Flyingcarpet’s Master Plan
What we’ve achieved and where we’re headed.
With plenty of assistance from our technical lead, Leopold Joy.
In 2016, Amazon completed its first drone delivery in the UK. Eighteen months later, patent filings revealed that the e-commerce giant plans to construct “beehive” towers in urban areas to act as drone fulfilment centres. This might seem far-fetched but the global market for drone business services is expected to grow to an enormous $127bn by 2020 [1].
The UK’s Civil Aviation Authority has issued permissions for 4,600 businesses, ranging from FTSE 100 corporates to SMEs, to conduct commercial drone operations for their various needs. Narendra Modi’s Indian government is using drones to monitor infrastructure projects in order to stamp out corruption and improve safety standards. In South Africa, anti-poaching organisations are using drones to protect endangered species.
These disparate drone use cases have one thing in common: they all involve the collection of vast quantities of visual data that often sits unused, meaning that valuable analytics are never extracted. Flyingcarpet’s mission is to incentivise data scientists to build models to extract analytics from this visual data and then fairly distribute the resulting value among our network participants.
Many of the world’s data scientists currently lack either the datasets to build cutting-edge machine learning models or, if they do, they operate within siloed organisations that reap most of the value that their work generates. Businesses that could radically improve their operations through the use of prediction analytics are often unable to because they can’t access the required talent, hardware and data.
Flyingcarpet is building a competition incentivisation mechanism to connect and align businesses that require analytics to inform commercial decisions with data scientists.
The journey so far (and beyond)
Q3-Q4 2017: Flyingcarpet takes shape
In August 2017, our CEO, Julien, built our first proof-of-concept: an algorithm that enabled a drone to autonomously gather visual data from a coconut plantation in Papua New Guinea. He also designed and built a solar-powered charging station that enabled the autonomous data-collection drone to recharge when needed, which rendered the entire process fully autonomous.
A 20-minute drone flight gave Namaliu, the plantation owner, all he needed to know in order to make better yield predictions and optimise fertiliser distribution. He could even monetise the analytics by selling them on a data marketplace, perhaps to a futures trader in a global financial centre who’s looking to better predict the market, if he so desired.
After this success, Julien envisaged a decentralised network that would incentivise developers to write machine learning algorithms for fully-autonomous, on-demand data-collection drone missions. In this way, data-hungry organisations would be able to obtain valuable visual analytics and, in doing so, grant hardware owners and rural communities an opportunity to build additional revenue streams.
Jane Thomason, a frequent commentator, blogger and thought leader in the applications of blockchain technology to solve social problems, joined as the first advisor to Flyingcarpet in the summer of 2017. In November, Viktor Tron, the lead developer at Swarm (Ethereum’s P2P data-storage system) joined our team as an advisor to help steer our vision.
Q2 2018: Winning at CogX and building partnerships
Julien presented the original Flyingcarpet vision at CogX, a London AI festival hosted by Fabric Ventures. Thanks to his efforts, Flyingcarpet took home the best in category award for “Best decentralised IoT-enabling project” and beat off competition from notable projects, such as Streamr and Fetch.ai.
(You can watch a video of his presentation here and a panel discussion on the future of IoT he was involved in here.)
Later, in mid-2018, we developed partnerships with several ConsenSys spoke companies. Token Foundry is assisting us with our upcoming token sale, which is presently scheduled for Q4 this year (see below). Gitcoin and the Bounties Network are providing crucial on-chain infrastructure to support the incentivisation mechanisms that we’re building.
Q3 2018: Earth Focus
We’re currently putting the finishing touches to our technical paper, which will be released imminently. The paper illustrates how our vision has changed since our early ruminations on automated drone flights. You can stay up to date with its release by joining our Telegram channel.
Although Flyingcarpet initially focused on decentralising all three stages of the aerial analytics process — data collection, analytics extraction and analytics monetisation — we concluded that our primary initial value-add is at the middle analytics layer. Therefore, we decided to focus our energies on processing the visual data itself.
It is an end-to-end solution for analytics-hungry organisations that require specific data-collection missions to be executed and analytics-extraction models to be built for their particular needs. Although this analytics-extraction layer is our primary focus, Flyingcarpet will, in the future, facilitate data collection via autonomous drones and charging stations and thus enable many different types of device owners to monetise their hardware.
