On SteemBots and Voting Errors

Introduction

In Maslow's Hierarchy of Needs and The SteemBot Revolution, I began to make an affirmative argument to demonstrate that steembots, including voting bots, contribute to the overall well-being of the steemit ecosystem. This article will continue that argument by considering voting errors and demonstrating that humans and bots complement each others' capabilities in a way that makes the steemit platform better overall.

Two Types of Voting Errors

vote
Image source: pixabay.com, licensed under CC0, Public Domain

In statistical testing, there are two types of errors: Type-I and Type-II. A Type-I error, or false positive, is incorrectly rejecting a true null hypothesis - finding an effect that doesn't exist. A Type-II, or false negative, error is when the null hypothesis is false, but not rejected - an effect exists but isn't found.

Likewise, in steemit, two types of voting errors are possible. If we take steemit's null voting proposition to be that an article is not valuable, then if I vote on an article that I wouldn't actually believe is valuable, that is analogous to a false-positive, or Type-I error. In this case, by "finding" value that doesn't exist, an author may wind up receiving an undeserved award, and I may even receive an improper curation reward. On the other hand, if an article is posted that I find to be valuable, and I have sufficient voting strength to vote, but I neglect to vote for it, that is analogous to a false-negative statistical error. This type of error is demoralizing to authors of valuable content.

This can be summarized in a table:

Article Quality

Voting Decision	Article is NOT subjectively valuable to me	Article is subjectively valuable to me
Vote	Type-I Voting Error	Voted Correctly
No Vote	Voted Correctly	Type II Voting Error

*Note: I don't mention flagging, because I consider it mostly harmful to the platform, and therefore don't use it. But it should be noted that flagging would complicate this analysis.

The Type-I form of error is often discussed, because it is highly visible and vulnerable to abuse. This is what seems to lead many people to dislike bot voting.

Type II voting error, however, is an important and infrequently discussed (in my feed, anyway) phenomenon. Any time my voting power is at or near 100%, if a valuable article exists and I am not voting, then I am committing a type-II voting error. In this case, I am foregoing curation awards that I might be earning, and I am missing an opportunity to help others to earn author awards. These missed votes may contribute to author retention problems, limit steemit's growth, and otherwise prevent the steemit community from maximizing its potential. All of these things reduce the value of my own SteemPower (SP) holdings.

Additionally, and this is underappreciated, by adding authors of valuable content to the reward pool and thereby diluting payouts, a reduction of Type-II errors will also tend to limit the harm that is done to steemit by Type-I errors. It is my (unsupported) opinion that the vast majority of steemit voting errors are presently of the Type-II variety.

Humans and Steembots Have Complementary Strengths

As a person with a family, a job, other interests, and needs for food and sleep, I am constantly committing Type-II voting errors. It's unavoidable because it is simply not possible for me to manually read and vote for every quality article that's posted on steemit, even at steemit's relatively small beta size. Some quality articles will always get past me. This problem will only be exacerbated as steemit grows. I can reduce my type II errors by reading less carefully, not viewing videos, or even by voting from the thumbnail view, but this reduction comes at the cost of increasing Type-I voting errors.

On the other hand, a voting bot doesn't need to eat or sleep, and it can vote whenever it has voting power, 24x7x365. Even a bot won't get every quality article, but it will do much better than I can do. At the moment, the drawbacks to bot voting are that the barriers to entry are significant, and the widely available heuristics for value are low-quality, so the few users who manage to launch bots will commit substantial Type I voting errors - at least initially.

It should be noted that the @robotev bot by @cryptos and the @steemvoter bot by @marcgodard both begin to lower barriers to entry with bot voting. I'm sure there are already other tools out there, and even more will emerge.

This table summarizes the human and bot relative strengths and weaknesses

Error Type

Voter	Type I (False Positive)	Type II (False Negative)
Human	Very good at avoiding	Cannot avoid
SteemBot	Difficult to avoid (but improving)	(Relatively) Good at avoiding

Conclusion

The intuitive solution to voting bots that vote on low-quality articles and deliver unsupported rewards through Type-I voting errors is to reduce or eliminate the use of bots. However, this action would increase the overall number of Type-II voting errors, which would also be harmful to the steemit platform. The better solution is to increase the quality and number of bots in the ecosystem, thereby reducing the Type-II voting errors and mitigating the harm from Type-I voting errors. Although counterintuitive, this measure should limit the damage that is done by abusive bots, and it should also generate more wide spread author satisfaction.

Therefore, as barriers to entry come down and quality goes up, all steemizens would be well served to make use of both bot voting and manual voting, in order to balance Type-I and Type-II voting errors. As I noted, here, I fully expect a vibrant market to emerge for non-technical users to harness the increasing benefits of voting by steembot. When that happens, all of steemit will benefit.

@remlaps is an Information Technology professional with three decades of business experience working with telecommunications and computing technologies. He has a bachelor's degree in mathematics, a master's degree in computer science, and is currently completing a doctoral degree in information technology.