The Crowdsale, What's New and What's Next

Return to site

The Crowdsale, What's New and What's Next

By Peronet Despeignes

Editor's Note: All figures and analysis provided below are preliminary. This post is likely to be updated.

The Augur Project is virtually 24/7/365. It never stops (except for occasional refueling on weekends). Now — with a refilled gas tank and a rejuvenated pit crew that has virtually ballooned by more than 2,000 new reporters (referees) —  we're just getting started.

The crowdsale was an extraordinary expression of worldwide interest and support for the Augur platform and prediction markets, but it was a means to an end, not an end in itself. The end is the deployment of an open, decentralized, robust prediction market platform powered by widespread use. 

The support and resources you've entrusted us with will be directed relentlessly to that end. Before going into more detail on that, below is a full report on the crowdsale, including the distribution of Reputation (REP) tokens. There is a lot to report, so let's get right to it.

Crowdsale Statistics

The 45-day crowdsale, which ended October 1, raised more than $5.2 million (http://sale.augur.net). It places the Augur Project among the top 20 most successful crowdfunding campaigns of all time and as the best in the cryptocurrency space after Ethereum (https://en.wikipedia.org/wiki/List_of_highest_funded_crowdfunding_projects). 

In the weeks leading up to and including the campaign, we were covered by The Wall Street Journal, International Business Times, Politico and Tech Crunch. We were named a finalist in CNBC & Singularity University's Exponential Finance XCS Challenge, an exclusive Wall Street event where ticket prices average $4,000. Within two months our "How Augur Works" video had become the most watched video about prediction markets in YouTube history with more than 150,000 views. We've received broad support from Coinbase CEO Brian Armstrong and a number of other cryptocurrency heavyweights including Erik Voorhees, Vitalik Buterin and Bo Shen.

Over the course of the campaign, we saw widespread global interest — from Oslo, Norway, to Cape Town, South Africa, and from Santiago, Chile, to Inner Mongolia, China. See below the Google Analytics dashboard screenshots of unique views, by city and by country.

Bitcoin (BTC) and Ether (ETH) crowdsale purchases came in the following waves:
15% bonus period: 
Average (mean) purchase (and standard deviation or margin of error): 
3.53165551876 (+/- 16.1878915742) BTC , 3477.75806287 (+/- 20966.1649591) ETH
Total number of transactions: 2670 BTC purchases , 284 ETH purchases
Total value: 9429.52023508 BTC , 987683.289854343324302095 ETH
10% bonus period:
Average purchase:  
5.70244987229 (+/- 44.2841315864) BTC , 574.5681616 (+/- 2220.66592877) ETH
Total number of transactions: 567 BTC purchases , 98 ETH purchases
Total value: 3233.28907759 BTC , 56307.6798367855209277 ETH
5% bonus period:
Average purchase: 
2.36504311093 (+/- 10.9087466941) BTC , 261.901177828 (+/- 536.201583122) ETH
Total number of transactions: 591 BTC purchases , 67 ETH purchases
Total value: 1397.74047856 BTC , 17547.3789145 ETH
0% bonus: 
Average purchase: 
3.72000479087 (+/- 17.8964006403) BTC , 751.105744539 (+/- 2814.05506866) ETH
Total number of transactions: 1231 BTC purchases , 127 ETH purchases
Total value:  4579.32589756 BTC , 110278.0819101943 ETH
Late buys (to be refunded):
Average purchase:
0.37484075122 (+/- 0.643699991857) BTC
Total number of transactions: 41 BTC purchases
Total value: 12.91411177 BTC

A key objective for Augur is end-to-end decentralization. There were a few crowdsale "whales," but as you can see below, the distribution was generally broad based.

BTC Contributions to Crowdsale*

Purchases of REP in BTC were made from 2,440 accounts.* The average (mean) purchase (per registered account) among them was 7.639293 BTC (or US$1854 at current prices) and the median purchase was 1 BTC (or $244 at current prices), which was also the most frequent transaction size (mode).** The overwhelming majority of accounts (more than 60%) purchased less than 2 BTC each. Fourteen accounts purchased more than 1% of the BTC total. 

The GINI Coefficient (a standard income/wealth distribution measure) of this BTC sales data is around .84, which is roughly equivalent to most major cryptocurrency distributions and is similar to the kind of power-law (Pareto) distributions found in economics, finance and nature.

The historical pattern of most cryptocurrencies (which typically start off with a small group of early-adopter enthusiasts who later sell off portions of their holdings to a wider audience), is for inequality to go down as tokens get more widely distributed away from early adopters. We expect the same phenomenon to happen here after REP tokens become available, exchangeable and tradable when the system goes live (which is currently projected to take place during Q1 2016).

