NISHIO Hirokazu[English][日本語]

Namera Conference 3

14:00 Opening, Opening Remarks 14:10 Self-introduction one by one 14:30 @miyabi @reira 「Semantic knowledge base」 15:10 @0xShj 「Napier protocol」 15:50 @SYORA | PixeLAW 「PixeLaw」 16:30 @tonfi_yuki「Algocracy」 17:10 Suto Lab "Blockchain Trilemma" 17:50 Dive presentation 18:30 Closing remarks

Semantic knowledge base

14:30 @miyabi @reira 「Semantic knowledge base」

Since this book aims for a smooth society, it must also tackle the difficult problem of smoothing out friend and foe.

  • Smooth Society and Its Enemies," paperback edition, p340.

Is the enemy complexity or cognitive limitations?

Governing Technology is a Communication Technology

  • nishio: Glen Weyl said that John Dewey said something like "[Technology changes the way people interact, so democracy needs to change with it.

Viewing Money and Voting as Communication

Life → self-enclosed system free energy minimization

symbolic emergence (springing up from a seedbed, esp. as a result of an explosion of symbols) to do free energy minimization as a group.

image

Joint attention + naming → symbol emergence

This naming game played by two agents is equivalent to Bayesian inference of internal representations in a cognitive system integrating the sensory systems of the two agents --- "Future Society and the Boundary of Meaning" Tadahiro Taniguchi, Shigeo Kawashima, Akito Inoue

Media interaction changes people's cognitive paradigms

island cosmopolitanization

Means of Knowledge Accumulation

Four classifications of PKM tools

  • nishio: 4 functions, not all necessary
  • Kobayu: These 4 classifications are smooth too!

auto-classification automatic folder Generate context from network

  • nishio: "Generating Context from Networks" Context is...

In trying to make better decisions, we move in the direction of multifacetedness.

coding

  • function (e.g. math, programming, programing) Words that can only be communicated between friends. symbolic emergence (springing up from a seedbed, esp. as a result of an explosion of symbols)

  • nishio: I wonder if you are talking about hierarchical clustering when you say it is displayed in a phylogenetic tree by a clustering algorithm.

    • →Yes

  • nishio: When you embed it in a vector space, you are assuming a distance space, but in reality, the relationship between meanings seems to be different from the distance space.

  • Kobayu:.

    • A special case of a network is a vector space

  • Ikumi Akatsuka:

  • miyabi:

    • Emergence, or is it emergence, may be difficult. Is there interaction or convergence?
    • While it may not be necessary to actively input data to create the knowledge graph itself, we believe that conscious input is important for its governance and political use.

  • nishio: I recently read about the HNSW mechanism used in vector search engines, and it was interesting, but the point is that when you make a graph, it's not enough to just link near things, you need to link far things, and to achieve that, you need to "link near things only between randomly selected vertices". I was making distant links with

Conscious information organization wisdom of crowds governing Collective knowledge of the stomach Non-linear voting ↔ one person, one vote What humans have tried to record

  • Human Conscious Information Organization

  • Would it make more sense to use this for governance?

  • nishio: is there an inherent need to actively put (information into) PKM or not?

  • toori:

    • I remember the one where the notes are grouped and look like a Venn diagram, and the SIMPLE and SYMMETRIC 11-dimensional Venn diagram looks like a mandala.
    • A New Rose : The First Simple Symmetric 11-Venn Diagram
    • https://arxiv.org/abs/1207.6452

  • nishio:

    • I think the decision of what to include is a value judgment on the content.
    • A sense that "this is worth the effort to write" is a statement of value

  • nishio:

    • Embedding in vector space and linking on its own assumes that the embedding method is correct, but I'm not sure if that's correct.
    • That's like getting fixated on one point of view...
    • It's a blur and hard to comment.

  • miyabi:

    • In designing, it is important to consider the aspects we leave to the LLM and "how different is the meaning and context of the LLM from ours?" is important.
    • The assumption is that LMM and human understanding of symbols are different. We also consider links to be semi-automatic (suggestions).
    • We should avoid being fixed in the LLM's own perspective.

