Thoughts after reading the [/tkgshn-private/DeCartography Whitepaper](https://scrapbox.io/tkgshn-private/DeCartography Whitepaper):
-
Regarding the mechanism:
- The discussions on building a social graph and the mechanism of decentralized oracles are interesting.
- However, I felt some doubt about combining these two aspects.
- Is this not about decentralized oracles?
- In the context of decentralized oracles, I think there exists something like “Social Distance” in a conventional sense.
- While it’s presented as a social graph, it’s unclear what exactly it should offer.
- Q. Is manual tagging the optimal method?
- My impression is that if a proper mechanism is established, it should work reasonably well, but I find it not very elegant.
- The manual tagging seems to focus on creating a social graph.
- But what if there is a need for common sense in cases like annotating along a timeline?
- I found a well-articulated explanation of this in [/tkgshn-private/From Users to (Sense)Makers: On the Pivotal Role of Stigmergic Social Annotation in the Quest for Collective Sensemaking].
- However, I felt some doubt about combining these two aspects.
- The issue I thought about:
- One challenge of manual tagging is that the information may be reduced to a level where humans can interpret it as words (tags).
- When tagging something as “Public Goods,” all the nuances within that category may be lost.
- It’s a difficult problem; what would be the best approach?
- Additionally, setting tags for choices could lead to arbitrary issues.
- For instance, by excluding specific tags from the choices, it might be easy to manipulate DeCartography’s determination of plurality.
- Generally, clustering is often done through unsupervised learning, and using methods like k-means might be beneficial.- When it comes down to it, it might be risky to rely on the implicit assumption that addresses are clustered.
- One challenge of manual tagging is that the information may be reduced to a level where humans can interpret it as words (tags).
- Other ideas that come to mind:
- For instance, asking a crowd questions like “How similar are these two addresses?” and then plotting them in a high-dimensional space for distance determination could be another approach.
- In this case, it could resolve the issue of information loss down to the granularity of words.
- Additionally, it avoids the arbitrariness of preparing tags, making it better.
- (Similar to pol.is)
- I see, that makes sense. If repeated multiple times, could it create an image in a two-dimensional or higher-dimensional space? (tkgshn)
- Since the question “How similar are these two addresses?” still involves a subjective human element, I thought maybe it should be judged entirely by machines. (u7693)
- I’m curious about the sense regarding subjectivity and the concept of Democracy.
- It feels somewhat different, like where do oracles stand on this.
- So, essentially, this maps to a vector space that can be expressed in human language?
- I think it’s fine if machines can achieve the goal, but I felt that determining the “good” diversity in voting requires human-world information, which intuition tells me can’t be done without it. (blu3mo) (tkgshn)
- So, would using humans or natural language models be a means to this end?
- Using a human’s natural language model sounds great.
- However, since natural language models involve arbitrary model selection, I feel using crowds might be better than that.
- Rather than asking the question “How similar are these two addresses?”, using a natural language model created by the crowd seemed to reduce the subjective parts. (Intuition) (u7693)
- Generalizing this point here, the methodology of “creating a natural language model that eliminates subjectivity,” seems applicable and interesting in distributed systems.
- [/tkgshn-private/Can a distributed oracle be replaced by a large-scale natural language model?](https://scrapbox.io/tkgshn-private/Can a distributed oracle be replaced by a large-scale natural language model?)
- +100 (u7693)
- I also want to delve deeper into what “good” diversity in voting means. (u7693)
- That makes sense. (blu3mo) (blu3mo) (blu3mo) (blu3mo) x100
- As mentioned below, 636b156f79e1130000b1e256
- Is that really a good thing? (u7693)
- That makes sense. (blu3mo) (blu3mo) (blu3mo) (blu3mo) x100
- Well, it’s kind of meta, but: [/tkgshn-private/DeCartography Whitepaper#6369fb4009c5f20000fd83e1](https://scrapbox.io/tkgshn-private/DeCartography Whitepaper#6369fb4009c5f20000fd83e1) (tkgshn)
- Generalizing this point here, the methodology of “creating a natural language model that eliminates subjectivity,” seems applicable and interesting in distributed systems.
- Rather than asking the question “How similar are these two addresses?”, using a natural language model created by the crowd seemed to reduce the subjective parts. (Intuition) (u7693)
- However, since natural language models involve arbitrary model selection, I feel using crowds might be better than that.
- I’m curious about the sense regarding subjectivity and the concept of Democracy.
- For instance, asking a crowd questions like “How similar are these two addresses?” and then plotting them in a high-dimensional space for distance determination could be another approach.
- Q. What is tagging information used for in the first place?
- Is it about considering that “two votes with different tags are more valuable than two votes with the same tags”?
- That’s correct. (tkgshn) (tkgshn) (tkgshn)
- So, is this unrelated to social graphs?
