NeuroChain and Matrix seem to be very similar projects. Ahead of NeuroChain ICO, we tried to evaluate pros and cons of our approach as opposed to theirs. The underlying analysis tries to be as concise and transparent as possible.
The resulting comparison between NeuroChain and Matrix lies in the table hereunder. For a more in depth description please find the description hereunder:
| Criteria | NeuroChain | Matrix |
| Prototype | 1 | 0 (not clear) |
| Product life cycle | 1 | 0 |
| Consensus | 1 | 0 |
| Security & hacking | 1 | 0 |
| Speed and Nb. of transactions | 1 (tested) | 1 |
| Public or private blockchain | 1 | 1 |
| Data capture | 1 | 0 |
| Hardware requirements | 1 | 1 |
| Use cases | 1 | 0 |
| Partnership & coroporate integration | 1 | 1 |
| Team and expertise | 1 | 1 |
| TOTAL | 10 | 5 |
Round 1 : Code generation.Matrix would propose a way to program with: inputs, outputs, scripting from the user and voila. The scripting would be realized through a technique I suppose called natural language processing. Natural language processing is a field of algorithms that cauterizes key words and pair it together to expose a solution were counterparty keywords are found. It is not possible currently to create a full algorithm with it. Web articles mention voice recognition and scripting. The experience could be very frustrating because it requires voice recognition, syntax analysis, vocal habits, and custom transcriptions and ends up with very simple cases. All this in the same box? Wow. The tools used in CNN and RNN. It cannot do all these steps. CNN are used to make pocket cars drive by themselves. It is very hard to make it run and fine tune. We have been trying it since 1998. Never seen anything around yet. What about NeuroChain?NeuroChain is a lot more conservative on that point. A meta-language is expected that takes in charge plenty of the fuss related to the decentralized aspects. Few people really know how to code. Therefore, an interface that eases the manipulation of complex underlying blocks. Ingestion of data: one block with a few parameters. Cauterization: one block with an arrow to simulate the workflow. For more complex things, the meta-language takes all the fuss regarding transaction and main security issues. Once validated, code is generated. The result is very visual. You understand what you program does.Winner: Matrix? Code generation is known for decades. Current new level, though highly unlikely to be reachable before the end of this decade, seems to be an amazing stuff.Round 2: Product lifecycleThe code ‘reuse’, order and management is not at all tackled in the whitepaper of Matrix. By NeuroChain a program is deployed on a network. A server, virtual or not, behaves as an independent entity. Use the conventional product lifecycle approach with continuous delivery approach and you are done.For the new comers, the deployment approach will be as follow. Go on a web page, select your program, next, next, describe app, finish.Winner: NeuroChain, granted no such explanation provided in the whitepaper. Round 3: ConsensusNeuroChain identifies the best nodes available thanks to a metric. It creates a pool out of them and pick some of them at random. They receive the transactions to validate and validate them. The protocol give them a few tokens as a reward. No one pays. The overall value of the tokens is diluted through token generation. You use your token : it is good. You do not, value dilution, just as euro or dollar would. If you own the validating Bot, you get the funds. As the owner of a Bot, you do not need to pay the network though.Matrix use miners. Plain and simple as for Bitcoin and Ethereum networks. Nodes where Matrix protocol works are selected at random to compute an algorithm called Markov Chain Monte Carlo (MCMC). This would be better because a set of problems are sent to the nodes to use this power to compute real life problem. Second, they send the problems to very node and use both proof of work and proof of stack. In reality, you select a pool of miners based on the amount of tokens they hold, and apply proof of work. Technically-wise it seems absurd. You still consume a lot of energy to validate a puzzle where complexity is poorly measurable because chosen by men to feed the MCMC.Winner: NeuroChain. It brings real innovation and uses efficiently resources with less bandwidth. Round 4 : Security and hackingThis one is a very important aspect. The security is a three layer cake:
- Protocol
- Application development
- Connection with outside blockchain world