  • 0xShj:

    • Is there an inherent need to put in active or not?

    • This sounds like a great question for Tonfi, but is it just me?

  • nishio:

    • sympathy
    • I think the comment was "governance by what humans have consciously tried to record is more satisfying," and the reason it's more satisfying is because "conscious choice" is a statement of individual "values."
    • But if everything we see is recorded, we are just "a lot of cameras placed in the world.

  • shudo:

    • (It's frustrating because the quantitative information that we want to use from software (like the Semantic Knowledge Base), such as the strength of relationships between words, is definitely in the LLM, but it's only implicit and not always explicitly extractable. I wish we could all use something like the Google Knowledge Base.

Napier protocol

15:10 @0xShj 「Napier protocol」

Different countries have different senses of nation. Even grandfathers DEX in countries with low confidence in their own currency. I'm just not aware of it because Japan is peaceful.

Curve Finance

  • Interest rate swaps

  • CRV tokens become a vote in distribution decisions.

  • nishio:

    • The more CRV tokens you have, the more you earn, so does that create an incentive to compete to own them?
    • Well balanced, it seems to be converging on a monopoly of the rich...

  • nishio:

    • In "The value of voting rights for CRV tokens is hard to quantify, so valuations higher than revenue expectations are occurring."bubble Then...
    • 0xShj:
      • I know this sounds like a bubble, but this is a situation where value is being added for businesses.

  • KOJO:

    • I think you're right that the CRV token voting rights raised expectations excessively (maybe Curve is the first system where staking gets you the right to vote?)

  • toori: Let's fully understand Curve War - by ta1suke

  • renchon: Curve had an exploit happen 6 months ago and it was a lot of fun.

Napier AMM Solving the problem of low pool liquidity?

Napier gives you more options for risk management

  • Be an option for those who want to avoid risk.

View Curve War as an investment in public goods Austrian School

fixed interest rate I want to benefit from Curve War. CRV tokens are priced differently than expected earnings.

Assets with maturities are less liquid. With Curve's pool, several assets of different maturities can be connected into one. Liquidity can be shared among different assets

Longer maturities and more incentives until the ecosystem matures, then more short-term ones as it matures

Real World Assets corporate bond issuance US Bond

A means of bringing people with different risk preferences together Risk averse people seek fixed interest rates. Chasing Yield at Variable Rates

  • shudo: "people with buggy risk appetite like the ones tinkering with DeFi right now" 😁.

  • renchon: I think DeFi is easier to master by actually touching it than by thinking about it at the desk, so you might want to touch it (risk preference bug).

  • nishio: bug that estimates the value of knowledge that can only be gained by trying to touch it too high()

  • Ikumi Akatsuka: I understand that the price will be higher than the expected revenue from fees, because businesses that want to list stable coin on curve will want to do so.

    • 0xShj: exactly.

  • shudo: I'm impressed with your eye for creating DeFi for interest rate swaps > Napier

  • KOJO: There are few PJs that have access to long-term interest rates.

  • miyabi:

    • Difi, it is interesting that the fact that many people are risk averse is a problematic factor for the product.
    • Environmental bug?

  • nishio: beyond chasm?

PixeLAW

15:50 @SYORA | PixeLAW 「PixeLaw」

Web 2.5 Games centralized authoritarian rule PONZI Structure AXIE STEPN

  • KOJO: Yoshiki Okamura, I like you too!

full-on chain game FOCG

Autonomous World Three ways to make the world https://movingcastles.world/three-eras-of-world-generation

  • nishio: you mean to create a membrane to separate inside and outside?
  • nishio: "three ways to make the world" power word.
  • shudo:
  • nishio: will god create smart contracts instead of "let there be light"?

HiroHamada: The Vanishing Mediator

ECS framework instead of object-oriented

  • Permissionless addition of new components and systems

OPCraft

  • Diamonds are the currency.
  • Own a parcel in Staking
  • We built a country.

nishio: too interesting > founding of the country Kobayu: Real World Domination 0xShj: There are aspects of the meta-game that can be revolutionary!