- Is it like determining the diversity of addresses using ① social distance information from social graphs and ② tagging information?
- Is it about considering that “two votes with different tags are more valuable than two votes with the same tags”?
- The discussions on building a social graph and the mechanism of decentralized oracles are interesting.
-
On a more philosophical note:
- Understanding of (blu3mo):
- Based on the premise of [/tkgshn-private/50 actually independent thinkers are worth more than 1000 NPCs who all consume the same media and vote the same way](https://scrapbox.io/tkgshn-private/50 actually independent thinkers are worth more than 1000 NPCs who all consume the same media and vote the same way)‘s vitalik-like ideology, it asserts that “votes from individuals with different tags or distant social graphs are more important than those from close individuals.”
- Well, if you put it that way, it might make sense?
- However,
- I haven’t fully grasped it, but I anticipate that there must be various discussions among people around vitalik, and there should be a stable foundation for this assertion/ideology, so I would like to know if there are references. (blu3mo)- “The idea that diversity is good is a premise, so any questions regarding that premise should refer to that,” said someone.
- Understanding of (blu3mo):
-
For example, if vitalik (or anyone, really) were to write a lengthy explanation supporting this claim, it would be great if that could be referenced (blu3mo).
- (tkgshn) It might be quite challenging, but should we start by looking at the Ethereum Foundation?
-
Regarding the last two or three aspects of crypto:
- After NFT Summer (2020) and DeFi Summer (2021), there has been a slight downturn.
- Although DAOs have been created before, they have not been entirely decentralized in terms of ideology (including organizational structure).
- Therefore, there are aspects where project tokens resemble equity (turned into tokens) where there were discussions like “this is not just securitization.”
- Consequently, there has been a trend towards diversification, finally leaning towards “isn’t it better if it’s more humanly done?”
- A sense of returning to the roots.
- [/tkgshn-private/Constructing a multi-dimensional ID (plural id) based on Social Graphs and Soulbound Tokens to prevent the threat of “collusion” in Quadratic Funding](https://scrapbox.io/tkgshn-private/Constructing a multi-dimensional ID (plural id) based on Social Graphs and Soulbound Tokens to prevent the threat of “collusion” in Quadratic Funding)
-
It feels somewhat like that, but the argument seems very vague.
- It’s not at all clear; you won’t know until you try. That’s why I want more people to challenge it (tkgshn).
- I want to know how we can determine if something is good after trying it out (blu3mo).
- (If the evaluation method is not established, there is no point in conducting an experiment.)
- Ultimately, as a human oracle, “previously, detecting collusion in Quadratic Funding as an SAD incurred costs and time, but can this mechanism make it faster and cheaper?” (tkgshn)
- [/tkgshn-private/The cost Gitcoin incurs to prevent Sybil attacks](https://scrapbox.io/tkgshn-private/The cost Gitcoin incurs to prevent Sybil attacks)
- Oh, was the purpose of Decartography to prevent Sybil attacks? (blu3mo)
- Well, from that perspective, it seems like a strong use case. If Gitcoin’s team is interested, it’s probably best to test it out quickly (tkgshn)
- Preventing collusion for Quadratic Funding and promoting diverse voting are probably separate issues, so if there are evaluation methods for the latter, I’d like to know (blu3mo).
- For example, if two unrelated people cast the same vote,
- For the purpose of preventing collusion: it’s fine
- For promoting diverse voting: since they probably saw the same media NPC, we might want to downplay their weight.
- That’s my understanding (blu3mo)
- Yeah, I get that, but well, there are still some doubts or things I don’t understand, so wouldn’t it be best to try it out to see? That’s my consistent impression. I’m more curious about what should be done or what you think should be done (tkgshn)
- I agree with “you won’t know until you try,” but I’m curious about how to evaluate it after trying it out (especially in point 2). (blu3mo)
- (For example, if presenting research on an AI model at a conference, you would write about evaluating the model in the paper.)
- In the crypto world, it seems like the approach is to publish the results on forums and let others evaluate them, right? (blu3mo)
- Well, I don’t think that’s really the case, haha (tkgshn)
- You don’t necessarily have to evaluate it or not…
- Well, I don’t think that’s really the case, haha (tkgshn)
- Hmm, it seems like having one oracle’s data source verified first, and then Gitcoin could use it effectively, right? That’s the current feeling (tkgshn)
- Well, in that case, it makes sense to take the hypothesis to the forum for discussion.
- Flow.- “I only understand to some extent lol” (tkgshn)
- Well, in that case, it makes sense to take the hypothesis to the forum for discussion.
