Why sigmoid? + Safecoin crowdsale generation

Curve:

I’ve heard a few times that the safecoin generation curve will be sigmoid in nature. Why was this chosen? Safecoin generation is supposed to first and foremost cover your usage while you browse. At least from my understanding. SAFE is also supposed to help out the “less privileged”, help bring access to those that couldn’t otherwise have a means. This may also mean they don’t have much to offer. Sigmoid seems to punish those people. It seems that a log curve would be more suited, topping out at 20% above network average.

Crowdsale coins:

Where are the 10% of crowdsale coins to trade in for out tokens coming from? Generated off the bat by the network? Taken from MAIDSAFE’s farms over time? If its the first option, how is that ability going to be taken away? Will they be generated on a “private” maidsafe network before it goes live?
Just wondering the specifics on this.

1 Like

Browsing cost == 0 and always will.

Sigmoid curve will allow very small nodes to get rewarded but encourage use up to average required storage. Otherwise we have all tiny vaults and that will be of no use to anyone. So sigmoid allows even small vaults to be rewarded but also encourages greater provision. A balance of reward verses requirement.

Most likely coded in, but several ways to achieve this, very low on our priority list at the moment to finalise it. All tokens will go to a single account, then all transfers will be done and account deleted. We can for instance make this read only for all so there is a view of it, but as I say it’s almost a zero considerations at the moment. Getting launched now is a 100% commitment for us in house and we are thinking of little else :slight_smile:

After launch I can happily spend days/weeks on each element of the network but I did learn last year of the cost to development of this so a mix of really good documents, more formal proofs and then have the convo’s as almost nothing is independent and there are effects of every single algorithm.

8 Likes

I meant light usage (posting a video in exchange for using my HD space) Just trying to get across that the main point is quid quo pro.

Touche. Allow People to get a usage out of it, but promote giving more. That makes sense.

Thanks for answering. My goal isn’t to take you guys out of the code. Was just something I though about and was curious if it had been planned out.

3 Likes

I was avoiding this discussion, but since it’s out there…

Mobile farmers (8GB vaults) are “unable” to store up to the Network Average, unless they start farming at the same time as PC farmers (100GB vaults) on launch day.

If the sweet spot is 20% above the Network Average, then vaults are “expected” to get bigger and bigger, chasing max reward. Eventually, Server farmers (+TB vaults + Very High bandwith) will take the lead over PC farmers.

The one thing slowing the inevitable is the ability to fill a vault fast enough per session, which is dependent on the Chunk Collection Rate. I suspect dedicated vaults will remain online long enough to collect more chunks when casuals go offline.

Why Vaults Get Bigger Over Time

  • Larger vaults have more GET opportunities compared to smaller ones. This means the incentive to provide more storage is already there, even without the higher reward ratio.
  • Larger vaults take longer to fill up if the Chunk Collection Rate is low. This means farmers will make multiple vaults to increase their Chunk Collection Rate. But still aim for 20% Above Network Average.
  • Multiple Small vaults have the advantage of speed but are restrained by resource costs. It is more economical to run an Above Average Vault than a Below Average Vault. Again the incentive is for vaults to get bigger.

Counter Point - Why Vaults Remain Small

  • If new vaults (0 chunks stored) affect the Network Average, then maybe it will not increase as I expect. It may even reduce the Network Average, and therefore we will end up with smaller vaults.

I’m not saying this is guaranteed to happen nor will farmers behave the same way. Regardless, these are my concerns when assessing the incentive structure, until TestNet3 proves otherwise. I hope I’m wrong. If mobile farming becomes viable long term, they can accelerate mass adoption.

2 Likes

I think you’ve missed one crucial factor in this, which works to keep vaults smaller by disadvantaging farmers who “chase” slightly higher farming rates.

The network will, even at lower farming rates, always favour those for whom farming has little or no extra cost - meaning anyone who just installs the vault and forgets about it, and just farms when they happen to have their device on for normal use.

It favours them in the sense that all their farming reward is profit.

Anyone “chasing” is incurring extra costs that must be exceeded by their extra rewards. Cost of energy, cost of bigger storage.

I think the effect of this is to put a ceiling on vault size, with a balance point (network average) arising out of the two types of farmer (and those in the middle!)