ETH Contributions to Crowdsale*

Purchases of REP in ETH were made from 304 accounts. The average (mean) purchase (per registered account) among them was 3782.545482 ETH (or US$2420 at current prices) and the median purchase was 202 ETH ($129 at current prices). The most frequent transaction size (mode) was 1000 ETH. The overwhelming majority of accounts (69%) purchased less than 1000 ETH each. Twenty one accounts committed more than 5000 ETH.

The ETH distribution was considerably more concentrated than the BTC distribution (with a GINI coefficient above .9), reflecting Ethereum's relatively younger stage of development and distribution, as well as enthusiasm for the Augur platform among some big ETH holders. Augur will be one of Ethereum's first active decentralized applications.

In many cases, account holders made multiple purchases across different early-bird bonus rounds.

The general formula to determine Reputation token holdings (factoring in early-bird discounts) is: 

Click to zoom

Late arrivals of BTC to the crowdsale that did not initiate their transactions before the deadline are being refunded and removed from the sale site's Total Sale and Reputation holding calculations. As a result, estimates of reputation holdings in individual accounts should adjust.

Before we move on to the update on other developments, a reminder — the crowdsale had three main objectives:

to raise funds for completion and ongoing refinement of the software platform and associated operational, promotional and legal costs
to determine empirically, based on preferences revealed by spending actions, the selection (essentially the self-selection) of referees, those who opted to buy a financial stake in the system
to determine empirically the voting clout and percentage of all platform trading fee revenue each referee gets (the more an individual bought into the crowdsale, the greater the voting clout and revenue share they will have).

Additionally, the crowdsale also provided market validation that helped increase awareness of the Project.

What's Happened

A lot. In case you missed it, here's a recap:

Augur hedged against bitcoin volatility during the crowdsale, but not as soon (or as late) as we would've liked in retrospect. Amid significant uncertainty over bitcoin's price stability through late summer and with the urging of Ethereum's former #2 Charles Hoskinson and other members of the community, Augur hedged a significant portion of its holdings (13,500 BTC).  Via itBit, Augur sold at an average of US$227 per BTC with $10/coin borrowing cost, netting the Project $217 per BTC. In light of the bounceback that occurred after that, it was not ideal timing but it reduced our downside risk — the US$ price of BTC, instead of recovering, could've easily continued falling. Our thinking was that it was better to lose a bit hedging than take the risk of losing a lot by not hedging. We wanted to guarantee Augur would have the minimum resources required to complete its mission no matter what happened to bitcoin's US$ price going forward.
Augur worked out a general plan for addressing referee-related scalability issues — specifically, we're focused on the possibility that if the platform turns out to be as successful as we aim for it to be, that the number of prediction markets requiring referees will grow significantly faster than the population of referees and their capacity to handle them all. 
As a supplement to this plan, Augur initiated an independent R&D initiative earlier this year codenamed ARGUS, an Artificial Intelligence (AI) / Natural Language Processing (NLP) / Machine Learning (ML) referee/reporter-assistant system. If successful, it could massively expand the scalability of the system by reducing the reporting burden on referees. In other words, it could massively increase the capacity of referees to report on vast amounts of predicted event outcomes. This could be critical as our ambition is for Augur to become the world's next big financial market platform and to host thousands of prediction markets with various topics - we need to make sure referees can managably report on all of those event outcomes.
blog post: http://www.augur.net/blog/i-robot-will-be-the-judge-the-case-for-artificial-intelligenc
whitepaper: https://drive.google.com/a/augur.net/file/d/0Bx8T_M1UUwjQRXplYVFVblBmUTg/view
Very preliminary demo: http://pasky.or.cz:5500/
Augur is moving toward partnership with IBM Watson, the Artificial Intelligence division of IBM, on development of a closed-source alternative to our open-source Argus as a "referee helper/recommender" application.
Four major banks — three from Luxembourg and one from Spain — have been actively considering participation in the Augur Project, either in the form of investment, collaboration or some other form of engagement. A number of banking officials, including officials with banks based in Switzerland, Luxembourg and Spain, have either suggested or explicitly told us they had personally bought into the Augur crowdsale on their own behalf.
At least three Luxembourg-based investment funds or startup incubators have been considering investment in Augur non-profit base platform or a widely anticipated for-profit spinoff.
We've begun initial work on SageNet, a media platform to identify, validate and monetize content from the world's "Hidden Sages." The system will build the world's first objectively and empirically validated meritocracy of the world's "true experts" across various topics and subtopics, as defined on an empirical 24/7/365 basis by systematic evaluation of real-time prediction market trading performance. Augur — through the open ledger that blockchain technology offers — will allow us to see, for the first time, who really knows their subject matter well enough to be able to anticipate future developments in their areas of expertise. We plan to build a media platform around that.
Blog post: http://www.augur.net/blog/find-the-hidden-sages
The government of Luxembourg, through the Economic Ministry, has said it's open to supporting a for-profit spinoff of Augur in some as-yet-undefined capacity if it does not present any significant financial regulation issues.
In a later meeting with a panel of officers of the Commission de Surveillance du Secteur Financier (http://www.cssf.lu), Luxembourg's financial regulatory authority, the CSSF determined that Augur is not under its jurisdiction and is therefore not subject to any financial regulation. This could have significant EU-wide implications to the extent that Luxembourg's regulatory determinations historically might have an implicit "passport" quality as far as their weight across the EU.
Augur recently presented an overview of its platform at the European Commission arguing for the business and social benefits of prediction markets and explaining how they can be technically implemented as distributed decentralized applications.
The demo of our alpha version works much more smoothly and quickly now with additional functionality including account registration.
We've made our third-party API publicly available (along with a nice blueprint). Augur intends to be an open platform for many other people to build their own prediction market applications, thereby expanding the platform's reach and depth.
Ethereum (http://ethereum.org), the cryptocurrency network for building automated financial applications and "smart (automated) contracts" that we're building on, went live this summer. In a matter of weeks it has amassed more than 300GH/s in hashpower, with nodes proliferating in Europe, the US and Asia. It is already ranked 4th in value among all cryptocurrencies, after Bitcoin, Ripple and Litecoin.
Senior representatives from at least two major cryptocurrency exchanges have already approached Augur (on their own initiative) about their strong interest in listing REP tokens on their exchanges when we go live (currently projected to be around Q1 2016).