- I agree with “you won’t know until you try,” but I’m curious about how to evaluate it after trying it out (especially in point 2). (blu3mo)
- Yeah, I get that, but well, there are still some doubts or things I don’t understand, so wouldn’t it be best to try it out to see? That’s my consistent impression. I’m more curious about what should be done or what you think should be done (tkgshn)
- For example, if two unrelated people cast the same vote,
-
“After that, it seems like the community evaluates (or unfortunately, ignores)” (blu3mo)
-
“Yeah, I really don’t understand, and it’s also a matter of luck where you get noticed, but somehow it doesn’t seem that interesting here…” (tkgshn)
- “But is there something like ‘high expectation value’?” lol
- “I want to know more about that/express that feeling” (blu3mo)
- “Why do you think the expectation value is high?”
- “I feel like there’s an atmosphere of ‘if you provide an oracle, it will be profitable’ lol” (tkgshn)
- [/tkgshn-private/Understanding Oracle Completely](https://scrapbox.io/tkgshn-private/Understanding Oracle Completely)
- “I feel like there’s an atmosphere of ‘if you provide an oracle, it will be profitable’ lol” (tkgshn)
- “I think that’s the basis”
- “Why do you think the expectation value is high?”
- “Expecting serendipity and then implementing it seems fine,”
- “However, having an image of what to evaluate through experimentation might lead to optimal actions” (blu3mo)
- “For example, a Research Question like ‘Does the diversity of ideologies manifest in the social graph and tagging?‘”
- [/tkgshn-private/Assumption on Wallet Transactions](https://scrapbox.io/tkgshn-private/Assumption on Wallet Transactions) touches on this definition. Also, it’s mentioned in the original white paper draft (tkgshn)
- “The differences between academic methodology and crypto methodology seem interesting around here” (blu3mo)
- “A link that might be helpful in thinking about collusion is Pairwise”
- [/tkgshn-private/Pairwise coordination subsidies: a new quadratic funding design](https://scrapbox.io/tkgshn-private/Pairwise coordination subsidies: a new quadratic funding design)
- “It’s not a Sybil attack, but prevention of collusion” (tkgshn)
- “A link that might be helpful in thinking about collusion is Pairwise”
-
“For example, if we were to write a counterexample,”
- “If 99% of people think A, but 1% think B, will B be reflected in the tagging?”
- “I think the goal is to reflect minority opinions the most, so suggesting that as a tag candidate seems tricky”
- “(This is more of a critique of the system’s flaw than a counterexample, the focus is different)” (blu3mo)
- “Fundamentally, ‘A. Not reflected’ is typically associated with being permissionless where forking without permission is allowed or designing a process for objection if it’s logical” (tkgshn)
- “So, if transparency on how those judgments are made is maintained, Gitcoin could negotiate that, maybe, I don’t know”
- “Since I’m not knowledgeable about the Crypto world, it’s hard to point out good counterexamples” (blu3mo)
- “Or rather, counterexamples can only be identified through experimentation, so it’s more about presenting the key points”
- “Q. Does the diversity of ideologies manifest in the social graph and tagging?”
- “Ultimately, one might be plagued by something like a 51% attack” (tkgshn)
- “However, since various ecosystems are continuously dropping things like POAP and SBT to wallets, it’s possible to incorporate tagging from a wide range of people while sharing those”
- “(Similar to Twitter) where not having a certain number of followers or tweet content can be seen as a bot”
- “Even with a human oracle, it might be okay to classify that 51% threshold in [/tkgshn-private/Tx based automatic generated social graph](https://scrapbox.io/tkgshn-private/Tx based automatic generated social graph)”
- https://www.youtube.com/watch?v=MsMsL5v2-Ls&ab_channel=GitcoinMedia
- “This refers to Glen from radical x change, I think it mostly points to this product, so please take a look”
- “Q. Does subjectivity enter into the social graph/tagging?”
- “I wonder how this will turn out, it’s about how much the mechanism of [/tkgshn-private/think what people think](https://scrapbox.io/tkgshn-private/think what people think) will prevail” (tkgshn)
- “Q. Does the diversity of ideologies manifest in the social graph and tagging?”
- And so on
- “Or rather, counterexamples can only be identified through experimentation, so it’s more about presenting the key points”
- “It seems that a strong logical connection and counterexamples would lead to a better understanding of DeCartography’s mechanism”- Initially, the discussion around DeCartography was about determining the variables of QF when distributing funds, and it was remembered that it was considered meaningless to have votes only from certain project stakeholders.
- In other words, the concept of “good diversity in voting” would likely vary depending on the voting process.
- [Decartography#636b24b909c5f20000e3aaf7]
- In the case of DeCartography, it is considered beneficial for the voters to be involved in diverse projects.
- “If 99% of people think A, but 1% think B, will B be reflected in the tagging?”
-
@kanair_jp: It’s interesting to note that there is a growing recognition that what algorithms choose is fairer than what people choose arbitrarily. While there is much discussion on how to create trustworthy AI technology, it may turn out that AI itself becomes more trusted in society.
The assumption of treating a physical body as ‘one person’ doesn’t quite fit