CfB discusses advanced future IOTA stuff on #tanglemath


Author
Message
EricHop
E
Attached to Tangle (468 reputation)
Group: Moderators
Posts: 9, Visits: 0
Q: @come-from-beyond You once said "for example, if we add anonymity that breaks links between inputs and outputs then we'll break IOTA .. it will have to have full tangle history in this case... swarm logic will stop working" .. can you give any kernel of knowledge on what you mean/ why this is?
A: remind me context plz, so I don't spend 1 hour explaining Economic Clustering
Q: A: local iotas = iotas spending inputs which are outputs of previous transfers and those transfers belong to the referenced subtangle too
  you can travel to another country and spend iotas there
  in this case you will be waiting quite long
  so it makes sense spending iotas to your own address again and again during the trip
  in this case you'll be keeping "locality" pretty high
  A: There are a lot of things which are not obvious
  for example, if we add anonymity that breaks links between inputs and outputs then we'll break IOTA
  it will have to have full tangle history in this case
  swarm logic will stop working
  pity that we can't have anonymity, but it's the price to pay for virtually unlimited scalability
  another things is: people ask what happens if someone spends the same iotas in 2 different parts of the earth
  answer is this: one of the spendings will have to wait a lot of time
  the one closer to the original "location" of the iotas will be confirmed faster
  another drawback comes to mind: iotas should be "refreshed"
  we have discussed the refresh .. and the local iota .. lol .. I now understand both of these ideas
  well "understand" at a conceptual level
  but I don't understand the inputs and outputs thing
  this was a saved conversation in this chatroom from probably a while ago .. don't know when .. and the first message was the one by you on local iotas .. not sure if you initiated the discussion or if there was more discussion prior to that comment that didn't get saved
A: swarm logic can't work if you don't see what others are doing
Q: couldnt the adresses get like hashed or something like that? so that only the tangle itself knows what is what?
and the key to the hash hardcoded into the tanglemath
A: To the best of my knowledge schemes allowing to implement that in practice and be quantum-proof don't exist
Q: you mean because hash functions arent quantum proof?
A: i mean because there are no such cryptoschemes
Q: then lets invent a new one like we did with the tangle :iota:
A: let's start now, coz according to some researchers you have to wait at least 6 years before you can start using those schemes in production :trollface:
Q: OK ..so my understanding is that this is directly related to the sub-tangle/local iota thing ... if there is only one main tangle which everyone can see and come to consensus on it is not important _from_ where and _to_ where iotas are going because in the end the total number of iotas remains the same ... but in the case of a sub-tangles .. you need to know to and from where because once you reconnect with the main tangle it would be impossible to reconcile who has what
is that kind of the gist?
A: hard to say, the claim that there is only a supercluster looks flawed to me, this kinda invalidates the rest of the sentence
do you mean real-world adoption with thousands transactions per second?
btw, this reminded me about SPECTRE, they will open testnet this year, right?
I heard they might offer bounties for hackers
I'm going to see in practice if their claim of feasibility of a single cluster is correct
I think we are not on the same page, I look at things from meta-level where Bitcoin and Ethereum Forks are taken into account and a theory exists explaining them. On that level "longest chain wins" rule makes no sense.
From that meta-level even existence of altcoins is explained, while you remove them out of the equation completely
others working on such things too - https://www.cryptocoinsnews.com/jeff-garzik-hopes-new-digital-currency-metronome-solves-bitcoins-shortcomings/
Q: Maybe I wrote that in a confusing way. I meant this:
Scenario 1: Supercluster ... obfuscation of inputs and outputs is possible .. but it wouldn't be infinitely scalable
Scenario 2: Existence of sub-clusters ... obfuscation is not possible ... but it can be infinitely scalable
A: If obfuscation is done you need to send all info to all neighbor clusters
in this case all clusters merge into a supercluster
Q: exactly .. so it is not scalable
A: Imagine clusters connected this way: A - B - C
ah, you got it
I won't explain it then
Q: I think we are on the same page. It basically comes down to the fact that it is simply not feasible to achieve 100% consensus on a global scale with a high throughput ... thus the formation of economic clusters is unavoidable
A: Economic clusters were discovered decades ago, extra researches only confirmed the phenomenon
I suspect that attempts to "map" single-sharded cryptocoins to global economy landscape will fail
because clusters which are not connected don't care about other's money. they don't need to secure them by issuing blocks.
even if they are irrational they won't spend extra resources for security of transactions between participants of non-related clusters
it's economic laws
scale Bitcoin to global world and Bitcoin will start falling apart
to mimic the same structure as global economy made of clusters
simple simulations show that
just WTF that noone attempted to do them
that's less than 1000 lines in JavaScript
(you have to guess correct assumptions regarding expenses though)
Q: yes .. so with IOTA you don't impose a protocol which requires people to care about things outside of their cluster .. you don't try to make water flow up hill .. you allow the clusters to form and only remerge only if they want to
A: aye
Ethereum follows the same route with their Sharding, but there is a flaw in their construction
I think Vitalik will notice it once he writes whitepaper or implements test version
