If the lights went out on 90% of the network

Given a network size of 30,000, what would happen if 300,000 bot controlled nodes created new accounts, and then turned them all off. Would data be lost or would rank have prevented them from storing all 4 copies of a piece of data? What would happen too if the bot accounts lay dormant (earning rank) for a period of time and then attacked (turned off all at once) at some point in the future?

I would love to know the answer to this too. Thanks in advance!

The scenario was already discussed on this forum.
Most data would be lost, but it wouldn’t be particularly useful for the botmaster. They’d have to spend a lot of resources and they’d gain nothing in return (except it’d be fun).

The most important is not the number of bot nodes but the total of capacity of this nodes. For this attack you need at least 3X the honest network size in petabytes. You need too to gain rank so you must aid the safe network to grow being able to enter, thereby, in a vicious circle.

Better network->New honest nodes->Need 3X new bot nodes->Better network


Would the amount held in peoples safecoin wallets also be lost, or is this data backed up on more than 4 copies ?

It seems like the network would recover from having data deleted, but if the safecoin values were lost it’d be game over I think (for this iteration anyway).

Not necessarily so (the most important). What’s very important is their rank as well. If owners recognize and eliminate the trojan (together with the illegally installed instance of MaidSafe), if the clients are slow and often offline and so on, they won’t be able to get a good ranking and really use their capacity well.
The more the node stores, the longer it’ll need to earn a good ranking and the more likely that the trojan will be detected by the system owner, while most semi-professional farmers will have nodes with a high ranking.

I assume you’ll be able to backup your wallet as with all other crypto-currencies known to man.

Would kinda suck if everyone who’s funds weren’t back up disappeared though. The permanence (resilience) of the network relies upon the continuity of every single safecoin IMO, not just ones which are backed up. It seems sensible to argue that the financial data in the network deserves a higher ‘security priority’, in order that confidence would remain even if a few data files were to be corrupted/deleted in a successful attack.

They are 4 replicas on the same system remember which introduces inherent weaknesses. If it was my life savings in safecoin I would either want my safecoin data backed up on a massive number of nodes in the network, or have it backed up offline/on a blockchain. If people don’t back up then they could lose everything if an unforeseen attack was successful it still seems. This should be made pretty clear to everyone who holds more than a trivial amount of safecoin if this is the case.

If this is an issue is it a scale issue? A teething issue?

  • I trust there must be a disclaimer somewhere around here which says don’t put in MAID/SAFE more than you can afford to lose
  • A replica (and that’s why I call them replicas) is not a backup, which is exactly the point of Mr. Wonka’s question. Even if you had 7 billion replicas, you’d still need to have a backup. It’s completely unrelated to the question of security.

I guess that is true of all crypto currencies. If a 51% attack was undertaken on bitcoin then there be a fork which maintained balances prior to the attack so whilst the $ value of bitcoin might go down the consensus as to who owns what is still maintained. If balances held in safecoin immediately prior to an attack cannot be agreed upon then it seems the whole network has to start from scratch again as opposed to from that moment?

Would be great if you could expand on that as I don’t quite understand.

Would be great if you could expand on that as I don’t quite understand.

Replica is a copy that’s being synchronized with the original (there are variants, e.g. some software has real-time synchronously replicated replicas and there’s no “original” which is always first to be updated, other s/w replicates asynchronously or lazily, etc.), so whatever change is made on one of the copies, eventually (0.001 or 1 week) it will be propagated to all the copies.

That means that something as simple as a wrong keyboard combo can delete all four replicas before you notice. Or that some virus can encrypt your MaidSafe data and promise to give you the password if you send 5 BTC to a certain address. 1 copy, 4 copies or 40 copies - there’s no difference.

Whatever the approach there needs to be a separate & independent copy of your data aka “backup”.

Each safecoin is a file so one backup is enough to restore the balances. I remember Irvine speak about voluntary nodes holding a copy of all safecoin.

However your question is somewhat tricky. Why not an attack of 95% or 99% or 99.99%…?

And in an attack of 51% the bitcoin value will be zero because it will have lost its core value who is trust.

Its not fixed to 4 copies, there in a minimum of 4 copies on the network.

Didn’t realise this - a cunning idea :smile:

So there could potentially be millions of copies of the same file spread throughout the network, and the network storage space is always ‘full’ (if that makes any sense)?

Is there a way this can be built into the system (i.e. automated) so that a backup is something the end user does’t have to think about?

I think @janitor is saying don’t trust one backup system (the network), always at least duplicate.

This is my position, although I’m open to being convinced that the network makes this unnecessary.

The issue with Safecoin is that (like bitcoin), they only exist on the network. However, all bitcoin could in theory be reconstituted from a single backup of the blockchain, and there are many such copies.

It sounds like SAFE has an equivalent mechanism for Safecoin, where certain nodes keep a complete copy of the Safecoin, in effect allowing all coin to be reconstituted from such a backup.

The same cannot be done for our files however, so we should all consider whether certain data needs a separate backup.

1 Like

I didn’t know this - a very neat solution.

According to Peter Todd:

“things like trusted hardware can provide weak pseudo-solutions.”

Is that what he was referring to you think, and why would they be ‘weak’?

I don’t know that quote or what he’s referring to sorry.