Literature survey on incentives and mechanisms to create trust

[synctext]

Towards msc thesis project on: fault-tolerant decentral market with proper security.

Read Torcoin, Lira, Multichain.

Storyline: tragedy of commons, I. PD, emergence of cooperation

2003 papers with 'solutions'.

[synctext]

This work prepares this quarter for a thesis on crafting a decentral market. As such, knowledge from threats and defenses is vital. Given central research question: how to build a trusted and reliable service out of unreliable and possible fraudulent donated resource?

This divides up in: building trust, reputation systems, currency, attack-resilience, sybil attack, and market fabric.

Starting points: Classical MojoNation paper: www.cs.kent.edu/~javed/class-FP2P10S/papers-2006/mojonation.pdf First ever documented collusion attack: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.84.7582&rep=rep1&type=pdf Fantasy designs (no security framework) from 2004 with 2000+ citations: www.cs.berkeley.edu/~adj/publications/paper-files/tapestry_jsac.pdf Excellent reflection of Tor crowd on their failure to create a cooperation currency: https://blog.torproject.org/blog/tor-incentives-research-roundup-goldstar-par-braids-lira-tears-and-torcoin Pioneering 2003 work on incentives "stamp trading": https://www.cs.ox.ac.uk/people/andy.twigg/pubs/2003-trading-trust.pdf

Silk Road measurements, showing honesty amonst druglords: https://arxiv.org/abs/1207.7139 in-depth 17-page decentral market follow-up study: https://www.usenix.org/system/files/conference/usenixsecurity15/sec15-paper-soska-updated.pdf MIT fantasy design (Sybil-resistant?) of a decentral market: https://eprint.iacr.org/2016/464.pdf Decentral market security implications and zero-day trading: http://www.aaai.org/ocs/index.php/IAAI/IAAI16/paper/view/12041/12326 Active Reddit community: https://www.reddit.com/r/DarkNetMarkets/comments/4ju0b1/beaver_a_decentralized_anonymous_marketplace_with/

[synctext]

Next steps:

• define scope
• find just titles for 6 sub-topics, aim for 30+ papers (not yet reading)

[synctext]

Possible title. state-of-the-art in self-regulating decentralized markets

Our idea from 2007: bandwidth as a currency. This has been the goal of the Tribler team at Delft for many years, create the first fully operational marketplace with high trust.

Interesting Papers: https://scholar.google.com/scholar?q=decentralized+matching

And more references:

PeerMart: Secure decentralized pricing and accounting for peer-to-peer systems Google: "Matching Through Decentralized Markets - Penn Economics" http://www.aaai.org/Papers/Symposia/Spring/1996/SS-96-02/SS96-02-002.pdf http://link.springer.com/article/10.1007/s10660-010-9056-y https://cs.uwaterloo.ca/~rckerr/KerrCohen-TREET.pdf www.mathcs.emory.edu/~lxiong/research/pub/xiong02building.pdf http://link.springer.com/article/10.1007/s10660-008-9014-0 https://www.computer.org/csdl/proceedings/hicss/2001/0981/07/09817008.pdf http://www.jstor.org/stable/10.1086/322087?seq=6#page_scan_tab_contents

7734 lines of Python code (decentral market with security weaknesses): https://github.com/Tribler/tribler/pull/2233

[synctext]

Survey .MD in progress: https://github.com/basvijzendoorn/literaturesurvey

[synctext]

Survey .tex in progress: https://www.sharelatex.com/project/57ea2058dee100682181ac73

ToDo:

• IEEE style in 2 columns, format like a paper
• Scope and storyline: fault-tolerant decentral markets
• sections:
  • intro
  • problem description (how to make a decentral market)
  • System model and architecture (DESCRIBE ONLY: trust, reputations, matching engine, attack-resilience{Sybils}, incentive alignments, price discovery mechanism, iterative prisoner dilemma, two-sided markets)
  • Operational markets and decentral designs
  • Cybercurrency markets
  • Security considerations
  • Reputations and incentives
  • the file sharing case
  • the Tor anonymity case

[ghost]

Lit_survey.pdf

All markets today have a centralized node.