or not
Q: i just thought about what about encrypting the amount of iotas spent, so that the amount gets encrypted with something like MD5 then the receiver has the key to decrypt it and so on? then the tangle would keep track of everything but the amount wouldnt be visible (I am really sure you thought about something like that before, or that its impossible, but just interested why)
Q: I am not very familiar with ethereums sharding .. but seems like a good explainer here https://github.com/ethereum/wiki/wiki/Sharding-FAQ
GitHub
ethereum/wiki
wiki - The Ethereum Wiki -
A: practical schemes don't exist
Q: I think the fundamental problem is that if there are separate clusters they will have no knowledge about what is happening in the other ... if you obfuscate inputs and outputs .. or amounts ... clusters would have no way to remerge .. obfuscation would force them to forever remain separate once they broke apart .. IOTA allows for clusters to form (if they want) and to re-merge (if they want) ..
or allow for just one participant in one cluster to move their balance to another cluster (again if they want) .. although they would have to do this adjacent cluster-by-adjacent cluster
A: funny observation regarding cluster: very fast emerging leads to iotas loss
because nodes (percentage wise) which know nothing about existing balance increase too fast (comparing to syncing)
Q: yes ... and the possibility of double spends right? its kind of important that clusters form somewhat slowly and naturally .. which seems like would be the case
A: depends if it's honest nodes which join
honest = loss
dishonest = creation out of thin air
Q: this raises very interesting questions about what the price of iota across clusters would be ... presumably an iota in cluster A would not be worth the same as an iota in cluster B
or what would the global value of an IOTA be? its almost as if there wouldn't be one .. IOTA's value would be cluster dependent
which is really no different from USD today .. i can buy a whole lot more with dollars in mexico than I can with dollars in NYC
same dollars
A: I think it will work this way:
You deploy a node in some area. You buy iotas from someone in that area. Without easy flow of iotas between that area and some other area there will be price difference determined by arbitrage opportunity.
I mean you deploy a device that needs to pay for services in iotas
not a node
Q: makes sense .. it will be interesting how easy it will or will not be to get iotas from one cluster to the next .. we discussed before the possibility of "iota exchanges" which set up nodes in both clusters ... maybe because of these exchanges and arbitrage the global value would actually converge
but 10 iotas in one cluster can buy what 100 iotas can buy in another ... really it would be value of the goods and services within an economic cluster that would be different
not the iotas themselves
this might be circular logic
A: yes, and even more itneresting things can happen if a cluster prevents flow of iotas into it's subclusters
Q: capital controls
A: this creates lack of iotas and rises their price
but it's not our problem, it's problem of AI that will be using that
I think humans will keep using USD
Q: I don't know ... as they say 'money is power' .. and I subscribe to the Kenneth Waltz's view of world affairs ... "?in international politics force serves, not only as the ultima ratio, but indeed as the first and constant one"
its the people with the guns that make the rules ... and until the AI apocolypse when they actually take over as the dominant species .. it will be humans who hold the 'ultima ratio'
aren't you Russian? you must be stooped in the cold war realism determining the current balance of power in our world
but i digress... this is not very tanglemathy
A: it's related to IOTA's cryptoeconomics
so it's tanglemathy
maybe we should open cards
like we revealed that PoW will actually be network-bound
or maybe should keep them hidden yet
to let the trolls troll
Q: open cards?
always hoped to end up in one historic chat, great discussion guys, happy to read you... give us the juice :slightly_smiling_face:
this one will be big :ok_hand:
Q: we should have an open discussion about this so people understand the limitations.
A: it's related to IOTA's cryptoeconomics
Q: the limitations are not unique to IOTA ... the limitations are inherent to the world we live in
Q: Yes I meant cryptoeconomics.
A:
> open cards?
I mean to explain all nuances, to publish results of simulations. But I can name reasons not to do that and they are more important than several guys screaming that IOTA can't work (even if they genuinely think so).
Q: Another thing im interested in atm, is that some guys said that they have been hacked and iotas were stolen.. is ther any info on this?
A: There were a lot of cases, some people used seed generators created by scammers, some reused addresses. Do you mean most recent issue?
Q: Yes, most recent
Think it had something todo with reused addresses
A: That was a bug in client software that allowed to steal iotas from some users. The details can't be disclosed now to protect those who hasn't upgraded yet.
The software is open-source and if someone from community noticed the bug he would warn us, but noone really cared and only black hat hacker reviewed our work and took the bounty himself. :trollface:
If you weren't sending iotas around then you don't need to worry, you can't reuse if you don't use at all :trollface:
Q: Oh thats bad.. sometimes i feel like its good for iota to not moon right now.. better in 3-4 months when bugs are mostly fixed, because this will be fudded to death again
A: not that bad, every crypto which reached 10B+ marketcap had such bugs. Think of it as a prerequisite to success. non-promising cryptos are ignored by blackhat hackers.
Q: You have a point there:white_check_mark:
Q: exactly what I did Joy
A: If you weren't sending iotas around then you don't need to worry, you can't reuse if you don't use at all :trollface:
A: I was going to send some iotas around but couldn't find the seed. Not sure if it's good or bad coz I still have no idea where that seed is.
:trollpokemon:
Q: is the network-bound PoW scheme you guys want to use: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.596.9426&rep=rep1&type=pdf
A: no
PoW or PoS are just special cases of Proof-of-Resource
If you use any other scarce resource it will work too
Spectrum is pretty scarce, so scarce that we have a lot of problems with interference in WiFi
Q: so how do you prove you expended some spectrum resource?
A: it's something that we don't reveal yet
just whitepaper is not that cool
I'm going to show it working in real life
There is a team working on LiFi now
I published video some time ago
Q: Not a simulation but in real life. sweet
A: As first level approximation you can treat it as flow of transactions when traffic is 100% utilized
attack from outside would stop at the edges
still possible to attack from inside though
with 34%
but for that you have to own 34% of devices
or equip your hardware with a lot of radio modules
here some interesting pics related to the issue - http://wraits09.di.fc.ul.pt/wraits09paperDiogoMonica.pdf
Q: thanks
A: scroll to Radio Resource Tests part
Q: pictures are nice but I needed more words .. I found this: http://homepages.gsd.inesc-id.pt/~ler/reports/wraits09.pdf
Q: what they describe seems simple enough .. a node assigns its neighbors different channels and randomly listens to the different channels to see if someone is trying to pretending to be more than one node ... they can only transmit on one channel at a time so if you run the test enough you'll eventually find out which nodes are sybil identities
While this makes sense for nodes communicating on radio waves ... how does it relate to the internet? nodes _can_ transmit on different IPs at the same time
A: IOTA is for IoT
You have to mimic IoT
we do it now partially with manual tethering
Q: but how does manual tethering replicate this? a single node can create 10 sybil identities and manually tether all 10 times with the victim node
A: partially
now people got used to manual tethering
and we can finish
mimicking IoT
it was important to educate people before going 100% on that
Q: how could you run 'Proof of Spectrum' on the internet .. because a radio channel in the spectrum is like an IP address on the internet .. but the critical difference being a radio device can only listen and talk to one at a time .. but a server can listen and talk to many IPs
A: links between nodes degrade over time
to refresh them you need to solve cryptographic puzzles
which require CPU
so a server can't talk to many IPs
you have to run a lot of GPUs then
from far away it looks like bitcoin mining
Q: so just to be connected to a node in the network i have to be constantly doing some PoW
A: yes, but not constantly, we can tolerate some leverage between honest nodes and attacking nodes
for example, with 12:1 you need to do PoW 5 mins per hour
but attackers need 12x less nodes then
you are not forced to use PoW
you can use PoS (similar to Hallmarking in Nxt)
you can just link to your friends without any PoW/PoS
you trust them anyway so you can make them 100% discount on PoW
if IOTA is deployed to another environment
not Internet nor IoT
you can use antiSybil mechanisms efficient in that very environment
Q: so the antiSybil will be environment dependent ... internet you can have degrading links... radio you have radio resource tests ... other enviroment something else
A: aye
Q: makes sense
is the degrading nodes thing going to be implemented soon?
A: dunno, IOTA team is overloaded with PoCs with different companies
Q: nodes could even just keep the standard nonce-PoW also no? it would be up to each node how they defend against sybil attacks
A: any antisybil can be used
Q: thats interesting
so when you say PoW will be network-bound .. thats just one of many options nodes can utilize depending on their environment and trust level with neighbors
not something that would be inherent in the protocol
A: network bound in IoT
Internet is not main environment for IOTA
so nuances of Internet are omitted most of time
Q: network-bound may make the most sense for IoT since they have limited computing resources .. but its not like you could impose this .. they are free to continue to use standard nonce-PoW if they want
or use nothing at all
but then how does the whole tangle get secured? it will become hard to make assumptions about what kinds of anti-sybil measures nodes are making ...
or not making
A: it's clusters' subtangles which are secured
Q: do you assume clusters will all use the same anti-sybil
A: I assume that IoT will rely on network-bound PoW for reaching consensus
Q: but a node in the cluster is free to connect w neighbors which he agrees can use some other antisybil right?
A: in IoT cluster?
Q: ya
A: there are no connections in IoT usually
Q: ya but i can run a node w a radio device and an IP .. ill connect my IoT neighbors w RRT antisybil and my IP neighbors w decaying node antisybil
A: theoretically it can work if majority uses your method
your node will be an edge node then
Q: why would it have to be majority? why cant i be the only one
A: I assume that you have limited bandwidth
too late to analyze it properly
let's continue tomorrow
Q: ok i need more time to think about it anyway
Q: Again great conversation :slightly_smiling_face: the way we think is different. I didn't have time too fully understand all sentences. But you really think that the TANGLE will be bound to be an aggregation of several clusters... it will be a pleasure to read that carefully this weekend. I more and more think about something close to IOTA but even more based on probability (if i can write something clever - i fear it will not be :smile: - i'll share for sure)

GO

Merge Selected

Merge into selected topic...



Merge into merge target...



Merge into a specific topic ID...




Reading This Topic

Login

Explore
Messages
Mentions
Search