RE: STEEM Contribution Score - From Concept to Model
Very interesting.. How will this account for things like that it's, who engage with enough intelligence to spam a few unique comments to various users with high engagement or reputation or other metric scores, simply to augment their own reputation by playing the numbers game of the algorithm. Almost "reverse calculating" who they need to interact with and how they need to interact as opposed to being human and interacting naturally?
Let's face it, we're playing a zero sum game here against the people who are building these bots for profit... There will always be an incentive to game the system or metrics that are in place to "protect the content". And bots will always win in the long-run based on the resources available to them vs the resources available to a valid and interesting content creator.
reverse calculations would not be impossible but the information gain probability on each data point used for the weighting would change each day so it would mean working out which metric is the best on the day, rather difficult for the average person. You are so right there will always be an "incentive to game" and so we would need to continuously evolve and we would not be able to eliminate this,
I'm not trying to say that your efforts are for naught, but this is something that I feel any "rating" efforts will need to consider. In my previous analysis of the blockchain, before SteemSQL was a for-purchase tool, there were many attributes that overwhelmingly identified bots/spammers/scammers, but also found valid users who were new and unaware of "acceptable" posting etiquette. The same will occur for providing rankings of valid users, except the "smart bots" will adapt to appear like valid users. The monetary incentive will always murky even the most complex and thoughtful algorithm.
I wish you all the best of luck, but I hope that expectations are tempered and considerable value is not weighted on any of these new scoring mechanisms until they are fully vetted. Looking solely at the top rank and bottom rank may be tempting, but sampling the middle ranks will be paramount to developing a truly reliable scoring method.
I value your input and my efforts my very well be for naught but at least I can say I tried to come up with solutions, and I will keep trying. Steemit allows us do that, help try shape the future of the platform. My intentions are good.
I hear you, and know that you are one of the (few) people who have good intentions for Steemit. I truly do hope that your methods work and improve the platform. Nothing in life is perfect, and I don't want to be a spout of negativity.
I just wanted to offer some feedback on how I've seen things like this pan out here in the past.
"The road to hell is paved with good intentions." -- sometimes even good acts have unintended negative consequences. Just look at the bidbots...