[ghost]

Solution: SOA, Blockchain, microservices. Two sided markets. http://dspace.unive.it/bitstream/handle/10579/7203/830275-1190055.pdf?sequence=2 Trust in Uber en Airbnb. Strategy proofness Darknet as an example market.

[synctext]

Current 3-page Lit_survey.pdf

Goal: 6 pages.

More academic fantasy designs for markets: based on decentralized network resource management

Table of content suggestions: \section{detailed market examples} Include matching engine (brief?) explanation + illustration from real-world. For instance, the orderbook, and bid/ask market depth from a liquid market as Bitcoin :

\section{mechanisms to create trust} 15 year old founding work

\section{current fraud schemes on online markets} Potentially 125 years in jail Excellent discussion of the eBay problem: no incentive to punish scammers!

\section{requirement analysis}

• sybil resilient
• strategy-proof
• distributed matching engine

[ghost]

Challenges on decentralized markets. Creating an online decentralized trust system. Why trust is important in markets and business. Failures and what problems causes these failures and successes in trust systems and what we can learn from them. Additional requirements in a decentralized market.

[synctext]

Lit_survey (4).pdf Converging to a final version. Keep writing tight and no double explanations of trust please.

• intro: explain the 22 years of history.
  • An algorithm automatically rates you in Uber and score between 4.0 and 4.6 score means real job trouble. Fired automatically and you need to be looking for a new job when dropped below 4.0 score in past 500.
  • eBay 22 years history (laser point story etc).
  • airbnb more hotel-like rooms then Mariot plus Hilton combined or something.
• operational systems
  • ecommerce (eBay, uber, airbnb)
  • Silk Road: honesty among drug dealers (90+% of deals have 4-5 star rating)
  • Bandwidth-as-a-currency: the 12 year Tribler struggle (ongoing effort of research team)
• 15 years of academic fantasy designs
  • 15 year small intro overview; 2002 paper, etc. tragedy of commons, repeated interactions,
  • System models and architectures (DESCRIBE ONLY: trust, reputations, matching engine, attack-resilience{Sybils}, incentive alignments, price discovery mechanism, iterative prisoner dilemma, two-sided markets)
  • Incentives for privacy enhancing technology (Tor attempts: LIRA, Torcoin, etc.)
  • often clean-slate designs, totally ignoring the past 22-years of operational systems
  • flawed or including magic "create trust" oracle black boxes (Beaver MIT)
  • An “Oracle” is a particular agent that is trusted by the participants in the ecosystem/blockchain.
• Sybil attacks and security of markets in general
  • also strategy proofness
  • ddos on matching engine, etc.
• Combine with: https://github.com/Tribler/tribler/issues/2559#issuecomment-274798560

Include real picture of architectures of 2002 paper, and beyond. [taken from PeerTrust]

[synctext]

Interpol red list, "FBI wanted" illustration of fraud on eBay: https://nakedsecurity.sophos.com/2013/10/29/fbi-hunt-for-seven-fugitives-involved-in-multimillion-dollar-ebay-car-scam/

Beaver not viable yet: survey shows that secure DHTs do not exist and are perhaps impossible to build no shortcut possible, need to solve the trust problem..

[synctext]

Conclusion draft:

Markets have developed significantly in the past centuries. The period 1050–1330 saw a remarkable multiplication and spatial diffusion of formal markets and fairs throughout Europe, accompanying growing population, production, and exchange. Our systematic overview of decentralized markets show the vital role of trust in the online age. After numerous proposals from 2002-2017 to create trust in markets we now have initial success stories. Sybil attacks.. etc.. Lit_survey (5).pdf

[synctext]

This 12-page survey with [65] citations has now been completed!

The challenge of decentralized marketplaces Published on arXiv.org also.

[synctext]

Ooops, on that got missed: Decentralized Prediction Markets | Augur Project https://augur.net/

[devos50]

Great explanation of how reputation in Augur works: http://augur.strikingly.com/blog/what-is-reputation

TLDR: Reputation (REP) tokens are only used to predict outcomes and are not required for waging. Each holder of REP tokens can vote proportionally to their amount of possessed tokens. Honest nodes get rewarded with a small fee for a correct vote; liars lose REP tokens which will be distributed amongst honest nodes. These tokens are exchanged by the platform and can be bought by users.