Big Data analytics, AI and Machine Learning (ML) have seen great advancements in recent years. Google’s Alpha Go beat the reigning human GO world champion (one of the world’s most complex games), voice recognition and assistants such as Siri, Google Assistant and Cortana have started to be highly effective, and insights provided by AI save companies billions of dollars every year. However, despite impressive growth in the AI industry, many bottlenecks exist to prevent the industry pushing onwards. One of the biggest issues concerns the acquisition of the data necessary to train the machines. Bottos (BTO) aims at fostering developments in the AI industry by creating a decentralized data marketplace to allow AI developers to quickly gather the data needed to train their algorithms.
To respond in a certain way (produce an output), AI models need to be injected with vast amounts of data points (thousands, if not millions). Each time a model is presented with new inputs, the machine computes them and provides an output. The more inputs the AI receives, the higher the likelihood of the accuracy and predictability of its response will be. Therefore, it is of primary importance for companies to have access to a large amount of relevant data, otherwise the robot will not return the expected (or wished) output. If not enough data is collected (or of high enough quality), the AI will be prone to error, and therefore will not reach a commercial standard; for self-driving cars, for instance, errors can be deadly, or for algorithmic trading, extremely costly.
The problem currently is that most of the available data is owned by a few big players (Google, Amazon, Baidu, Tencent, etc) which do not share it – making it extremely difficult for start-ups and SMEs to develop efficient machines. Bottos is a project that aims at changing the ecosystem. To that purpose, they will create a decentralized data and AI model marketplace – where companies request data and the “crowd” provide it to them – to enable easy and efficient data acquisition and computation for all – from micro-enterprises to major organisations. In later stages, the platform will also host a storage and computer power-sharing network.
The use of the blockchain has many benefits in handling data. Firstly, the data cannot be altered, as once sealed into the blockchain, the data cannot be changed and remains the same forever. Secondly, blockchain brings privacy due to encryption. This leads to the data being secure, as smart contracts allow the separation of data ownership and user rights. Thirdly, as blockchain in itself is hacker proof even if some nodes are attacked by hackers, the remaining nodes are securing the network, thus ensuring that no-one can hijack the blockchain and steal the data. Last but not least, blockchain enables low transaction costs, essential to enable micro-transactions as explained in our blockchain in the Internet of Things (IoT) article.
The network consists of 3 user types: Requestors, Validators, and Providers; with the Requestors paying other members in BTO tokens. The demand for the BTO token will be driven by data purchasers; who will pay in BTO to obtain the data or to request services (such as data clearing). Moreover, the network will allow Data Validators and Data Providers to earn BTO tokens if they run the node on their computer.
- Data Requestors – Project developers who need the data, and will pay other members of the platform in the form of BTO.
- Data Providers – The people who will provide the unstructured, raw data, and will earn some BTO when their data is acquired by the requestors.
- Data Validators – Who will not only clean the data, but also verify and structure it. Will be paid in BTO.
Ref: Bottos whitepaper.
To ensure the veracity and quality of the data, the platform will also incorporate a credit system. Basically, each time a transaction is performed, the quality of the data is being reviewed by validation nodes and by the data requestor and given a credit score. This system ensures that requestors receive only accurate and necessary data.
Another interesting usage of the token is that teams will be able to release their own token. This system works the way a crowdfunding website would. Companies release tokens and data providers receive tokens that appreciate with the value of the AI model being built with that data.
In the future, the team intends to build a complete layer of AI (storing the data, sharing the data, etc). For instance, in the case of the shared computational power network, companies would ask the network to perform certain computationally-intensive tasks. This mechanism would decrease costs for companies while increasing their computational power.
Ref: Bottos’ Medium
Team and Partnerships
Bottos is run by Tingting Wang and Xin Song. Xin has more than 10 years in the investment and digital transformation fields, having been the former head of the China division of Droege, an investment and consulting company focusing on SMEs. Tinting was Chief Marketing Officer of the NEO platform and created a Robot Exoskeleton R&D company called Rivexo. The fact that the leadership has years of experience in not only the AI ecosystem, but also in business development, is really a value-added to us as they not only have industry insight but they can also execute.
The AI Architect, Zhen Gao, is editor in chief of two AI and robotics journals, Chao Wang, the CTO, used to work at Wanxiang Blockchain Labs, and was the Solution Architect at Huawei Technologies Co. for cellular networks (4g and 5g), which is the area he mainly worked in during his career (including 11 years at ZTE). Lastly, Gao Zhen is an assistant professor in Engineering Practice and Technology at McMaster University.
The team had contact with Huawei (probably due to Chao Wang’s connections); however, nothing tangible has officially been released yet. The only official source mentioning the rapprochement states a “potential future partnership” – meaning that as of now, nothing has been signed. Additionally, Bottos is also officially partnered with High Performance Blockchain (HPB) but not much has been said regarding what they intend to do despite marketing collaborations.
The team seems knowledgeable, however, the lack of partnerships (especially compared with the competition, discussed in the section below) is a potential hold-up for us.