The front end is only an interpretation.

Ikumi Akatsuka: I wonder if the intrinsic fun of a game doesn't depend on the UI (I'd like to believe it doesn't).

  • 0xShj: Maybe AW is for people who are trying out the game without relying on the UI

EVE Online

  • Tens of millions of yen war
  • Real wars are in the hundreds of trillions of dollars.

PixeLAW Minimum AW Multiple FOCG moves.

nuun: https://x.com/0xpixelaw?s=21&t=LtTSMYFrOPuIH6wKijA-ng nishio: the L in Pixel is capitalized LAW?

Ikumi Akatsuka: https://ja.wikipedia.org/wiki/セル/automaton

Ikumi Akatsuka: The slide "What's been bothering me lately" is great.

0xShj: I think funding as a public good needs to be attracted to this area from Name Enemy

  • image
  • nishio:

    • "Erotic computing," also known as "elocutionary computing," is an idea driven by human desire.

    • Erotic Autonomous World! ()

What do users enjoy?

nishio: The "world-building" part is more interesting than the implemented game itself.

nishio: tkgshn's comment: positive externalities, NumerAI, on-chain Kaggle miyabi: like letting humans do evolutionary algorithms? nishio: So there are many ways to define customers, whether you take money from people who want a trade AI as a customer, or from people who want to create a world, or from people who play games

Ikumi Akatsuka: Hurdle: Latency nishio: Uh, board games can have low latency, and I see.

0xShj: I think we need to work out a bubble-like thing to launch public goods.

nishio: I think there are lots of people who want to make board games and have lots of people play them, and there are lots of people who want to show their strength in board games Ikumi Akatsuka: I understood that blockchain in AW is like playing cards

nishio:

nishio: momentum for NAME Enemy Board Game Association()

SYORA | PixeLAW:

  • This is the NOTE mentioned at the end!
  • Please come when you have time!
  • (I believe it is a new play that can be done with AW)
  • History of Token House｜SYORA

Algocracy

16:30 @tonfi_yuki「Algocracy」

  • Yuki Toki.

  • nishio:

    • Ruled by AI, not humans! AI dictatorship banzai 🙌

    • Support, not replacement

    • Let's replace stupid homo sapiens! I love this kind of thing! ()

Democracy is growing. I think the growth rate is high in the medium to long term. Negative for quick decision making

  • Then, on the side of the dictatorship, it would be advantageous to stir up the world and create a situation where quick decisions need to be made (Kora).nishio.icon

Communication becomes a bottleneck.

  • I see a connection with miyabi's "governing technology is a communication technology".nishio.icon

legitimacy and legitimacy. justness and legitimacy. Publicness on the ground floor and publicness on the second floor Isn't legitimacy fundamental? →Convincing.

  • nishio: I just recently gave a LT on "The Concept of Convincibility, English Translation difficulties" w Words as Public Goods.

  • KOJO: I guess I'm just saying that I'm convinced by the algorithm of majority rule.

  • 0xShj: I remember a messy argument we used to have that democracy should work better when it's a festival and everyone talks a lot.

  • Kobayu: always in the minority, it doesn't seem so

  • nishio: I talk a lot, but the listener is the bottleneck, so let's use AI to replace that.

Substitution of administrative functions: enforcement within the bounds of humanly determined law Alternative legislative function: deciding on laws and budgets Alternative judicial functions: determination of legality, review of unconstitutionality

argocracy

If an AI sets the agenda and votes on it, is it legitimate?

  • 0xShj: Choose an AI that you feel comfortable with and delegate. Is it possible to give legitimacy to algocracy through something like.

  • Kobayu: Human politicians who are committed to obedience to AI, and so on.

  • high efficiency

  • Enormous amount of data

  • You can't calculate on-time.

  • Economic Calculation Controversy

  • Now wouldn't that be calculable?

  • Objective function itself is a large-degree-of-freedom problem

  • miyabi: setting the objective function EBGM.

  • nishio: I wonder if this is achieved by a smart dictator who algorithmizes his values when he is not blessed with a good successor

  • Kobayu: I don't like that!

