typos as usual :slight_smile:

12 Likes
17

@dirvine Thanks David, you’ve explained the farming side, but I think the PUT cost is critical and is going to be affected by the fiat price of Safecoin.

I can see how if the PUT cost remained static, more people would just farm for coin, and that this could act as a stabiliser for fiat SC price.

But I believe that PUT costs can also change, so can you say good the two happen and how they intersect?

18

I agree with this, so we need to think about what the constraints are on “cheapest possible price”. The most obvious ones being 1) the SafeCoin cap of 2^32 and 2) harmful effects of an excessive rate of inflation. I think that the PUT cost algorithm should thus take into account 1) the amount of coins in circulation and 2) the ratio of coins issued to farmers and coins absorbed by the network.

I think it would be great if we could make statements about the PUT cost algorithm like this:

When X% of coins are in circulation, the network aims to absorb Y coins from uploaders for every Z coins issued to farmers.

Examples:

  • When 30% of coins are in circulation, the network aims to absorb 10 coins from uploaders for every 15 coins it issues to farmers.
  • When 90% of coins are in circulation, the network aims to absorb 10 coins from uploaders for every 10 coins it issues to farmers. (i.e. it’d try to keep 10% as a reserve buffer, the network is in equilibrium at 90%)

So the algorithm first computes a target of how many coins should’ve been absorbed up until that moment, and if reality doesn’t match, it increases or decreases the PUT price accordingly.

19

It won’t, the ClientManagers know how much space a safecoin will buy, so at the time of using a safecoin it may be 100Gb or whatever and as you use it you see it gets more space as the price of put drops (or should).

I hope this is what you mean. So a put is always a safecoin but could represent millions or billions of chunks. The network dynamically calculates on every store how much it is.

6 Likes
20

The price on fiat shouldn’t be a consideration with how the network rewards farmers.

It makes sense to a person but it doesn’t to an economy. Bitcoin is still at 300 and people still are mining, and also people were still mining when bitcoin had its greatest reward and worthless bits. additionally, there was no mountain of economic theory to reason with why 21m; people make it work. as we/they did there, so we/us will here

Safecoin affords a difference in that it will always buy a PUT with some data behind it. And I think in this the fiat denomination should not at all play a factor in how the network determines distribution and cost of bytes/safecoin.

2 Likes
21

@dirvine I realise the PUT cost changes, but I haven’t grasped how this works and how the various metrics behave (fiat price/user demand v PUT cost v farming rewards).

What I think I know:

  • Farming rewards depend on the amount of free space (too little reward increases, too much reward decreases). Does this effect override variations in the chance of a farming attempt succeeding (since this depends on the number of unallocated Safecoin)? It seems clear how this can work regardless of fiat price.
  • PUT costs are (as you explain) calculated on the fly - by changing the amount of storage delivered in exchange for 1 SC at the time of purchase. (Hehe, I just typed that as “putchase”! Now my copyright ;-))

What I don’t understand is how the PUT cost is calculated and how this would be affected by shifts in fiat price, since demand for PUTs will vary due to factors other than fiat price (whereas farming provision should be closely tied to it). I guess this is not finalised, but can you explain a little?

I can imagine you could vary the PUT price more slowly when the network is out of balance (too much or too little spare capacity). When out of balance, the “priority” adjustment will be farming rewards, but once the farming provision is in balance, the PUT price adjusts more quickly.

Is that along the right lines?

1 Like