38

What conditions need to be met for the central coordinator to be shut down?

Is there a number of transactions per second, a number of full IRI nodes or an update to the algorithm necessary?

Improve this question

2 Answers 2

Reset to default
26

The coordinator or "COO" will be shut down as soon as the network reaches a certain number of transactions per second (not known at the moment). This is necessary because "The Tangle" is vulnerable to attacks in its infancy. The network is secured if there are many people confirming transactions all the time. No exact date can be given as when the COO will be shut down as no one knows how fast the adoption will be.

"Coordinator: IOTA is currently in what should be considered a ‘transition period’ towards large-scale deployment and standardization. Like Bitcoin, Ethereum and all other distributed consensus protocols before it, the IOTA network need an onboarding mechanism to provide 34% attack protection in its early days. Due to the unique underlying architecture of IOTA, this takes the shape of a ‘Coordinator’. The ‘Coordinator’ or ‘Coo’ for short, is essentially training wheels for the network until the amount of organic activity on the ledger is sufficient to where it can evolve unassisted, at which point the Coo is permanently shut off. This does not mean that the IOTA ledger is currently in any way centralized, the network is 100% decentralized, every node verifies that the Coo is not breaking consensus rules by creating iotas out of thin air or approving double-spendings. In fact, any talented programmer could replace Coo logic in IRI with Random Walk Monte Carlo logic and go without its milestones right now, so technically even at present, the Coo is entirely optional. The only role the Coordinator serves is to protect against attacks in this temporary infancy stage of the Tangle ledger, if we shut the Coo down the network would continue to evolve as it will in the future when it is unassisted by these ‘training wheels’."

David Sønstebø - Founder

EDIT 30-nov-2017: enter image description here

Sources: https://blog.iota.org/the-transparency-compendium-26aa5bb8e260 https://www.reddit.com/r/Iota/comments/6lspeq/the_coordinator/

Improve this answer
6
  • How will we know when the TPS is high enough?
    – RevMen
    Commented Nov 30, 2017 at 15:17
  • I don't think the developers put a number on it yet. Maybe they are not sure yet. It's experimental technology. I asked on answers.thenextweb.com/s/… Didn't see an answer yet
    – IOTTOI
    Commented Nov 30, 2017 at 17:00
  • I understand that the number is not known. I'm asking how that number can be learned.
    – RevMen
    Commented Dec 1, 2017 at 19:29
  • 8
    "the network is 100% decentralized". This is repeated over and over again. However, it is still not true. The coordinator is THE central component in the network and the network would break down without it immediately, otherwise the coo would have long been shut off already.
    – lex82
    Commented Dec 8, 2017 at 21:10
  • 6
    So the short answer is: there need to be enough tx/s by honest participants. The devs chose a central authority to avoid attacks. Why don't the devs just trust their userbase to take care of the problem and let them provide hashpower in terms of load-levelled good-spam tx? The longer the Coo sticks around, the higher the IOTA market cap and adoption, the greater the fear to risk shutting it down. Performing research simulations is fine, but you can only know if it really works if you try it in the real world with humans in the loop
    – brenzi
    Commented Dec 10, 2017 at 19:46
4

Summary

Saying that IOTA can run without the Coordinator is like saying:

Our car can run on water instead of gasoline. But for now, it has a lawnmower engine under the hood. But that is only temporary, until the engineers (very brilliant guys) finish developing the water-burning one...

What conditions need to be met for the central coordinator to be shut down?

The number of assiduous honest transactions must always be found to be in the majority. In order to guarantee that, they must, by all practical estimations, be constantly generating new honest transactions at all times.

For further reference to this, see As per "The Stability and the Security of the Tangle", how will IOTA ensure that all honest nodes are continuously using their hashing power?

This is further specified by the IOTA Foundation in their FAQ documentation:

The most critical factor needed for the removal of the Coordinator, for example, is the greater adoption of the IOTA technology increasing the throughput of transactions on the network to meet the fundamental security assumption - that the cumulative throughput of honest transactions is large compared to that which an attacker could feasibly produce.

Is there a number of transactions per second, a number of full IRI nodes or an update to the algorithm necessary?

Yes. In 2015, a developer mocked up a scratch scenario on bitcointalk and predicted that over 28,666,666 transactions per second would be required to secure the Tangle.