What's Going On Now

Except for occasional downtime, we never really stop building Augur: our Github
We're hiring (details are here): We're taking applications for front-end and back-end programming roles. Specifically, Javascript and Ethereum smart contract programmers. We're interested in programmers familiar with cryptocurrency and fluent in JS frameworks and Serpent.
We're looking for scholarly review. We seek extensive, rigorous assessments of the platform incentive structure and other underlying assumptions. We'd like experts in economics and finance (particularly game theory and behavioral finance) as well as computer scientists and cryptologists to review our models and code. We are looking into the possibility of submitting the entire platform to a formal peer review process. Interested? Have suggestions? Contact us at team@augur.net.
This is an update on ARGUS, our Artificial Intelligence, Natural Language Processing reporter-helper application R&D initiative being led by Petr Baudis: "Argus is currently in the early prototype stage, with the basic data management functionality, machine learning and general framework ready, and a very simple baseline that's right about 2/3 of time on our test set of questions. We also have a filter for questions that we are able to answer, which currently accepts about 1/3 of questions. We are now going to improve both of these numbers (especially the first one - when we do try to answer, we should be really reliable) - by adding more news sources besides The Guardian and New York Times, and by extending our evaluation engine with more advanced semantic methods." 

What's Next

As you can see from the above, the Augur Project is entering a more serious phase and is retrofitting to deal with new anticipated needs and challenges as effectively, efficiently and professionally as possible. Here is some of what to look for in the weeks to come:

There will be more engagement with regulators and other government officials in the US, Europe and elsewhere. 
New features will be added to the codebase, including other types of prediction markets besides "yes/no" binary (including, but not limited to, scalar, categorical and combinatorial) and options to filter by topic and sort by volume.
A budget will be prepared and there will be regular, perhaps quarterly, reporting of our financials to the general public.
We plan to have a fully featured beta version of the platform out by the end of this year, maybe Christmas
Again, we're projected release of the full live real-money version by Q1 2016, but that's a forecast that depends on what unexpected security holes third-party security audits, penetration testing and bug bounties uncover. 
REP token holdings would be released shortly before the platform goes live.

So, what should referees/reporters (REP holders) do in the meantime? Not much for now beyond spreading the word that Augur is coming, but please keep an eye out for updates. We'll be emailing all REP holders (very likely more than once) as we get closer to going live and distributing REP tokens.

The Augur team will remain available via Slack, Reddit, Twitter, Facebook, Tawk and email. We're also available, as always, to respond to any comments down below in Disqus. But as we move from the crowdsale and promotional phase and re-concentrate our efforts on the platform itself, our top priorities are to complete development, secure the platform, secure its future and deploy it.

We again thank you for your interest in, and your support of, this ongoing journey. Please check your seat belts.

Resources:

Raw data: https://drive.google.com/file/d/0B3bxHDL2cCFbaktqdmtUQVY1Mlk/view?usp=sharing
Analytical tools: https://github.com/AugurProject/token-sale/tree/master/stats
GINI Coefficient calculator: http://www.peterrosenmai.com/lorenz-curve-graphing-tool-and-gini-coefficient-calculator

What are Reputation (REP) tokens?: http://www.augur.net/blog/what-is-reputation
General Q&A: https://augur.zendesk.com

Notes:
*for practical reasons, we use unique full names as proxies for BTC accounts and use unique ETH addresses as proxies for ETH accounts
**mean is the average transaction size, mode is the most frequently occurring transaction size and median would be the transaction in the dead center of the list if you listed all transactions from largest to smallest.