phd placeholder: identity, reputation, trust, something

[synctext]

Phd track: 21 Nov 2021 - 21 Nov 2025
First step in 4 year phd track: get paper from master thesis work.
Venues: https://github.com/Tribler/tribler/wiki/Scientific-publication-venues-for-ledger-science
After first publication move beyond strong identity to a Repucoin or other Tribler-lab interest. Ideas like generalisation of IPv8 communities into "networking fabric for social intelligence", or beyond 1988 distributed file systems into the first self-organising global commons information space 🚀

The graduate school needs a blurb of a possible phd description:

Big Tech dominates the Internet economy and forms a natural monopoly.
Alternatives based on non-profit approaches are needed, commonly called the
decentralised web. This thesis will produce innovations in the field of strong identity,
trust frameworks, self-organising systems, and commons-based leaderless
communities.

Publication options: the decentralised web storyline, see Mastodon article. co-next and IMC. Publication from almost 40 years ago about "managing the global commons", still unsolved. A big "reality gap" exists between the people that talk about sovereignty, public data commons and (darknet) technology which provides it.

[InvictusRMC]

Progress update:

• Not the best two weeks
• Restructured paper: --> SSI storyline reduced to revocation SSI
  • SSI Architecture deleted, now referring to "assumed architecture", making the revocation more suitable for generic SSI architectures (if such a thing exists)
  • Added four more entries to the related work table: Speelman, Quinten (2x), Baars (for completion's sake).
  • Changing (Trusted) Autorities concept to Issuer paradigm. Thoughts?
  • TODO: incorporate public data commons? SSI seems to not have been related to this topic yet. Perhaps next paper?
• Devised new simulation experiment: am now able to simulate 6 digit client numbers, working on tuning that up to 1 million. Want to follow UK storyline --> 400k revocations, 56 million citizens.
  • Now using DES simulation: --> no more sluggish real-(execution) time simulation, but using events associated with a cost
  • Removed randomness from simulation timings (as +- 500ms hardy has an impact over the overall number, yet might be good to reintroduce)
  • Different network topology --> using clusters with of 100 clients with 1 link between them, only within clusters graphs are complete.
  • Simulate "sporadically online nodes" by having x% of revocations dropped in a transmission
  • Simulate evil nodes (?)
  • Multiple revoking nodes
  • Graphs are not done yet 😢
  • Got access to cluster for running simulations

Next steps:

• Got some bugfixes in implementation that needs to be merged
• Want to redo smartphone test: timings are quite slow (need to use different network protocol/something smarter with resends)
• Better metric than execution time?
• Math/proofs changes: Fernando disliked the assumption of a complete graph network. I believe the math is sound (if I clarify some parts), but the assumption is the deal-breaker here. Need to incorporate/acknowledge possibility of Sybil attack/evil link in case we use a regular graph network for math portion.

[synctext]

This thesis will produce innovations in the field of strong identity, trust frameworks, self-organising systems, and commons-based leaderless communities.

Ancient problems of freeriding and Sybils are unsolved. They have proven to be exceptionally hard, unsolved by Big Tech, and are now popping up in the latest hipster topics. After 20 years we now have a new context for "freeriding&Sybil research": Web3, decentralised web, and federated machine learning. Already in 2011 (!!!) the pioneering work by Márk Jelasity proved the viability of gossip-learning "Gossip Learning with Linear Models on Fully Distributed Data". This worked also showed that the P2P problems of freeriding and Sybils are still the same in federated machine learning context. Sample of recent work:

• Sybil
  • The Limitations of Federated Learning in Sybil Settings
  • This code accompanies the paper 'Analyzing Federated Learning through an Adversarial Lens' which has been accepted at ICML 2019.
  • Code and Notebooks: Manipulating the Byzantine: Optimizing Model Poisoning Attacks and Defensesfor Federated Learning
  • FL-WBC: Enhancing Robustness against Model Poisoning Attacks in Federated Learning from a Client Perspective
• Freeriding
  • Hierarchical Incentive Mechanism Design for Federated Machine Learning in Mobile Networks
  • Free-rider Attacks on Model Aggregation in Federated Learning
  • A multi-player game for studying federated learning incentive schemes
  • A Survey of Incentive Mechanism Design for Federated Learning
  • Incentive Mechanism Design for Federated Learning: Challenges and Opportunities
• collaborative learning, gossip learning, or decentralised learning in general
  • 121 page overview Advances and Open Problems in Federated Learning by countless Stanford, MIT, Google, INRIA, etc. authors
  • Decentralized Collaborative Learning of Personalized Models over Networks