That estimate is actually very conservative a number since the N is much bigger now than when the example was written.

The formula - using SHA-256 and a Core 2 Duo for Proof-of-work calculations:

  • Let J be the average Joe hash rate
  • You cannot ask Joe to wait more than 60 sec to issue a single transaction, so the minimal PoW cannot be more than 60 * J
  • Let E be the attacker's hash rate

The minimal number of transactions per second that you need in order to keep the system secure is N = E / (60 * J)

So for SHA-256 (in fact, what hashing do you consider?):

  • Let's take the Core 2 Duo hash rate for Joe J = 2.5 MH/s
  • Today's hash rate of the Bitcoin network is around 430 PH/s. It is plausible to assume that a single entity owns 1% of that hash power E = 4.3 PH/s = 4 300 000 000 MH/s

=> The minimal number of transactions per second is the astonishing N = 28 666 666

What are the TPS and POW specifications for IOTA at the moment?

MH/s and TPS for POW Function

The first FPGA boards can now easily do 25 MH/s (for 5 TPS). Higher performing boards are now available that can do more for a reasonable price. According to Thomas Pototschnig on reddit:

I tried to port the core to the DE10 Nano board ($130) and it would archive about 80MH/s :)

Furthermore, it is expected for low cost trinary chip designs to do POW even more efficiently.

Current BTC Network Hashrates for Formula Calculation

The BTC (only BTC and not any other coins such as BCH, ETH, etc. etc.) network is today running a hashrate at 51,000,000 TH/s, and 1% of that is 510,000,000,000 MH/s.

What is a real estimate based on recent BTC hashrates using this calculation method?

If we take that a 25 MH/s FPGA can do 5 transaction per second ( so that 1 TPS rerquires 5 MH/s), and with the average user IoT device being able to do 1 transaction every second, then J would be 5 MH/s.

The formula yields a required network TPS to secure against 1% of the BTC network hashrate, of 1.7 Billion TPS by honest participants.

Tangentially: the attacker also has all the power at their disposal at a facility that is directly into the internet's backbone (T1 etc.) and can direct the attack to the correct network nodes in a strategic order. This isn't necessary but it surely enhances the attackers abilities.

Just the Facts

It is in the order of 1.7 billion transactions per second that must be constantly generated.

The constancy of the number of transactions depends on this fact, that at any time it is possible to double-spend a transaction and the only way to guarantee that is to ensure that the honest hashpower is at all times in the majority.

As confirmed by Come-from-Beyond:

if transaction flow drops then an adversary can catch up and double-spend. It's a never ending race.

Come-From-Beyond says a few posts down that this example "Looks good".

Improve this answer
8
  • tbh i think both answers are leaving variables out. The first one ignores where do we start the random walk without milestones w.r.t things like splitting attacks and sidetangles? Your answer kind of leaves out variables like bandwidth and node hardware limitations. Assuming that the infrastructure can handle ~28 million tps, i dont think nodes could handle it so an attacker is more likely to achieve a massive ddos than a double spend. Also an attacker would need to neighbor with a significant number of nodes in order to avoid only referencing their own transactions, which would increase cost.
    – Johnny Milkshakes
    Commented Jul 12, 2018 at 0:37
  • 1
    This is a really good answer. I am under the impression that relying on PoW alone to secure a COO-less Tangle is impractical. A rumor has been spreading around for the past year that IOTA will "move away from PoW". I'm eagerly waiting to see how things will turn out. IOTA Foundation is really pushing the limit of innovation here.
    – Bayequentist
    Commented Jul 12, 2018 at 9:21
  • 1
    It's the ratio between the Joe's transaction speed (and whether that speed is utilized or left idle!) and the adversaries transaction speed that matters, not the absolute numbers in hashes per second.
    – kfx
    Commented Jul 22, 2018 at 12:35
  • My point is your reasoning can't be correct. If you think it is, consider what would happen if IOTA's PoW complexity would be changed by a factor of 10. Since this is not a parameter in your estimate TPS, it would not change, which is absurd. Also, I got the feeling that you don't realize that IOTA uses a different hash function than BTC.
    – kfx
    Commented Jul 25, 2018 at 9:01
  • Why is the entire honest network assumed to not have FPGA/ASICs? Wouldn't some honest participants also have them?
    – user482
    Commented Jul 25, 2018 at 17:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.