Sounds like a good idea. The caretakers can do additional things that a normal witness does, filtering the content based on keywords and stuff. Ie not everything can be manually done. Facebook for example has some 1k manual reviewers but it’s mot of much use as we can see. Also it’s not an easy problem to sort out as YouTube for kids is plagued with all sorts of weird stuff and google is finding tough to handle the situation. So we have a problem - a difficult one here and answering it with something like kid STEEM will be great idea
You are viewing a single comment's thread from:
There are certainly many things that can be left to a simple algorithm like inappropriate language and links. No humans are needed for that. But banning/nuking users and investigating trolls and bullies must have some human input. Think cheetah and steemcleaners. The users themselves should be able to report inappropriate content and algorithms can help prioritizing by measuring severity.
Small offences will get deleted and be given a miner slap on the reputation. But we can't entrust anything bigger to an algorithm at least until we develop some great AI down the road.
Thanks for stopping by and commenting.
I was thinking about it and also about @steemcleaners and @cheetah - Bayesian algorithms are enough to track most of the abuse as you said. NOOO - this is not AI. These days everything is AI but as you said blocklists and tools like spamassassin which is used to catch spam can help a lot.
I had written about it in 2004 (yea, 13 years back) If quickly look, the article here, you will see that it already talks about proof of work in section 4.1
https://www.academia.edu/3888983/Spam_I_hate_it_so_do_you
Such an approach will make sure that cross posting is not frowned upon as well.
Your post is really good and its thought provoking. Do check the old article and the proof of work mentioned there ....
I read your link. It's really great and should be able to take care of most things. Possible inappropriate posts can also be pre-identified and stored temporarily or in server cases (like multiple cursing or lengthy descriptions of violence etc.) we can not store the post at all and even give a warning/smaller punishment using algorithms.
Researching your post led me to a bunch of things like Probability interpretations, Dutch book and Quantum Bayesianism. So thanks for that. You might find the following interesting:
https://en.wikipedia.org/wiki/Quantum_Bayesianism#Relational_quantum_mechanics
https://steemit.com/philosophy/@vimukthi/relational-quantum-mechanics-solution-for-the-paradox-of-the-ship-of-theseus-with-a-touch-of-zen-and-cryptocurrency
https://steemit.com/philosophy/@vimukthi/breaking-down-this-statement-is-false-a-solution-to-another-paradox-from-vimukthi