[synctext]

2021 related work: "Building a Sybil-Resilient Digital Community Utilizing Trust-Graph Connectivity" (covers all our ideas)

we explore the use of a trust-graph of identities, with edges
representing trust among identity owners, to allow a community
to grow indefinitely without increasing its sybil penetration.   [...]
Of particular interest is keeping the fraction of byzantines below
one third, as it would allow the use of Byzantine Agreement 
for consensus as well as for sybil-resilient social choice
(Shahaf et al., 2019). This paper considers incrementally growing
a trust graph and shows that, under its key assumptions and
additional requirements, including keeping the conductance or
vertex expansion of the community trust graph sufficiently high,
a community may grow safely, indefinitely.

[synctext]

Real-world trust and reputation laundering: https://restofworld.org/2022/documents-reputation-laundering-firm-eliminalia/ and the details: https://www.qurium.org/forensics/dark-ops-undercovered-episode-i-eliminalia/ HN discussion with more.
300 fake newspapers, fake DMCA, Google de-indexing attack, spoofing European Commission, etc Potential dataset partners?

There are 17,000 URLs that were apparently targeted on clients’ behalf between 2015 and 2019.

[synctext]

Related: User accounts are central to tracking people. Strong identity and web-of-trust are separate issues?
Thoughts on future platform for experiments, superapp is less unique than I originally though. commercial competition around integrated instant-messaging, wallets and "web3 browsers". Trustworthy decentralised Google replacement has no known commercial competitor (or any significant academic initiative). Its hard to do a few phd thesis chapters for 3 years with commercials competitors also doing competing innovations.

[synctext]

brainstorm {problem: selling the engineering of a unique system is not deemed scientific by reviewers; acsos2022, 6may or taas}
Our Web3 toolkit creates the critical infrastructure for identity,money,trust,data and AI. Our Web3 architecture is academically pure: it is 100% Big Tech free, does not use any third party, does not include a server, or single PC. Our Web3 stack only exists on edge devices. We demonstrate the strength of our Web3 stack with three examples:

• A Spotify alternative
• Whatsapp alternative. Conversations, self-sovereign identity wallet, verifiable attestations, Bitcoins, and digital Euro support.
• Trust graph engine. We distributed 1 billion trust records across numerous edge devices, offer trust queries for any Web3 application, and conducted extensive performance profiling. {inspired by SQLite VFS abstraction and 2017 SQLtorrent}

[synctext]

Possible thesis focus with ideal publication prospects:
"Adversarial Learn-to-rank in Web3: thriving while subject to relentless attacks"

Brainstorm: https://2022.acsos.org/track/acsos-2022-papers#About

Web3: a revival of Self-Organising Systems
Our work is designed to specifically address the following gross defect in our current
socio-economic systems: _the conservation of centralisation_
https://cacm.acm.org/magazines/2021/3/250710-the-decline-of-computers-as-a-general-purpose-technology/fulltext
Within this article we outline how the distributed Web3 movement represents the
renaissance of self-organising systems. We show operational technology with an
as decentralised as possible architecture. Based on common smartphone hardware
we demonstrate 1) how users can form a secure communication network; one which
is extremely robust and does not require any server, laptop, or other coordinating device.
Decentralisation also offers autonomy. 
{reformulate politics below into science-speak}
Our core Web3 principle is that everything is owned by the people themselves, the
direct ownership rule. Direct ownership is an inalienable right. You own your own data,
your own identity, your own money, and your own means of production. Markets,
states and megacorporations have no voice, no influence, and hold no power.
All capital is directly owned, no central bank, commercial banks, or financial
intermediaries exist.

https://eriktorenberg.substack.com/p/software-is-eating-the-world-revisited?s=r
Web3: replacing the first industry with software
We present a self-organising system. The word 'politics' rarely appears in SASO proceedings. Empowering citizens to self-organise against monopolies is deep political.