5 Likes

While I don’t plan to be a “commercial” farmer (trying to do it for a living) I do have a rather large media server, ~13tb that may put me in this type of category based simply on size. The vast majority of people don’t have anywhere near that… I’m just a geek and like to play with raids. Is there / will there be a way to slowly increases my vault size as time goes on? Or will I end up starting new vaults if I want to offer more space? I don’t want to just lump that space on the network all at once.

That should naturally happen. When we say cap the farming rate at 20% over average what it means is you still farm but not at a continually higher rate. To maximise farming then it’s best to run 2 vaults etc. I think it will be a matter of measuring and I am very sure apps will be created to try and maximise farming etc. it’s a human thing to get max value for min effort and we work with that. IT’s not bad it’s natural so should be no problem.

Should be no problem when beta is running. If you so it slowly I suppose you will earn the safecoin to store. Certainly be good as a test in early days as well (when no safecoin payment is needed/testnet3:-).

1 Like

That sounds contrary to what the graph shows. There is an obvious increase between (Network Average) ~> (20% Above Network Average). Let’s run the math.

Assume the Network Average is 100GB.

If I have 120GB stored, I will earn significantly more Safecoin compared to an 80GB vault. Yes, I will have more GET request to service, but reward is not linear. In terms of (Resource VS Reward) it looks something like this…

Assume the Network Reward is 10 Safecoins.

120GB Vault = 90% Reward Ratio, 9 Safecoin, (13GB provided per Safecoin)
100GB Vault = 30% Reward Ratio, 3 Safecoin, (33GB provided per Safecoin)
80GB Vault = 10% Reward Ratio, 1 Safecoin, (80GB provided per Safecoin)

These numbers are not meant to be exact calculation. It’s only meant to show the disparity of resources needed to earn 1 Safecoin due to the Sigmoid curve. It looks like the Network favors Above Average Vaults. This is why vaults will keep getting bigger until, as you said, the costs exceed the reward.


Here’s what I’m trying to address.

If the goal is to make mobile farming viable, and they cannot reach the Network Average, it will discourage most of them. Will people run mobile farms anyway? Probably. But the amount of resources needed per Safecoin will be much higher than the Average or Above Average PC Vault.

If I’ve completely misunderstood the way the Sigmoid Curve works, then I apologize for wasting everyone’s time.

MaidSafe is doing an incredible job and my intention is to help, not distract. Having said that, I’ve raised my concern and will leave it there. They have far more important things to work on. :smile:

There are a couple of things to consider here. Ultimately relatively few vaults will be farming at above the network average, and those that do will have earned the position of increased safecoin by staying continuously functional for a long time. Just turning on a vault with a lot of storage space doesn’t mean it will be filled up very quickly. It will be filled as more data is stored, and as other vaults in the address range (or otherwise as Data Managers determine) go off line, requiring their chunks be duplicated.

That’s all going to add up to more consistent online time and probably UPS and solid data connection, all of which cost the farmer resources to set up and maintain. Such “archives nodes” should be valuable to the network, so they should be rewarded a bit better. The curve allows for a bit better reward for this, but shouldn’t suck all the safecoin out of the ground for others, either. Everybody will have chunks, chunks will get used, and delivery rewarded.

I just had a realization about this. Farming is done on a per successful request basis. Vaults are also “wiped” on a reboot. So it’s not just space offered, it’s space filled.

If the curve follows that logic - space filled, not space offered - that alleviates a majority of my concerns. Someone can’t just dump a huge amount of space on the network and profit, it has to be filled. I guess I knew that but it hadn’t clicked yet. Maybe it helps someone else.

2 Likes

Just a thought here but isn’t processor speed and bandwidth also a viable resource? And unless I’m mistaken some nodes will be specialized not for storage but to do calculations and track other nodes and do various tasks for the network. (I’m forgetting my terminology here, forgive me. I think they were called arbitors or something.) Anyway the point I’m making is while mobiles might lack data storage they still excel in processor power so couldn’t they devote that as a resource in exchange for safecoin? Wouldn’t this translate into additional and much needed bandwidth for the network? As near as I can figure we have multiple resources that need to be fullfilled.

  1. Data storage space.
  2. Relay points in a mesh network.
  3. Processor power to computer and transmit data.

So what if mobiles don’t have a lot of data space? They’re excellent for acting as nodes in a mesh network and taking the processor load of desktop computers, which people willl likely be using to play games or whatever, and doing the heavy duty computation for the network. SAFE is about proof of resource. It doesn’t say that resource must be data.