  • nuun: naive feelings w

  • nishio: dictatorial AI-ization as perpetual youth and longevity.

  • 0xShj:

    • Is it possible to draw a tangent to algocracy from a place like neuromarketing?
    • Maybe for the transposition of a model created by working backwards from the choices made by consumers in their consumption behavior?

  • miyabi:

  • nuun:

    • I feel that algocracy can be achieved in a democracy in the form of each person delegating their vote to an AI agent.

  • nishio:

  • nuun: I think there is a very likely pattern that delegating to your own AI agent is more accurate than deciding your own voting behavior yourself.

Technicians, communicators

Board of Directors and Executive Committee Supervisory and executive branch Shouldn't the judicial branch be replaced by AI?

  • nishio: It was very interesting that the discussion was not a binary argument about whether to leave governance to AI or not, but rather, after breaking down the system of governance into several parts, it became like "leave the administration and the judiciary to die", it was like the resolution of understanding was raised, I want to think about it carefully!

tkgshn's comment.

  • AI Delegate, Pol.is2.0

  • Accelerating Direct Democracy with LLM

  • That is exactly what is being done in Taiwan.

  • This argument is a leap too far, we should look at what is being done in reality now.

  • miyabi: [/tkgshn/first-exploration Polis 2.0：邁向關鍵評論網絡](https://scrapbox.io/tkgshn/first-exploration Polis 2.0：邁向關鍵評論網絡)

  • nishio:

    • Delegates were put in place to scale communication, and decentralization is no different, but could this be solved by digital deliberation tools?"
    • is now taking place in Taiwan and elsewhere.
    • This is a note from tkgshn pointing out that

  • The support itself, people will be doing it, I just wanted to see how far the algorithm would go to replace it, and how far we could let it replace us."

What will happen to diplomacy?

  • nishio: "diplomacy" another difficult topic.

    • What is diplomacy?

  • shudo:

    • Today's DAO...
    • For example, in DeFi's DAO, most of the execution is left to smart contracts, but the management is still done by humans (with votes based on their share of the governor's tokens).
    • In the case of DAOs, which require more complex execution, the execution is eventually done by a human being.
    • Company or DAO: Execution ~ Management ~ Shareholders' Meeting(?)
    • State: Executive ~ Legislative ~ Judicial
    • There seems to be a similarity between the two.
    • 0xShj:
      • When you create a delegated democracy among DAOs, you could do something like that, and part of it could be AI.

  • nishio:

    • I don't know if the country should be assumed to be in good standing, or if membership would be smoothed out."
    • shudo:
      • We would like to rethink the framework, such as the country, and pretend that it never existed. Clean Slate Approach.
      • But since people are dependent on the resources they get from the territory (like agricultural products), it seems hard to decouple that from personal security.
    • We need a new slide on what diplomacy is all about."
    • Let's make the distribution of resources among local governments even smarter than the taxation of hometowns (...)

  • nuun:

    • If the U.S. and Russia contract with smart contracts, I wonder how enforceable the performance would be.
    • How much can the state intervene in the blockchain consensus mechanism?

  • nishio: is there any mention of democracy in homoludens?

  • KOJO: Judiciary becomes correctability?

  • nishio:

    • 'Leave it to the people to believe what they believe, trust the algorithm, it's a democracy.'
    • When choosing something, do you choose what is safe or what is interesting? ()

  • miyabi

    • Plasticity, evolutionary novelty and mutation, correctability
    • As long as we can use algorithms to decide if it's a good idea, it's democratic.