[synctext]

DAO economics

We explore the fundamentals of Decentralised Autonomous Organisations to organise the global economy. Our DAO is run exclusively using code and governs itself. We present a generalised DAO architecture around identity, trust, money, data, markets, and AI. Our experimental work provides an operational economy which proves the viability of our DAO architecture. Unique quality is that no legal entity or world government holds any special power. We successfully realised the first DAO with an academically pure level of decentralisation, no centrality of any kind exists. Our DelftDAO is democratically organised based on the principle of self-governance. We as the original original founders have no special power or influence.
DelftDAO is organised as a decentralised venture fund in which artists can freely share their work. Our core economic organising principle is direct ownership. Artists remain in control of their art. Citizens control their economy. Markets,
states and megacorporations have no voice, no influence, and hold no power. All capital is directly owned, no central bank, commercial banks, or financial intermediaries exist.
The economic primitives we rely on are permission-less rental of data storage and rental of computational power. We assume sufficient supply of storage and computation by profit seeking actors which are honest. We implemented a Tiktok-ish, Youtube-like platform for video streaming without relying on any cloud technology, Javascript, or browser technology of any kind. We show that Bitcoin can be used in a scalable manner using the Trustchain distributed ledger for full scalability as layer two. On top of the storage and computational primitives we are the first to successfully realise the economic primitive of collective ownership. A group of humans of unlimited size can own an unlimited amount of Bitcoins. No legal entity, platform, or government permission is needed. Critical to our success are our trust function and automagic investment function. Algorithmic translation of who is trustworthy, appetite for risk, and speculative investment. ToDo: proxy service, creative commons, pay-to-seed, self-expanding, git-commit-for-pay, Big Tech zero taxation habit, robot economy converging to a single Web3 DAO stack.

[synctext]

Democracy-as-a-service for Web3

To strengthen democracy we devised principles for digital self-rule and conducted a unique field experiment. We meticulously crafted a new digital community with thousands of real citizens in order to understand how to organise the global economy in a more fair, honest, impartial, and equitable manner. Self-governance of digital communities has been presented as an alternative to Big Tech. Despite decades of effort to re-decentralise The Internet, we have failed to make progress. The concentration of wealth and capital outside the reach of democratic oversight is relentlessly growing. Academics have documented Big Tech strategies for avoiding taxation and hiding profits, thereby hurting the public good[ref].

We believe designing The Internet as a business was a mistake. A privatized internet will always amount to the rule of the many by the few. Our Web3 vision is an Internet based on public interest, based on self-governance principles. A cardinal part of self-rule is the ability to adapt and grow. Without any exception, existing Web3 work uses central elements to offer their services. Existing web3 work uses central servers, committees with special power, or untrusted third parties. Our work is unique: it is based on academically-pure decentralisation. Nobody has any special powers. This provides a unique ability to grow and adapt. No permission, upkeep or maintenance is needed.
{real details now coming} Our self-ruling community is based on rules which are digitally encoded into smart contracts.
Evolving smart contracts has been an open scientific challenge.

We present a trustless on-chain governance model for refinement of these smart contracts. Any part of the community can self-organise and propose changes to the rules, collectively pool their funds, jointly commission the crafting of new smart contracts, and propose a global vote for change. When a majority approves the new rules, these smart contracts are deployed to the whole community and activated. Special property of our architecture is the defence against regulatory capture. Our serverless architecture uses mobile phones exclusively. Like Bitcoin and Bittorrent, it has no centrality. We carefully avoid the need for any bootstrap server, legal entity, or mining community to minimise our attack surface. {insert something big, like:}The United Nations is harnessing the inspirational power of a better future. The digital equivalence is a distant dream, but we believe our democracy-as-a-service can gravitate any socio-economic activity. We need democratic control and our work demonstrates viability. We need global search for a new core of democracy.