To an extent, therefore, Flyingcarpet will be to visual data what Numerai is to financial data. Where the two projects differ is that unlike Numerai, which exclusively uses the generated analytics for its own hedge fund, Flyingcarpet will distribute its insights to analytics-hungry businesses. However, the Flyingcarpet team may additionally use the generated analytics itself in order to, for instance, make commodity predictions using extracted crop yield estimates.
In a nutshell, we’re doing this by leveraging three essential pieces of blockchain-based infrastructure:
- A Token-Curated Registry of Opportunities (TCRO), which will list and rank potential jobs for data scientists. The opportunities are ranked based on the amount of token staked against them — the more token staked against an opportunity, the sooner a corresponding model competition will be run. Meanwhile, bounty hunters — participants who are paid to collect raw data using drones and satellites — browse a web-based heat-map UI to discover data collection opportunities.
- Model competitions are used to produce analytics-extraction models for businesses. Competitions are run based on the current top TCRO entry. Competition reward is implemented using bounties on the Bounties Network and open-source training datasets for machine learning models are annotated by “Human Classifiers” via Gitcoin incentives. Data scientists have a fixed window of time to complete and submit their models. After a submission window is closed, models are evaluated and data scientists are rewarded based on the success of their models and how confident they were in their work.
- Nitrogen (NTN), our native staking token, is central to the model competition mechanism and is used by data scientists to stake against submitted models. The amount of NTN that a data scientist stakes against their work, in addition to a confidence metric they provide, determines the data scientist’s reward if their model is successful in the competition. Thus, data scientists are incentivised to hold NTN in order to profit from their model creation work.
It’s worth noting that businesses don’t need a complete understanding of the above infrastructure, nor will they have to set bounties themselves in order to obtain their desired analytics — this will all be made possible through Flyingcarpet’s web-based user interface.
Our present direction also enables us to open up the field to other IoT devices that gather visual data; the space industry, for example, is undergoing a renaissance of sorts and satellite data analytics are expected to be worth around $18bn by 2027 [2]. Similarly, the number of static cameras worldwide continues to grow and as do the heaps of visual data that they are amassing.
There are a myriad of use cases for Flyingcarpet, some of which have been outlined in our technical paper. To name but one, analytics from disaster zones can be used to enable underwriters to update their loss-earn ratios, and offer customers automated claims. This won’t just improve underwriting decisions, it will likely also incentivise more individuals and businesses to insure against catastrophe risk; after all, just 38% of the economic cost of environmental catastrophes in 2017 was insured.
Q4 2018: Token sale and testnet
There are, as with all token sales, two essential questions that must be answered: why is a token needed and how does it derive its value? Regarding the former, this post has already outlined the function of the Nitrogen token and more detail is provided in the upcoming technical paper.
As for the value question, it’s imperative to appreciate that Nitrogen is a staking coin and not simply a medium-of-exchange token. Ultimately, medium-of-exchange tokens derive their value from whether or not people are willing to store economic value in what is a relatively illiquid asset (when compared to USD, BTC or ETH) — something that most would rather not do over the long-term.
Nitrogen, on the other hand, will derive its long-term value from the utility that data scientists derive from staking the token against their submitted models. The amount of reward data scientists receive for successful models is determined by how much of the Nitrogen token they stake. Therefore, as demand from businesses for Flyingcarpet analytics models increases, so too will model creation opportunities, meaning that Nitrogen demand from new and existing data scientists seeking to capitalise on these opportunities will increase as well.
Stay in the loop
We’ll soon be releasing further articles that dig into the specifics of Flyingcarpet’s token economics and value flow, as well as the network’s governance mechanism. We are also putting together a series of posts on GDPR’s impact on decentralised networks — here is Part 1.
In the coming months, we intend to launch our testnet, which will include a web-based heatmap UI that displays TCRO data collection bounties. Businesses will also be able to create analytics model opportunities via bounty smart-contracts.
Join our Telegram to keep up to date with our progress, including the testnet launch and upcoming token sale.
Citations:
[1] ‘Global Market for Commercial Applications of Drone Technology Valued at over $127bn’, PwC, <https://pwc.to/2y4t3wh>
[2] ‘Big Data Analytics via Satellite, 2nd Edition’, Northern Sky Research, <https://bit.ly/2LqYRNR>
[3] ‘Global Catastrophe Recap’, Aon, <https://bit.ly/2P3iMEy>