  • KOJO: sounds like safe or funny=>conservative or innovative

  • nishio: it's funny, let's release it()

0xShj:

nishio:

  • Synthetic Technocracy, one of those "21st century ideologies" like the Plurality book written by Glen Weyl and Audley Tang.
  • There are three ideologies, you know, organizing the three ideologies.

nishio:

  • The resolution needs to be increased in terms of how to use and implement specific algorithms, and I have a sense that the resolution is increasing for such cut-off points as "whether to replace the legislature" in this presentation.

nishio:

  • Phase transitions due to accumulation of unconscious frustration, interesting.

nishio:

Ryuichi Maruyama:

  • I felt that whether administrative or legislative, the issue of legitimacy would arise not only in "whether or not to substitute an algorithm" but also in the choice of "which algorithm to use". If that is decided by an algorithm, it will be infinitely retroactive, so where and how will human beings make the decision?

nishio:

  • "By all means, keep throwing interesting examples at us, we work fast and there is no question of including them all," reckless w

0xShj:

  • I felt something similar to interpreting DAOs as a way to govern something as big as google or twitter that should no longer be called a public good.
  • Even if it is difficult to integrate AI into the real world state today, the need for AI will arise as DAOs expand their sphere of influence. It is likely that development incentives will be created in a context like this. (I didn't directly answer your question.)

nishio:

  • I wonder if it will spread as a set with the DAO or if it will go from support to replacement as a DX of an existing organization, unrelated to the DAO, and I'm relatively sure it won't be DAO related.

miyabi:

  • The fear of some kind of cultural invasion depending on what language AGI is created in is similar to colonialism and imperialism.
  • What kind of literature makes what kind of AGI?
  • nishio:
    • When what is distinguishable in one culture is not distinguishable in another, an overwhelming number of LLMs who do not distinguish impose extra costs on the part of minorities who do distinguish.
    • Just like having a hard time eating halal food in the countryside in Japan.
  • miyabi:
    • This is based on the assumption that LLM is currently difficult to operate simultaneously in multiple languages.
    • The imposition of segmentation by the dominant language model. That's how you might see the importation of concepts in the Meiji period.
    • At the last Plurality Tokyo, Mr. Nishio's slide was about the multiplicity of languages and their representation systems.

blockchain trilemma

17:10 Suto Lab "Blockchain Trilemma"

Who knows trilemma! You know almost everyone!

  • nishio: I don't know trilemma!

nuun: https://x.com/web5_?s=21&t=LtTSMYFrOPuIH6wKijA-ng

nishio: "only 2 out of 3 can be done" and "the product of 3 is a constant" the latter is higher resolution

1/F

0xShj: The inverse of the fork rate is not clear enough to understand security.

  • shudo:
    • When a fork occurs, the chain (the first block) that the miners are mining is split, not 50/50, but say 80% vs. 20%. Then the total hash rate would be lower, making it more vulnerable to various attacks. So after some time, hopefully it will converge.
  • 0xShj:
    • You said, "In terms of double-spending attacks and other attack patterns, the higher the forking rate, the easier it is, so the lower the forking rate, the safer it is." I am very happy to hear that, thank you very much!

H

  • Vector of hash rate percentages P
  • propagation time
  • Assuming constant communication time between nodes is HHI
  • Simulations showed strong correlations even in the general case

nishio:

  • I got the addendum where I was wondering, "The forking rate is not a parameter that can be directly controlled, but is determined by other parameters."
  • Simulations show that this equation is nearly constant

nishio:

  • What was the attack on the one that wasn't Sellfish Mining? I can't hear it because I don't have the knowledge. SYORA | PixeLAW:
  • I think you mean double-spending attack (double payment).
  • I was wondering about security too, so I'm glad you asked!
  • I'm also curious about the wording when it's PoS and not PoW.

There are two ways to improve

  • Decrease block size
  • Optimize P

Q: Why is simulation necessary?

  • A: I don't think it's necessary, but I did it because a lot of people asked me not to.
  • Since the fork rate is actually measured, it indirectly verifies that the theoretical fork rate is correct.

Geographic distribution of nodes Relationship to collusion It could be split into separate blockchains in the East and West. When it comes to PoS?

  • Forks do not occur
  • It's a different model. Maybe we should think of collusion in a different layer.