[synctext]

Self-regulating capitalism: the Delft experiments

Despite many examples in history of effective decentralized political systems, cultures and
businesses, the predominant perception of decentralization is that it is inherently fragile when
confronted with the utilitarian effectiveness of centralization. The reason for this is simple and
intuitive, one voice is more effective than many when it comes to clarity in decision making.
Uniformity in vision and clear delineation of authority are powerful tools when creating laws,
products, strategy and infrastructure because they avoid duplication of efforts and harness
creative vision towards accomplishing shared objectives. In contrast, decentralization requires
communication and compromise. Whereas in a centralized system a dissenting party may not
like the decision that has been made, in a decentralized system a dissenting party may be able
to keep a decision from being made at all.

intro from here

We present an experiment which provides irrefutable proof for the viability of a new economic paradigm. Self-regulating capitalism might be possible. Capitalism ensures economic efficiency because it is self-regulated through competition. When competition is no longer genuine a capitalist economy composed of private self-serving businesses will cease to be self-regulating. Key underpinnings of self-regulating capitalism are compliance-by-design in combination with extreme decentralisation.

We present "the Delft experiments" around a robust decentralised economy. Our economy is unique because of the academically-pure level of decentralisation, dubbed extreme decentralisation. Our cardinal design principle was to remove all central governance, to cut out any central company from the architecture, eliminate all central servers, and abolish all third parties. Similar to Bittorrent and Bitcoin our Delft experiments carefully avoids a single point of failure. Within our experiment Delft University of technology bypassed all intermediaries in finance, technology, and the music industry with a direct donation to musicians. We also demonstrate the viability of collective decision within our decentralised economy. Our experiment is focused on collective money allocation. A number of Internet volunteers donated money and participated "freely" with 100 Euro worth of value. The first 10 volunteers to successfully join the Delft experiment could vote on their favourite artist. After these volunteers joined in, we automatically doubled the available funds. The collective fund was disseminated proportional to the outcome of the democratic voting process by the 10 participating volunteers.
Our second experiment demonstrates collective decision making around adaptation and growth. We conducted a field trail of our trustless on-chain governance model. We offered a bounty to any software developer who provides our robust decentralised economy with new functionality.

Architectural principles: democratic, non-custodian, permissionless, openly auditable, trustless, self-organising, zero-server, modular, en composable (e.g. lego fit of APIs).
Related work (white paper only, nothing yet realised): https://colony.io/about-us

[synctext]

Decentralised Netflix: towards a better Internet

The Artist Investment Token is operational inside our MusicDAO. With the maturing DAO prototypes we are getting ready to add investments to our scientific portfolio together with identity, trust, money, markets, and AI. Next: investment in independent movie makers! Reading:

• https://www.kcrw.com/culture/articles/filmmakers-hit-100-million-mark-on-kickstarter {Tribler movie investment ranking}
  • https://www.kickstarter.com/blog/100-million-pledged-to-independent-film
  • https://deadline.com/2022/09/netflix-patrick-wachsberger-apple-coda-zurich-summit-1235126355/
  • https://fortune.com/2023/04/07/hollywood-economy-struggles-residual-checks-streaming-profit-cutbacks-writers-strike/ {$4 income from streamers by screenwriter Kyra Jones}
  • https://www.kaggle.com/datasets/kemical/kickstarter-projects {Kaggle dataset on Kickstarter movies}
• https://hbr.org/2022/05/web3-is-our-chance-to-make-a-better-internet {a better Internet}
• https://www.newyorker.com/business/currency/how-mondragon-became-the-worlds-largest-co-op {existing post-capitalism model}

Brainstorm, Tribler DAO roadmap for 2026: list of indy movies asking for investments. Ranking by MetaMerit of these movie ideas. Proof-of-investment badges, decentralised kickstarter in DAO form with Bitcoins.