On the protocol layer both have a fairly honest approach and are both weak by the same type of attacks, with a short advantage for NeuroChain. Proof of stack may select poor nodes more frequently than NeuroChain. The bad behavior or malicious attacks are more dynamically caught through the Proof of Involvement and Integrity of NeuroChain. NeuroChain wins.On the app development, both provide a formal assessment. Draw.The connection with outside world is exactly the same with secured channel and firewalls. Draw.Winner: NeuroChain, but with a short advance. Round 5 : Speed and number of transactionsThis is a very common question on the discussions feeds. VISA is expected to manage 60.000 transactions a minute. No. In practice in exchanges 60.000 messages, not the validation of transactions. Divide this figure and you get reality.Both claim they can reach 100.000 transactions per minute in developed countries. Really reachable. Note however that the average bandwidth in the world is 0.7Mb. A small Excel spreadsheet is 0.01Mb. The performance will be greatly impacted by hardware and service providers.Winner: Draw. Any expectations will be proven by facts. Speed issues are multi factored that protocol cannot manage by themselves. Round 6: Public or private blockchainNeuroChain is by default a public blockchain. You can however have your private blockchain with NeuroChain. In practice, it is very similar to a testnet. Please note that any transaction realized on the private chain cannot be accepted to the public chain. Via API, the call to other blockchains is made available.Matrix seems to be have a different approach. It provides an inclusion with private chains, yet specifically not mentioning how. A reinforcement learning framework is stated. What for? Private chains means corporates and government agencies. A simple LDAP is enough. The exchange between private and public chain is very general.Winner: Draw. Though very shady and blurry, the possibility for integration with private blockchains seems to be possible. NeuroChain provides similar approach but with practical description. Round 7: Data captureMatrix uses peer to peer network. Same goes for bitcoin. Not much of an innovation.NeuroChain proposes multiple protocol with the best channel used dynamically between two bots to have a better use of bandwidth. Storage is expected on IPFS. Matrix does not seem to state this feature.Winner: NeuroChain due to more detailed approach. Round 8: Hardware requirementsNeuroChain runs on any device. It can be called from what are called APIs, and it may be the privileged approach in the future. The motivation lies in the fact that ML and AI required a lot of power for computation. Validation process will very likely happen on huge servers, although not mandatory.Matrix behaves the same. Any device may use it. Servers are required for validation. Not an option.Winner: Draw. Round 9: Use casesNeuroChain is a running ecosystem where the algorithms are shared and the output dynamically transferred. The first one computes best drone route, the second one manages the pictures received from the drone, the third transfer the best wildlife spot for tourists for instance. It is also a market place where people seamlessly pay on usage. The best algorithms are used above the lame ones, because less expensive on usage for instance. Use cases are limited to the creativity of the users and the machine learning, artificial intelligence and deep learning current frontiers. Although any apps can run on NeuroChain, the hardware and configuration elements shall be optimized for these tasks. Targeted use cases are meant to improve education, prevents theft, improve logistics, and many other use cases so as to leverage the power of machine learning and artificial intelligence.Matrix is using AI to generate code. The underlying applications are not meant to be used for AI. No ecosystem. At best a market place, but little is mentioned on how to do it. This vision may change in the future. While use cases are potentially limitless, like many Ethereum programs, AI and ML does not seem to be the target.Winner: NeuroChain. Use case are much more described. Vision is clear regarding ML usage. Round 10: Partnerships and corporate integrationOne of the main reasons why blockchain technology is not used in corporates include lack of use cases, little integration with industries, lack of skills in development teams.NeuroChain has already started partnerships, many of which wish not to appear on the website. The underlying motivation: use NeuroChain as a secret strategy over the competitors. In the pool we include banks, payments companies and shipping companies to name a few.Matrix has not mentioned much on this. Granted the deployment process of the applications on Matrix network, it may be possible NeuroChain has an edge, but this is very speculative,Winner: Draw. Round 11: Team and expertiseBoth teams are amazing. The founders have interesting backgrounds? Unfortunately, most communication on websites are focused on CV more than on technology in the case of Matrix. Many chiefs and managers, few makers.NeuroChain seems to have a more complete and experienced team, but few of them declare their participation to the project on their LinkedIn. More makers than managers. This can also be seen on the whitepaper, a lot more complex to read and understand.Winner: Draw.
Invest Smart, Invest NeuroChain
Photo credit: Pexel