Roadmap and Advancement
Bottos’ roadmap isn’t descriptive at all, nor exhaustive. Despite the main milestone, it is hard to see where the project is at in its development. As of now, the project only lists their targeted months for platforms (V1, V2, and V3) and mainnet releases. One red flag we saw is that despite the team saying that the first version (beta) has been released, we were not able to find it. In March, the testnet (version 2 of the platform) is supposed to be released and May should be the time for the mainnet to show its face. Without being able to see the version 1 and the breakdown of the steps undertaken by the team, it is hard to be fully convinced about the advancement of the project. However, to be totally fair, the team started to describe some of its R&D progress in a bi-monthly infographic (only 1 edition for now) – we will have to see if those updates come often from now on.
The platform would be of great help for entrepreneurs, SMEs and the AI industry as a whole. However, the obstacles the company faces are tremendous and numerous. As explained in the overview, to be commercialized, AI needs to be trained with vast amounts of data. Therefore, the platform will need to attract many contributors from the start in order to provide value to companies requesting data. Until now, details on the reward mechanism have not been announced, therefore we cannot pronounce ourselves in regard to how attractive it may be to data providers. The project having the most rewarding mechanism will have a competitive advantage in terms of accelerating the network effect.
The figure below shows the Big Data marketplace revenue in billions of US dollars. In 2017, the industry generated $34 billion dollars, with IBM being the market leader – generating $1.5 billion dollars from this business segment. IBM has the supercomputer Watson – processing 67 million pages per second and delivering actionable business insights to companies – making it a direct competitor to blockchain solutions wishing to become a computational powerhouse. Moreover, IBM is known to have endorsed the blockchain ecosystem since its inception. Therefore, we expect IBM to enter, the decentralised the computational marketplace ecosystem, one way or the other – competing with Bottos – while having already immense brand awareness and reputation.
Ref: Statistica, 2018
Moreover, data powerhouses such as Google, Baidu, and Microsoft would probably not join the platform as it would dilute their competitive advantage. Moreover, the power of Google and the others is to know everything about you all at once, enabling to perform great insights and data aggregation. From the data acquisition side, companies can always turn to data aggregation companies, where companies share their data – such as Dun Bradstreet, Nielsen, Datasift, and many others.
Also, many different crypto projects operating in the data, computing power and model marketplace ecosystem exist, such as Deep Brain Chain, SingularityNet, and IOTA (read more about that on their blog), with others aiming at doing the same, further down the line, such as Matrix and CPChain.
One notable competitor is SingularityNet which aims at connecting different AI solutions together (but might offer a data marketplace later on). Its Robotics Lead is the founder of Hans Robotics, the company behind Sophia, one of the most advanced humanoid robots, and the project has more than 50 AI developers and states that 30 partners are ready to use the network at launch. Another project competing directly against Bottos is Deep Brain Chain, which has a clearer roadmap, a more experienced team than Bottos in the AI industry and, last but not least, is partnered with the NEO council.
Sophia the Robot
Ref: International Telecommunication Union
On the positive side, people are willing to share their data (see figure below) and they may, at least in Europe, have more control over them due to the General Data Protection Regulation (GDRP), a reform of the EU data protection framework, coming into effect in May 2018. However, an obstacle we see is whether or not the people have (or can provide) the necessary data. Furthermore, as mentioned earlier, to attract individuals and data providers, the platform that will offer the highest payout, will likely have a competitive advantage; therefore, the business model and reward mechanism for validating the transactions will be crucial. Unfortunately, no information regarding this aspect has been spoken of until now.
Ref: AIG, the data sharing economy
To summarize, Bottos (BTO) is a great project, with a good vision and a forward-looking team. However, despite the value the project brings, it operates in a highly concentrated and competitive industry, where the aggregation and quantity of data is crucial; therefore, in order to compete, Bottos (BTO) would need to have a huge network effect (more than other platforms evolving in other industries which can work in isolation with steady growth). The fact that one of the founders, Tingting, used to be Chief Marketing Officer at NEO is highly positive, as she will have a major role to play in attracting a user base and therefore in the overall success of the project.
Without a clear picture of the rewards provided to validate the data, we cannot infer anything regarding the network effect that the platform may create, in comparison with the competition. To attract individuals and data providers, the platform that will offer the biggest payout will likely have a competitive advantage; therefore, the business model and reward for validating the transactions will be crucial. As the team has not yet communicated details regarding this, we will “assume” for the sake of the analysis and the uncertainty, that the reward model will be average.
Will Bottos become the go-to AI marketplace? Given the competition in the space and the need for an extensive network effect, we think that Bottos has a smaller likelihood than other projects (blockchain and traditional) to succeed. Despite not investing in Bottos, we will keep an eye on the advancement of the project (especially in terms of platform usage, once released) and will review our position if there is significant progress.
Disclaimer: We do not hold any BTO.
As always, this is not financial advice. This Bottos analysis is based on personal opinion, and we invite everyone to read Bottos’ white paper, available on their official webpage, before investing. Do your own research, contact the team on their Telegram channel, and consult a financial advisor if needed.