[InvictusRMC]

Progress update

• Missed DICG deadline 😞
• Lesson learned: plan better (especially during overlapping events)
• Progress on experiments:
  • Two main experiments:
    • The main experiments are based on simple simulations of the donation receiving. The clients are based on a dataset of 10k bitcoin nodes which I scraped from https://bitnodes.io/nodes/leaderboard/. Their scoring is increased by a factor of 10 in order to have more difference in the rankings (originally only ranked from 0-10). This could be improved to be more realistic.
    • 4 graphs showcasing the relative number of donations received vs the relative number of work done after n donations:
      • 
    • 1 graph showcasing the convergence (difference between the number of donations received vs work put in):
      • 
    • 4 more graphs to showcase the committees' influence in the "bootstrapping problem". This shows the influence of committee sizes. Each client has different scoring for clients (uniformly distributed [-2, +2], bound by 0).
• TODO:
  • Finalize story line
  • Find other venue
  • Convert to 8? page paper.

[InvictusRMC]

Roadmap to a Leaderless Organisation on the Tribler Network

Goal: deploy a Decentralised Autonomous Organisation on top Tribler using MeritRank. We aim to create the first leaderless organisation on a decentralised network without global consensus or knowledge in which power is distributed through performed work rather than through wealth. Using this concept, we would be able to reward those clients that consistently perform any type of beneficial work for the network. Beneficial nodes can be provided with certain privileges (e.g., exit node priority).

November 2023

• Brainstorm about usable features for MeritRank: currently Rendezvous is the sole feature that is not yet deployed. Ofcourse, this would provide merely a view on clients that are online and is no indicator of actual useful work. Possible Tribler features of nodes: 1) Online time 2) Age 3) Metadata enrichment (voting on tags) 4) Seeding.
• Brainstorm about MeritRank deployment on Tribler network.

December 2023 - January 2024

• Finish up Rendezvous design. Derive statistics (e.g. age).
• Create crawler.
• Create test nodes.
• Generalize design for future features
• Deploy Rendezvous (?)

January - March 2024

• Implement additional features:
  • Adjust crawler to compensate for all features
  • Test nodes
  • Derive first statistics
• DAO Design: DAO within Tribler can be an implicit design. Clients are all part of the DAO by design. First DAO design can be quite trivial: simply reputation and privileges. Future design will require DAO primitives: shared resources / voting / communication overlay. Additionally, we need to solve the non-trivial issue of converting MeritRank's subjectivity to objectivity (committees(?)). Furthermore, what would Tribler vote on? What would be a shared resource? What perks could well-performing clients receive?

April - Aug 2024

• Implement DAO.
• Deployment iteration. Deploy all features and make clients subjectively score one another. Gather data from clients over several months and observe convergence of scoring.
• Run crawler extensively.
• Get useful data 😄

[InvictusRMC]

Progress update

Durable Decentralised Economics for Distributed Ledger Technologies:

• Rewrote introduction and (hesitantly) finalized story line
  • New economic paradigm for DLT (not Web3!)
  • Recentralization due to mining and staking pools --> generalization to "pooling of resources"
  • Staking pool angle is very interesting because blatant centralization is current ongoing and not yet achieved (LIDO staking pool)
  • Enough papers showcasing that few entities exist --> centralization
  • We cannot simply replace mining and staking pool (in particular mining pools) because they provide efficiency to blockchains. --> We need to design a decentralized mining pool. This needs proper marketing, but can be dubbed a new economic paradigm for DLT!
• New planned experiments:
  • Gather ETH and BTC data
  • Showcase wealth distribution in current form
    • Gini coefficient (w/ Lorentz curves)
    • Nakamoto coefficient
    • Shannon Entropy
• Apply simulation of our mechanism (code already exists in simplified form)
• Repeat metrics and showcase improvements

Other activities

• Worked on three DA exercises (this should now no longer require any effort from me for the remainder of my PhD 🤞)
• Busy with DA lab setup (we're now using IPv8 😃)
• Have been busy with HREC & TCCC
• Still planning A27 ASCI

[synctext]

Graphs of centralisation: https://www.coindesk.com/layer2/2022/04/20/is-ethereum-staking-pool-lidos-growth-an-omen-of-centralization/
Solution: competitiveness function? (not a goodness function, this is not a cosy collaborative, hypercapitalism, out-compete all for-profit firms by ruthless re-investment of all profits, win the winner-takes-all game)

Storyline: Not an alternative for consensus layer! With ledger economics we are consensus-free and replace proof-of-work + global consensus with a competitiveness function + fitness ranking + income distribution function.

