When I speak of I in this context I'm taking on the persona of a single or set of nodes in a decentralized environment using a spec based protocol.
I had assumed as much. I mostly just did the "we we we" thing to drive the point home for our audience (who I assume, in the average case, only "gets" about half of what we are discussing) more so than for you.
Quote
that there is where you have lost people.
Probably, but I'm not concerned. The important fact is to be sure that this isn't where I have lost you!
If "people" get lost on this concept then we just send them over to metamath.org until they "get it" ok?
Quote
Yea, I obviously had an Ah-ha moment here
Good, this is precisely the critical "Ah-ha moment" to be had.
Quote
By this point I'm removing this restraint. I'm assuming coordination with the event actor is rare and potentially biased.
Excellent. I think that constraint would've been an Achilles heel in many ways, both technical and social.
Quote
not custom puzzles but instead:
I was thinking a simple hash PoW, with low difficulty:
The server sends ( DIFFICULTY, salt, pow_challenge ) to the client.
The client iterates nonce value (nonce ← 0; nonce < inf; ++nonce) and calculates HASH(CONCAT(pow_challenge, nonce)), and when it reaches a hash that ends in DIFFICULTY, proceeds.
Easy for server to validate but less easy to perform the mining.
Work but not complicated... but has the ability to support something like automation and ever increasing DIFFICULTY. I really want to de-value cheating, for example the first giveaway we did included obvious automation. Resulting in the removal of a youtube video. The spec should be able to handle this on it's own and lower the reward.
This is a touchy subject, in some ways. On the one hand, I agree that this sort of an "embedded work" step may be absolutely necessary. (Have you followed my work with securing and re-securing Motocoin, by chance? We came to a similar conclusion!) However, I implore you not to get to hung up or stuck in the mud on this concern, and for precisely the reasons you outline below. Bots should be combated/controlled through the qualitative parameters as much as is possible.
Quote
adding this to stack: http://kjur.github.io/jsrsasign/
I hear that Kenji guy is one smart fellow. I haven't had the pleasure yet, myself, but we have some mutual acquaintances who speak highly of him.
Quote
Whoa whoa whoa there... slow down.
Hehe, yes this "let's retain a hard record forever" goal might turn the problem from a difficult task into a moon-shot task. As I've said this isn't critical, but there is huge advantage to doing it this way as I suspect (from the rest of your response) you've become aware of.
Quote
I kid, but this brings up something amazing that i hadn't thought about before.
If you're familiar with my background at all, you'll probably suspect that I had thought of this. If you're really familiar with my background, you'll probably suspect that it is the primary reason that I appeared in your world to begin with!
Quote
Yes its planned to have an action type(s) that judge quality, but up until now I wasn't planning to try and persist beyond that. In short I had assumed the content or the "creative" to use a flowery term to be a blob, some piece where the quality (while organically accurate based on consensus) was arbitrary and consensus of quality subject to change over time, making the actual "creative" content inconsequential (it could be represented as a thumbtack or pencil).
The creative content is anything but inconsequential. (I suspect that Chris might bonk you on the head for having ever thought otherwise?) The creative content is "the whole point." This is kind of like comparing proof-of-work via prefix collision with proof-of-work via prime chains or other "scientific hashing." In bitcoin, the work output itself is mostly "meaningless" but in primecoin/riecoin/gridcoin/etc the work output is potentially meaningful. For a PoA coin, we should hope that it ends up in this latter category, where the work output, itself, is of some potential direct value.
Quote
But!!! Maybe I was wrong, There is value in recording this meta data. The way you described it made me instantly think of the block chain as machine learning or neural network persisted storage. What if, the chain was able to eventually determine this quality or predict consensus on it's own? Capturing things like wordsenses, sentiment analisys and other meta information makes for a much richer "snapshot" and makes the whole chain more of a model of behavior than just a ledger of activity.
And now we begin to get into the really really juicy stuff. If you'll look back to my very early posts in your thread you might realize that this point was my "intended destination" over this whole discourse. The amazing thing about a proof is that it proves a thing. The amazing thing about a set of related proofs is that it proves not only individual things, but properties/relations over those things. Once we jump from "looking at" a single proof to assessing a structured lattice of related tautological bodies we begin to see where the real "long term" value of PoA is, and it certainly isn't in raising some facebook like count.
Again, if you are familiar at all with my background you will likely not be surprised by my interest in, and focus on, this notion.
Quote
I have no doubt whatsoever that I/we can come up with a more generalized concept of an action. I also want to keep the simple case of "use walks to xyz, user performs task, user receives reward" the ability to track outside the context of a client/server is powerful.
This "simple" case is indeed powerful, but sadly it is the furthest case from "simple." This goes back to an age old problem that, as far as I'm aware, has not been well resolved.
An example that I've been working with in my head is the idea of an action "get our logo tattooed on your body." Proving such an action is fraught with complication. As yet, I don't see a way of reasonably asserting such an action as performed without reliance on some independent mediation. Great care must be taken, here.
Quote
A friend of mine is working on just this issue related to asset naming within crypto properties. Currently leaning toward the domain name model but having discussed more experimental techniques that might be well suited for our use case.
There's certainly something to this. For some reason thinking about it keeps bringing the "Waterken web calculus" to the forefront of my mind. If the nodes, capabilities, etc within a web-calculus could be married to cryptography security (forming an "authenticated web calculus") you simply and elegantly solve some problems that have gone unmet in the "grid computing" space for decades.
This notion could be it's own whole project waiting to be started.
Quote
With time... for the sake of iterative development we will do easy cases and build from there. That being said my experience at Mastercoin will come into play, as I've watched and participated in a protocol development that had to both introduce new features, change existing features all while preserving historical consensus within the chain "don't change history"
It is only natural that this be taken incrementally. I think an ideal would be to establish a basic framework and a few example of action cases, and then to open up definition of additional action cases to the world. I'd certainly be interested in contributing a few, and I'm sure I'm not alone!
Quote
I'd like to invite you to connect any time over skype: genecyber is my user there
Ewwww skype. I don't use it, as a matter of policy. Too much bad history, there. I'm sure it is not as horribly insecure as it used to be, years ago, but my experiences with it left a sour taste in my mouth that hasn't yet faded.
IRC?
Quote
I also want to always thing of attack scenarios, one that jumps out at me would be the "mining" of someone else's action by issuing quality actions on the victim's action. decreasing the victim reward while rewarding the attacker... Other users should come along and average out the attacker's quality attacks... how do you alert a horde to do this action... Does content that is being buried become temporarily more valuable to try and determine a why and check for attacks.
I'm not so sure that I would qualify this as an "attack." In any case, the qualitative modifiers should be able to handle this "naturally."
Quote
I think at the end of the day we want meaningful, valuable action. Not mindless re-tweeting. I can write some automaton to upvote me through the roof but I don't care about the vote, the vote should be a measurement, votes that are blindly placed lower the value of the whole system. There will be some gaming, it's necessary imo to get a truly adaptive and reliable system.
I've said from the beginning that if you are successful you will, as a side-effect, just breed an entirely new type of automated spammer. One that is nearly indistinguishable from someone carrying out legitimate discourse.
I'm not so sure that it is a bad thing on a long enough curve.