0xtkgshn Awesome talk about blockchain trilirnma image

17:50 Dive presentation The New Science of Private Organizations Connecting to the World by Reexamining Crypto from the Goods

nishio: Talking about Funding the Commons

  • Three ideological slides came out.
  • Would it be interesting to create AI delegates of the three ideologies and allow them to vote on any allocation they want.
  • We'll have an online reading of Plurality on 1/20 if you're interested in talking about this area.
  • 1/20 Social Hack Day [/plurality-japanese/1/20 Social Hack Day](https://scrapbox.io/plurality-japanese/1/20 Social Hack Day)

nishio: distinguishing between charity and philanthropy, interesting

  • Generative mechanisms that produce public goods
  • Peter Frumkin

ryoar:

Closing Comments

  • Dr. Suto commented after being waved off by Ken.
  • I'm usually in the field of technology, it's good to be here and have my eyes opened about finance and society.
  • The lack of interaction between academia and businessmen is mainly due to the fact that they cannot afford it

nishio:

  • I am going to make a summary page again, but it will be after Monday, so I would appreciate it if you could keep posting things like, "I enjoyed this talk," or "I understood more at the reception after this!

nishio:

nishio:

HiroHamada

  • By the way, regarding what @miyabi was talking about yesterday, I submitted a manuscript to the Association for Language Processing that there is a correspondence between vector space and semantic relations for distance relations using psychological constructs. https://osf.io/nsqba/

  • Reconstructing Psychological Constructs with the OSF Language Model miyabi

  • Thank you, Mr. Hamada. I will read it! HiroHamada

  • The embedded model was not bad either, the semantic relationship and the vector space relationship seemed to have a reasonable correspondence. However, I think the LLM system can be used to constrain the vector space with a pronto, but it costs a lot of money, so the embedded model seemed to be better for everyday use!

  • https://speakerdeck.com/eumesy/chatgpt-and-intro-of-ot-for-nlp?slide=49 I thought this might be helpful as well.

    • Sho Yokoi
    • ChatGPT and Natural Language Processing / Computation and optimal transport of language meaning
    • Slides used in "Workshop OT 2023 Optimal Transport and Beyond - From Machine Learning to Thermodynamic Optimization".

  • nishio

    • Looking back at miyabi's slides, I think they are very interesting slides, but they require about 10 slides of background knowledge to understand one slide, and I feel that it is impossible for me to understand an oral presentation at that speed. I think it would be great if we could create a network-like structure using Scrapbox or a home-made PKM, and make it possible to communicate from there.
    • There's so much information that by the time Mandala of Both Realms came out, I'd forgotten the story about the two agents' naming game being equivalent to sensory system integration. w
    • LLM and human sensory system integration is hot 🔥.

Dr. Suto's diary of the day

X/Twitter

realNuun2 Found a reference to the Namera Conference in Shincho 2024.4! image image

realNuun Interesting article! Just at the after-party of the last Namekura conference, Dr. Suto told us about the word "flat world" which was apparently popular in 2005 or so, and in that context, the argument seems to be "the world is not as homogeneous as the flat world said"... image image

:

rmaruy The other day at some place (Namera Conference), I belatedly learned that "decentralised" is one of the regular translations of decentralisation. I thought that translating "decentralised ○○" as "decentralized ○○" would be another big change in image from "decentralized". "Decentralized" is too neutral, "decentralized" is too strong.

shudo He discussed the relationship between long-termism and extending the cognitive limit, which is said to be 150 people. The after-party of the #NameraConference

miyabi_00008 #Namefoekai Many people have supported me in making this presentation. It is not a measure of Pareto optimum or monetary evolution. I would like to continue to show the context of reaching a "smooth world" from a different perspective. Devote yourself to the thought technology that makes symbols, meanings, bodies, people and AI smooth.

ojoknek #Namera Conference [/tkgshn/ 初探 Polis 2.0:邁向關鍵評論網絡](https://scrapbox.io/tkgshn/ 初探 Polis 2.0:邁向關鍵評論網絡) Interesting image

0xtkgshn Argocracy is too anti-Argentine #NameraConference

realNuun The third #Nuun Conference, a group of readers of The Smooth Society and Its Enemies, is underway! image

This page is auto-translated from /nishio/なめら会議3 using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.

(C)NISHIO Hirokazu / Converted from Markdown (en)
Source: [GitHub] / [Scrapbox]