What difference does XOR space matter, we are talking about one person or group controlling a large % of the network, the only way they can feasibly do that is if they own a datacenter or something, if theres no way to protect against datacenters then it wouldnt take much to wreck the network.
But maybe you have a point, if its not a datacenter then maybe its a botnet, or maybe a bunch of small farmers form a pool like bitcoin does where theres a monetary incentive for everyone to join under one banner, thats threatened bitcoin with a 51% several times now.
I dont get why the network topology could not be made visible, if every node in the network is known then its IPs are known as well as its share in the network, then it should become apparent where centralization is happening and steps could be taken to minimize its impact by blacklisting further expansion from the ip range. And by that i dont mean allow any one datacenter to become a signficiant percentage, like for say even 0.1%, as if someone were determined enough they could then find a bunch of other datacenters and disrupt the network in short time.
As for botnets maidsafe sounds like its going to be too much of a resource hog for its victims to not notice. I dont know if pools will be a thing but if the network is handing out a singular reward every n minutes then you’ll end up with a bitcoin centralization all over again.
I really hope the maidsafe devs have something planned, it only takes watching whats happening to bitcoin to see the very real threat centralization poses.
In my head, if I was gonna try to answer this thread’s main question in like one sentence:
Centralisation will be less of a problem for SAFE because everyone already has a computer / phone / etc, with an internet connection, And farming will be super easy and accessible (compared to btc farming… Yeeesh…), so everyone can just jump in with what they already have at no marginal cost.
But big data center guys will be spending alot, so their returns won’t be as big (cuz u gotta subtract that cost), compared to everyone else (who just uses the computers they already have).
I don’t think it’s profitable to run millions of tiny vaults due to the non-optional routing burden that comes with a position in XOR space. Since all chunks are routed through the DHT, a chunk takes X amount of hops from the serving vault to client. If X is for example 5 (probably higher in practice), then serving a 1 MB chunk causes 4 vaults to route (download and then upload) that 1 MB while only 1 vault is farming (upload only). So in this scenario an average sized vault would be using about 80% of it’s upload bandwidth for routing and only 20% for actual farming (this is simplified for the sake of argument).
Since routing burden is independent of a vault’s size, a tiny vault will have to route just as much as the previously mentioned average sized vault. If that routing burden is 1 MB/s, creating a million tiny vaults would force you to instantly start routing at 1 TB/s, and you’re not compensated for this routing burden.
The routing burden acts as a pretty good first line of defence against sybill attacks. Many smaller vaults would actually be good the network, but they’ll likely never be profitable for their owners. For a vault to be profitable, it has to be big enough so that it’s farming income compensates both for the routing burden and farming burden together, and that routing burden is “fixed”, while farming income rises with vault size.
The point is that farmers are paid in proportion to the resource that is contributed to, and found to be useful by, the network. I don’t view it as punishing the smaller, individual farmers. They will be paid proportional to the resources provide to the network in a useful way.
The network will be useful, ultimately, to the degree that A LOT of individuals are using it. These individuals will also be contributing to it. If huge farming interests predominate the network, they won’t be getting paid much, because the value of the network is directly proportional to the amount that it is used by people.
A server farm could fork the network and run a huge network with hundreds of thousands of nodes, and create automatic clients to store and retrieve data, and it would earn all the safecoin, but that’s like being a bizillionaire with unicorncoins created in ones own mind.
Large farmer will actually have to apply good husbandry of real users, or the endeavor is profitless. Such huge server-farmers are an attack vector, but not a concern from a profit hunger concern, I think. As long as farmers are paid proportionally to their contribution, and moderate vaults are rewarded preferably to larger ones, mega farmers will only be shooting themselves in the foot if they get too big a portion of the network. If you cut real users out of the money, you destroy the value of the network, earn less safecoin, and the safecoins earned become less valuable.
And remember, the marginal utility for the home farmer using resources they are not paying for solely to farm is much, much greater than that of a large farmer maintaining all that infrastructure just to make a profit.
As I say, there is definitely a vector for malicious attack. But that’s another matter.
So in short it’s a trade off. More vaults means more chunks and more GETs so your space fills up faster but more vaults cost more resources and more bandwidth.
Because that would be a security risk. IPs are private information and extremely sensitive at that. They can be used to identify you and would pose a grave security risk to the network and it’s user’s anonymity if they were exposed. Just look at what’s happening with bittorrent and how the feds are tracking pirates by tracking their IPs.