2 pages of math! Prove: more decentralisation

System	NO 51% risk of central party	natural decentralisation (or global consensus-layer needed)
Mining pool	❌	❌
Staking pool	❌	❌
Ledger Economics	✔️	✔️

[InvictusRMC]

Progress update

• Reiterated on Introduction
• Reiterated on problem description (have decent ideas to formalize our problem)
• Worked further on experiments, no new graphs yet. But am hopefully close.
  • Making experiments on Bitcoin and Ethereum is difficult: not easily tracable which clients receive rewards due to pools
  • Two currents idea:
    1. I pull 12 months worth of blocks and cross reference pool payouts.
    2. I analyze the Cardano blockchain, which is supposedly much more transparent, allowing for easy analysis.
  • Still difficulties in simulating network without pools.
• Hopefully graphs at end of week.

Rowdy_Chotkan___Web3_Economics-draft-28-11-23.pdf

Other points

• Two ongoing other paper ideas: (MEV Cardano; Offline Token Transfer)
• Releasing on Tribler network
• What did you think about last weeks meeting? 🙂
• DICG?

[InvictusRMC]

Progress Update

• Progress is mostly related to the two new graphs:
  • Graph showcasing centralisation on mining pools
  • Graph showcasing geographical and therefore political centralisation
• Continuing with streamlining introduction. Am getting there, but will need a few more days.
• Note yet able to analyze the Cardano network properly, working on it. Got a good dataset from Bitcoin, however, which would allow for great mining pool analysis. But not individual clients. Labelling clients continues to be difficult. Will dive into this tomorrow. Hopefully some results at the end of the week!

Other stuff

• busy with TCCC stuff;
• helping with Distributed Algorithms;
• planning A27;
• nektarios MSc thesis; &
• HREC weekly.

[synctext]

CFP ACM Ledger Journal: https://dl.acm.org/journal/dlt/calls-for-papers

Bitcoin, DAOs Decentralization Theater https://www.coindesk.com/consensus-magazine/2023/06/08/are-daos-doomed-to-decentralization-theater/

[InvictusRMC]

Progress

• Limited progress with Christmas break and all
• Worked further on simulations: small simulation done. Currently finishing up simulation with Sybils. I pulled 5 years of block data which is distributed to clients based on their reputation score as weight.
• Came up with concept that fixes our distribution problem (not yet written up yet): when forming a block, a miner already includes the nodes selected to receive the reward. This allows miners to validate the selected nodes.

Other

• ASCI in 1.5 weeks (currently preparing)
• DA wrapping up
• Giving first lecture next week 😄

[InvictusRMC]

Progress

• Busy with ASCI, so little progress
  • --> But five credits!
• Tidied up introduction and problem statement
• Working on system requirements
• Working on simulation for introducing Sybil. Currently busy with IPv8 simulator. Thoughts?
• ACM Special Issue best fit, topic:
  • --> Incentive and consensus algorithms for Metaverse and Web 3.0

Discussion points

• Blockchain Engineering
• Yearly Progress Meeting is overdue(!)
• Master student?

[synctext]

CS people creating an economy, please read these papers and their cfp scope: ACM Transactions on Economics and Computation

[InvictusRMC]

Rendezvous Design Iteration

We simplify the rendezvous design, in turn lowering performance overhead and storage requirements. We now fully rely on the peer lifecycle. Per known peer $i$ we store the following information:

• $c_i$: Number of ended connections
• $t_i$: The timestamp at which the last connection ended
• $s_{prev}$: The previous trust score saved for the peer

Now, the trust score $s_i$ for peer $i$ at timestamp $t$ is calculated with the following formula, where $d_i$ is the duration of the current connection (if any) and $\Delta t = t-t_{i}$:

$s(t) = d_i / c_i + e$^{$-\lambda \Delta t$}$\cdot s_{prev}$

In the example below, this leads to the following scores for nodes A and B at $t=7$, with $\lambda=0.05):

