[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/biz/ - Business & Finance


View post   

File: 8 KB, 235x215, 945678231.jpg [View same] [iqdb] [saucenao] [google]
20591621 No.20591621 [Reply] [Original]

Just read through the chainlink whitepaper 3 times and couldn't find an answer to my question.

If a smart contract is requesting data, lets say its for the price of ETH in USD, and 6/10 nodes give an untrue, but similar, response.
How does chainlink handle the fact that 4 honest nodes will be penalized and have their collateral slashed as a result of this.

Not only would the 6 lying nodes stand to make a profit from providing the false data in the first place, and abusing the smart contract that relies on this data
But the 4 honest nodes would get LINK slashed from them.

How is this handled? Are the 4 honest nodes just fucked if it's a high value slash contract agreement
Does someone at chainlink go in and manually make a decision to override the protocol, effectively eliminating all decentralized aspects of chainlink?

Does anyone have a legitimate response to this

>> No.20591651

>>20591621
dr;ns

>> No.20591850

Statistics even it out. assuming all nodes are basically honest the chance that one node will be punished all the time is fairly low, But yeah something something sybill, probably no issue in such markets like x/usd as there are many nodes and the collateral requirement close to 0, so not much reason to attack. For more exclusive data, licence the node operators before they even get access to the data point and put up retarded collateral requirements

>> No.20592016

>>20591850
I agree that my example was a bad one, as simple price feeds will likely not have much at stake, just wanted to give an easy to understand example

>For more exclusive data, licence the node operators before they even get access to the data point and put up retarded collateral requirements
I'm not sure I follow exactly what you mean by "license the node operators"
But yes, with more exclusive data you would expect very high collateral requirements

If there was extremely exclusive data and all 10 nodes had $1M USD as collateral, and 6/10 nodes colluded to provide false information
The other 4 nodes would take a massive penalty on their $1M staked, and the 6/10 nodes would also make whatever profit they stand to gain from providing false data in the first place
whether thats abusing the smart contract requiring that data directly, or being paid off by a third party who is abusing the false data feed

how is this handled?
is the general idea that the 6 nodes would lose all their reputation and would be effectively barred from ever filling jobs again
doesn't this just create an exit scenario for long-term nodes who are done being nodes, and so as a final way to milk profit they collude and provide false data, then lose all reputation and ability to provide data
- but they dont care because they were done anyways

>> No.20592455

anyone?

>> No.20592522

>>20591621
I Just want to know How much I Will get from each 1k link I stake on linkpool

>> No.20592557

>>20592016
Every Nigger can provide data on the price of eth/usd. To get access to medical data apis, you need a bit more I guess, so getting that access is already somehow of a sign the node operator can be trusted

>> No.20592582

Yes. Read the BFT paper from the 80s, retard

>> No.20592631
File: 20 KB, 290x330, 1575042270513.jpg [View same] [iqdb] [saucenao] [google]
20592631

will trolls be able to hack my node and force it to give wrong data, forfeiting my links

>> No.20592675

>>20591621
This is a problem with all blockchain technology.

But I know you are a troll, and I fell for your juicy tasty bait. I am a fool.

>> No.20592720

>>20591621
most og linkers that actually know this shit probably dont browse biz anymore. all the new linkers care about is making fiat gains.

>> No.20592747

Read this guys early papers. Its basic BFT

https://en.m.wikipedia.org/wiki/Leslie_Lamport

>> No.20593471

I guess I figured it out myself, the general idea is that while this is technically an unsolvable fault of chainlink, the very fact that it would take collusion from multiple parties makes it inherently more secure than any other Oracle design.

If a smart contract was feeding data into itself from a single provider, it would only take malicious intent from that provider.
Whereas, with chainlinks design, it would take collusion by a number of parties, specifically at least 1 more than half of the nodes that aggregate the data response.

So while it is an issue with the chainlink network, the design of chainlink makes it inherently less of an issue than literally any other design possible.
just thought id post this in case anyone else was wondering, thanks again niggers

>> No.20593666

>>20592016
I believe that nodes are unaware of the other nodes in their designated network for a singular task. I’m pretty sure it’s randomized. So in order to collude, it would have to start with a malicious node within the randomized node circle communicate with ALL other nodes to figure out which nodes are working on this specific task in order to collide. Thing is, probability of randomly contacting a node 1) within your circle for the task and 2) one that is ALSO willing to collide is extremely improbable.

This is at least the way I understand it. Could be missing something.

>> No.20593706

>>20591621
cant read still buyin

>> No.20593827

>>20593666
Checking myself 666.

I’m kind of just thinking intuitively here about the circle of accountability. I haven’t read that it works this way exactly but I imagine it’s something similar