Junaeth Series #3 – Reputation System

July 30, 2021

The Stealth blockchain owes its unsurpassed efficiency to a set of block validators, called StealthNodes, that are chosen through purely economic means. StealthNode operators buy their irrevocable validation rights through irreversible transactions on the Stealth blockchain. Validator selection and the reward mechanism for validation forms the heart of the Junaeth consensus protocol.

One can assess StealthNode performance with a variety of metrics. The Stealth blockchain explorer website, stealthmonitor.org, displays several metrics that include daily average block interval, daily blocks missed, and the block interval RMSD (root-mean-square deviation). Each of these metrics represent essentially similar information in slightly different ways.

 

Average Block Interval

The average block interval measures the average time between blocks over the course of a day. This value is defined by the equation:

 

This equation simply says to add up all of the block intervals (Δt with subscript i) for N number of blocks, and divide by N. For example, if we consider three (N = 3) blocks with intervals 4, 5, and 6, then we add these up (4 + 5 + 6 = 15) and divide by N: 5 = 15/3. In other words, for these three blocks, the average block interval is 5 seconds.

The average block interval says how consistently StealthNodes produce their assigned blocks. If one StealthNode is a second late with a block interval of 6 sec, then, if no StealthNodes miss any blocks, we expect another StealthNode to have a block interval of 4 sec, restoring the average. Thus, the average block interval metric is somewhat forgiving to StealthNodes as a collective, eventually erasing any degradation of the statistic resulting from late blocks if the collective is producing blocks dependably.

As of writing, the mean block interval for the last 24 hours is 5.0116 seconds, which means that StealthNodes have collectively only missed 0.2% of their blocks over the last day.

— — — — — — —

 

Hourly Blocks Missed

Obviously it might be more intuitive to simply look directly at the number of blocks missed in a given day. With 5 second blocks, the Stealth blockchain is expected to produce 17,280 blocks every 24 hours. As of writing, StealthNodes have collectively missed 32 blocks in the last 24 hours. This number represents about 0.2% missed blocks, consistent with the estimate calculated from the daily average block interval.

StealthNodes can miss blocks for a variety of reasons. The most common reasons are network latencies (unavoidable lags in sending and receiving communications) and temporary disconnects. Disconnects happen on all networks, and can be due to denial of service attacks, or outages that temporarily completely disrupt communications. The latter can often happen with ISPs (internet service providers) who have scheduled outages or disruptions related to reconfiguring their networks to balance traffic.

— — — — — — —

 

Block Interval RMSD

The block interval RMSD is defined by the following equation:

 

In this equation, N is the total number of blocks over which to sum, Δt with subscript i is the block interval for block i, and Δt with the bar over it is the target interval (i.e. 5 seconds) for each of the N blocks.

The RMSD is a complicated equation that probably looks like alphabet soup to most people, but the final value gives an amazingly good sense of overall performance for a certain number of consecutive blocks (N). That is, if N is 120, then the RMSD is calculated over 120 blocks, or 10 minutes.

If we break the equation down we see that at the heart of it is the parenthetical expression that calculates the difference between a particular block interval and the expected block interval. For the simplicity required to get an intuitive understanding of RMSD, let’s imagine we want to look at a single block interval, where N is 1 block. Imagine that this block happened 6 seconds after the previous. That makes the interval 6 sec (Δt with subscript i). The absolute value of the difference between this block interval and the target interval is 1 sec. If one evaluates the RMSD equation for this single block, the result is 1 sec.

This analysis gives an intuitive understanding for the block interval RMSD, which is the amount of time the average block interval deviates from the target interval.

As of writing this post, the block interval RMSD of the last 24 hours was about 0.25 seconds, which means the average block deviated from 5 seconds by just a quarter of a second. This small deviation has a barely perceptible impact on the user experience and is dominated by missed blocks.

— — — — — — —

 

Reputation

Collective statistics help set a standard for the performance of individual StealthNodes. This relativistic assessment is embodied in the Junaeth consensus protocol as a reputation system. In this system, the total blocks produced and total blocks missed by each StealthNode is tallied and used to calculate a reputation score, called weight. Weight is the square root of the difference between total blocks produced and total blocks missed.

The square root was chosen because it has convenient scaling properties and allows newer nodes to play catch up with older nodes, assuming good performance. The following graph shows the behavior of the reputation function up to 10,000 net blocks.

 

Notice how a constant difference in net blocks leads to a decreasing difference in weight as the total net blocks increases. This behavior is what allows for “catching up”. But also notice that a newer StealthNode will never fully catch up with an older StealthNode, assuming equivalent performance.

Reputation (weight) is important because a StealthNode’s earnings is proportional to its weight. Additionally, the influence a StealthNode has on consensus in the event of a conflict is also proportional to its weight. So not only do StealthNodes have an incentive to have good performance, less reputable StealthNodes have an incentive to agree with more reputable StealthNodes. The reason for the latter is that, in the event of a conflict, StealthNodes want to produce blocks on the eventual winning chain so that they can be credited with block rewards.

reputation

Fast. Secure. Reliable.

Get the Stealthsend Desktop App

OSX

Copyright © 2023 Stealth R&D LLC. All rights reserved. The Stealth main blockchain “StealthCore” incorporates all of the features to ensure FATF Travel Rule compliance.