Build an app or even write something into the core to determine if a node is x amount close to another node and you can then give mobiles credit for acting as geographic locations in a mesh network. Say I have a cell phone and I’m the only node linking 2 other nodes, that’s worth something. if I wander out of range that would be bad for the network so incentivizing me (reletive to the next guy) to stay put would be a good idea. Conversely if I have a cell phone in the middle of a crowd I’m not going to be paid as much safecoin because there are other nodes all around me. This also encourages users to disperse the network to outlaying areas. While you get more connectivity in a city because there are more people there to connect to you get paid more if you move further out (and thuse connect new people to the network, say those living in rural areas that may not even have internet at all).

@dirvine Just wondering but could (or is it currently being) something like this actually be done? Or am I totally misunderstanding the network here?

1 Like

The problem with this is the built in assumptions.

Suppose the farmer is a US Citizen, with a bandwidth cap? Storage is abundant but bandwidth is not. So why do we keep talking about storage and vaults when we think about the Safecoin instead of talking about what could have the greatest impact on Safecoin which is bandwidth?

In rich countries, where electricity, bandwidth, computation and storage are abundant, then SAFE Network would work fine. Bandwidth is not abundant. Electricity also is not yet abundant even if at some point nodes could be solar powered.

I haven’t finished the thread on safecoin valuation where you seem to keep bringing this point up, but I guess it never got answered to your liking there, so I’ll give it a shot.

You’re comparing apples and oranges. You say that someone who supplies bandwidth in an area where there are bandwidth caps should be paid by the safe network. This is just not the case. Safe is a destributed secure data management software.That’s its goal. In order to do that it must move data. That’s a side effect. People dont get paid for side effects. They get paid for primary goals.

If you’re a farmer with a cap, get a new ISP. That’s where you could come in with your solar balloon WIFI. Be an ISP without a cap that takes safecoin in payment from your newfound farming friends.

You’re cant fix an imposed hardware “problem” with a software solution who’s goal is to fix an unrelated issue.

I don’t think it’s apples to oranges. I see all of these components as connected. I’m seeing it in a holistic manner rather than seeing SAFE Network as just secure data storage.

If you want SAFE Network to be used you have to figure out how to allow people to access it. SAFE Network can be throttled by ISPs who restrict access.

Suppose it were China? In China the ISPs can just block access to SAFE Network and that would be the end of it. You’re right that service providers will have to accept Safecoins in exchange for access and become mobile ISPs.

But how to do this in a way where the bandwidth isn’t unbearably slow? If it’s got bad bandwidth then it has no advantages over just uploading your data to a centralized cloud. Ultimately you have to be able to upload or download your data at a reasonable rate for SAFE Network to have good utility.

The key seems to be Safecoin. Safecoin can provide the incentive for people to bring bandwidth along with storage but at this point in time no one can know how that will influence the distribution of Safecoin. It could be bandwidth is the most scarce globally which could mean most Safecoins will go to bandwidth providers.

They can’t. SAFE traffick consists of just UDP packets with encrypted payload between random IP addresses.

They can determine that users are running SAFE Network and then reduce upload speeds to those users. They can put in place caps to reduce total bandwidth for all users who are not business users.

How can they do that? if:

1 Like

I’ve once heard the TOR-protocol gives away that it’s TOR by some “handshake”. The Devs of TOR made it in a way that they used a different handshake and make it look as if the protocol was just a normal website doing some browsing. So ISP’s and Gov’s may use deep packet inspection to find out that you’re most likely using SAFEnet. But it can be disguised as well. Maybe the Devs already did.

What if a solar flare hits the earth and takes out the internet. We need to pay safe farmers for shielding their computer in lead. Lead is expensive, we should pay them more for lead shielded computers…

How far do you want to go with this “what if” game? The beauty of a fluid open protocol and service is we can adapt to issues as they come. If its an issue, we’ll fix it. Let’s not drive ourselves crazy trying to fix something that most likely isn’t an issue in the long run.

Deep packet inspection is just one technique but there are others. Unless you specifically go out of your way to mask your traffic the traffic patterns emit a signature which can reveal exactly what you’re doing to the ISP.