$s_i$   $= s(7) $
      $= 1/5 + e$^{$-\lambda 0.5 $} $\cdot s(5.5) $
      $= 1/5 + e$^{$-\lambda 0.5 $} $\cdot (1/4 + e$^{$-\lambda 0.5 $} $\cdot s(4))$
      $= 1/5 + e$^{$-\lambda 0.5 $} $\cdot (1/4 + e$^{$-\lambda 0.5 $} $\cdot (1/3 + e$^{$-\lambda 0.5 $} $\cdot s(2.5)))$
      $= 1/5 + e$^{$-\lambda 0.5 $} $\cdot (1/4 + e$^{$-\lambda 0.5 $} $\cdot (1/3 + e$^{$-\lambda 0.5 $} $\cdot (1/2 + e$^{$-\lambda 0.5 $} $\cdot s(1))))$
      $= 1/5 + e$^{$-\lambda 0.5 $} $\cdot (1/4 + e$^{$-\lambda 0.5 $} $\cdot (1/3 + e$^{$-\lambda 0.5 $} $\cdot (1/2 + e$^{$-\lambda 0.5 $} $\cdot (1/1 + e$^{$-\lambda 0 $} $\cdot 0))))$
      $\approx 2.13$

$S_{B} = 7/1 + e$^{$-\lambda 0 $} $\cdot 0 = 7$

[synctext]

Overall thesis storyline idea. Then we have the following 4-layered trust framework. Go for straight LCN storyline with networking part.

Layer	Reputation Name	Description
Application	MeritRank	Feedback at application layer, database storage, 100k people around you. Strangers which helped you in the past and you slightly trust. GPU exchange market and lottery function
Autonomous System	ASRank	Full AS list with their ratio of sybils, attacks, and honest peers. Identify clusters of bad actors and poorly managed IP-ranges. (AS count 74,200). Note, UDP spoofing
Network	NetworkRank	Currently online peers with their network stability, responsiveness, and aliveness. database storage, 10k currently around you
Personal identity	SSIRank	Self-sovereign identity with legally binding signatures under EU law. Total 448 million unique citizens in the EU realm, with enhanced legal protection from Big Tech.

NetworkRank idea comes from assignment for master student for EBSI passport-grade EU ID.

Intro from our prior "universal trust Machine" paper
### NetworkRank: A Decentralised Trust Network
Since the dawn of human civilization, trust has
been the core challenge of social organization. Trust functions to
reduce the effort spent in constantly monitoring others’ actions
in order to verify their assertions, thus facilitating cooperation
by allowing groups to function with reduced complexity. To
date, in modern societies, large scale trust is almost exclusively
provided by large centralized institutions. Specifically in the
case of the Internet, Big Tech companies maintain the largest
Internet platforms where users can interact, transact and share
information. Thus, they control who can interact and conduct
transactions through their monopoly of online trust.
Algorithms which provide some for of trust such as PageRank
have been notoriously hard to decentralise.
Our contribution is Internet-deployment of a large-scale trust
network. Prior work on decentralised trust mistakenly conjugated
the trustworthy overlay network with the human trust graph. We
show the effectiveness of splitting those two orthogonal dimensions
of trust. We propose NetworkRank for expressing the stability and
liveliness of any participant in the decentralised network overlays.
MeritRank represents the state-of-the-art algorithm to express
human collaboration and trust with resilience against fake identities
(sybil attack). Our NetworkRank ensures a single fully connected
network graph of opinions on network stability, responsiveness, and
aliveness. This enables graph-based algorithms such as MeritRank
to flourish. Key failing of prior work was the cost of maintenance of
trust. We studied all the various algorithms proposed in the past 20
years such as EigenTrust, TrustRank, PowerTrust, SybilRank,
SybilDefender, NetworkRank and extracted ideas to address the
overhead issue. NetworkRank is the first Internet-deployed mechanism
with deep packet-level integration at the network overlay layer,
thereby adding zero bandwidth overhead. All information is kept locally
and exclusively used to boost the effectiveness of MeritRank.