This is a false assumption @whiteoutmashups I know plenty of people that don’t own a computer or smartphone and even of those that do a lot don’t have internet. And that’s in the first world. Poverty can be a bitch. Or in some cases it’s just geography, I have a friend who rents a cabin out where there is no internet just cell service. And I’ve lived places with bad or intermittent internet simply because of snow on the mountains or interference from trees. Things happen.
Have there been any recent discussions or solutions proposed for this problem? I see the last post is over a year old.
It does seem like a bit of a scary attack vector for the network, especially in the early days, that a large entity with massive amounts of bandwidth and storage capacity could destroy an enormous amount of random data on the network if they are contributing a majority of the space.
If they could do this and take a huge short position on the price of safecoin, it could be very profitable for them and disastrous for the rest of us right?
I guess maybe the best defense is that anyone with a really large position in Safecoin should also contribute huge amounts of resources in the early days of the network in order to prevent this… but they wouldn’t stand a chance against a Google or large government agency.
Yes there has been a lot of discussions on this and its not as bad as you seem to think it is.
One aspect is nodes (vaults) are (will be) require to progress through an aging process and so massive disruptions will be costing the attacker real money to run centralised vaults in order to attack.
Also with the number of copies kept of each chunk the attacker has to have a lot more than 50% of the total nodes/vaults. 8 copies of each chunk in aged vaults, so the attacker needs to have more than just a majority of nodes for a long time to be able to cause data loss. Think in terms of 2 to 3 times the number of good nodes.
Ah thanks, I didn’t realize there’s an aging process for the vaults. I’ll have to read more about that. Is there another more recent thread you can point me to that covers this?
Try searching for “data chains” or “aging” or “centralization”
I cannot at the moment.
But you are right that there is a degree of vulnerably while the network is small and even aging will not be able to stop it. In the last testnet there was a invite system to help slow down the attack that someone did on the previous testnet. Since then there has been a lot of work gone into implementing the planned security protocols before starting another testnet. So the attack you mentioned has already raised its ugly head and I’m sure will again. Although next time it won’t be so easy.
How large the network will have to be is uncertain, but when farming rewards are introduced then there wil be a lot more farmers out there than there was in the previous testnets. Probably 100 to 1000 times more once rewards are introduced and likely to grow at a very fast rate in the first days/weeks.
Great discussion BTW. I dont think anyone large enough will attack. And even if you get some hackers in eastern Europe or China who want a few bucks out of it by the time they get online there will already be hundreds if not thousands of nodes which will mitigate any potential attack. Just look at all of us here not to mention the non English speaking SAFErs out there. And just because there is a big data whale, so what. That whale needs data to farm.
Personally I think there are just to many well thought out hurdles for any big mob to be more then interested. First you have the lottery, then XOR, then optimum sizing, then competition with thousands of average joes and finally they can’t force the data onto the network and even if they do happen to capitalize on a large percentage of it, it won’t necessarily be valuable let alone profitable for them.
But then again everything can be gamed, I just don’t know if gaming it will be worth while is all im saying.
Yeah it depends on how quickly the price of safecoin goes up due to speculation in the early days. If it goes up a lot and a powerful malicious attacker can gain a lot by shorting safecoin and causing massive data loss & bad press for Maidsafe, then that could be a big financial incentive. But will it be very easy to take a huge short position on safecoin early on? not sure
The faster the price goes up though, the more people will be joining the network to farm. And even if the price stays at current levels, you’re right that a lot of people in the community already have a huge incentive to farm a lot in the beginning. So maybe it will be too expensive for most attackers paying attention to pull off.
It’s an interesting situation, because unlike Bitcoin, where gaming the system early on would not have resulted in a big price drop, because BTC had little to no USD value, Safecoin will already have quite a bit of value when it is launched, so the attack incentive is greatly increased. If someone pulls this off in the first month or two, it will not only crash the safecoin price, but will also undermine people’s trust in SAFE securing their data for a long time potentially…
Edit: On second thought, maybe the market won’t react badly to an attack in the very early days. I’m sure during the first couple months or longer, most people will not be storing important information on the network. They will simply be testing it out. So some loss of data would not necessarily guarantee the attacker that the price would drop significantly, so probably not worth it…Maybe I just answered my own question…