UPDATE: informed ddos mitigation based on reputation

[grimadas]

A small update on the progress:

We are developing a simple crawler that establishes aggressive connections with peers in the network and logs the timestamps of their IPv8 connections. The data is recorded in the following format:

<sha1 of peer key>.csv: 
connection_start_time, connection_end_time

This format allows us to process the data as needed later on, including calculating the stability and recency scores, as well as analyzing the dynamics of these scores. Each file documents the connection to a single peer, simplifying post-processing.

We have set the response timeout to the default IPv8 value of 60 seconds. At this time, I see no reason to adjust it.

[qstokkink]

Script complete: https://gist.github.com/qstokkink/29422919c1f28a5d0774b6447a582280

Crawler started at 11:54, 4th of March 2024. Aiming for shutdown at 11:54, 11th of March 2024.

[synctext]

So for the upcoming 7.14 experimental release - 1 we aim to prepare for 2 new novel features. Also hopefully a broad code cleanup.... 👀 Both @InvictusRMC and @mg98 require Internet-deployment of their phd thesis ideas. With this step we prepare for MeritRank and decentralised machine learning deployment. MeritRank deployment is future work. So we build a fully connected graph of network stability and responsiveness. Second, pure rumour mongering of search results pairs for our operational (yet highly experimental) decentralised federated machine learning training using Google T5 transformer. So we randomly spread without any security measure at this point keyword searches and SHA1 of successfully downloaded torrents. We rely on the rumour spreading for privacy-preservation and need to further enhance this and add spam hardening in the future.

[synctext]

(btw bsc team of 5 == 4h/week)

Towards a concrete final publication

• "AI economy: the DAO financial architecture" (Yes / NO / Other)
• scope: fully automated AI economy requires a architecture for capital, investments, market design (Martijn), financial flows, transaction processing, and ... (Yes / NO / Other)
• DAO scaling is the fundamental problem, 1 billion and beyond (TikTok as storyline example with US breakup) (Yes / NO / Other)
• DAO without any single-point-of-failure, financial architecture for the AI economy (global GDP scale, daily volume: 373,468 payments €1,877 billion)
• No invoices, No shared wallet, No deterministic solution; stochastic inspired by proof-of-work (Yes / NO / Other)
• Value Function ??? (core of paper) (Yes / NO / Other)
• Related work ALL "Decentralised In Name Only", not scalable, not fit-for-purpose. (wild idea: DAO maximum performance experiment)
• DAO GPU farming, open market, start experimental section with small prototype (Yes / NO / Other)
• DAO GPU farming, DAS6 farm (Yes / NO / Other)
• DAO seedbox use-case for Creative Commons content (Yes / NO / Other)
• DAO seedbox use-case, but general fiancial architecture with also in-scope GPU farming (Yes / NO / Other)

31 July deadline - learn-by-doing progress

• Burb of text Experimental section Why/What (no How)
• 3 Figures with random genererated datapoint
• Start of experiment design process

[synctext]

Resource donations to a fully decentralised collective?
This needs a status update from @InvictusRMC 😁

Could you please think for a day about "the spam paper process" and document a post-mortum in small 8-10 text blurb. In the flow progress: 1 week of continuous writing.

[ ] DAO seedbox use-case, but general financial architecture with also in-scope GPU farming (Yes / NO / Other)

On Mar 13, 2013 issue #23 was created, "Credit investments test: donate 1 TByte anonymously". We could not close that issue for over a decade. Niels,Ardhi,Egbert, and Sandip all created running code. However, never ready for production. That experience feeds into the 5th victim 5th researcher to work on this.

update:
Financial cryptography conference: blockchain research never died there! Does Proposer-Builder Separation Preserve Decentralization?. by a16z Crypto people. Similar: A DAO need financial transaction-income separation. Just like Amazon marketplace is rulled by the A10 algorithm, Google by the infamous ranking algorithm, Youtube algorithm with meticulous recording of how long you watch, etc.
A DAO needs to have a "prize winner function" which is more flexible, faster, and agile than bank transaction, clearance, and settlement overhead.