MeaCORA - Crowd-solving platform for compute problems

MeaCORA is building community based supercomputing capabilities specifically targeting academic research. Through the community's own resources, the solution will allow researchers to reduce costs and expertise needed to access heavy compute problems.

Existing options require expensive hardware (servers), software and specialist personnel just to maintain the systems. To keep costs down, access to these resources are typically scheduled in a queue system, requiring prioritisation assessment of projects wishing to use the compute power. Many worthwhile projects can be delayed by this pitching process and/or don’t meet the cut because of pressure on the resources.

The MeaCORA platform provides tools to address this while maintaining low cost, expertise and unlimited compute power.

‘The world is rapidly running out of computing power’ - Satya Nadella (CEO, Microsoft) @ World Economic Forum in Davos, 2017

Corporations and individuals have more compute power than they use, and yet we are in a period where there is not enough compute power for the world's requirements. Many crave unaffordable computing power to develop their ideas into usable solutions for mankind.

The cost or expertise to set up and maintain those resources is often difficult or prohibitive. 

From medicine, environmental issues, climate models, behavioural economics, energy and artificial intelligence, we can contribute to a better future, using already existing resources.

We are primarily looking to help scientists to run computationally heavy problems at lower cost, higher speeds, and higher availability. We aim to provide research teams and amateur scientists a tool to experiment with data in ways that would otherwise be restrictive, reducing financial investment and specialist skills to access compute power.

We are also looking to engage with the public to allow them to actively participate in the solutions without needing to be a scientist. With the low barrier to get involved in these projects, we want to facilitate the generation of conversation. We believe the increase in dialogue and ease of access to resources, the development and creation of ideas that can benefit humanity will be pursued more frequently from a wider variety of sources.

MeaCORA breaks down big problems into small parts that are then distributed across a network of available devices. MeaCORA determines the optimal utilisation of the devices available to the platform and with the preferences defined by both the problem and the researchers.  Examples of preferences may be, to process only within university controlled devices, or devices that are operating within a specific geofenced area, or devices with a minimum of 2 cores and available for at least the next 2 hours.

People with computational power, including the general public with smartphones, can get involved with projects that interest them. While they provide power to these projects, they can earn credits or use apps that are integrated with the platform by any third party, such as content or games.

Each device performs a small part of the operations needed and returns its part to the platform to create the completed result. Great care has been taken to ensure both the data transmitted and the device that receives the data is kept secure and isolated from any interference. The device is only used when it has idle compute power, such as when the owner is sleeping. Account holders earn credits based on the amount of computation a device performs.

The MeaCORA platform supports multiple options for those writing the computational scripts; including on-premise installation, cloud executions and several hybrid approaches. We are looking to reduce costs and simplify access to High-Performance Computing (HPC) capabilities.

The MeaCORA architecture does away with the need to run duplicate processes to validate authentic results or network failures. Distributed computing usually needs to deal with unstable networks by distributing the same problem to multiple nodes in the hope that at least one of the packets is returned with the results needed. To combat malicious or non-malicious interference in the packet, a comparison of all the results returned is assessed and the highest number of equal results chosen as the correct result. This can waste around 50% of the computational power that is used. 

MeaCORA packets are tokenised and much smaller than existing citizen science distributed computing methods, so that we do not have to wait hours to validate packetes. As the signed results are returned to the MeaCORA distribution server, they are checked to see that the packet is still intact and as expected, otherwise, a new node picks up the package if required.

MeaCORA exists on a network of distributed internet connected devices (smartphones, PCs, servers, IoT devices) so there is no direct investment required for hardware or software ownership. MeaCORA runs on these devices only when they have an idle capacity from their main purpose. 

We have chosen JavaScript as the first supported language for MeaCORA for its ubiquity and ease of use with existing technology.

Problems can be restricted to a private group so that only devices in that group are used to process problems or parts of problems. Data or code can be limited to just running on internal devices or MDM controlled employee devices while still taking advantage of the MeaCORA platform.  This also allows for concepts such as recycling credits by redirecting credits generated by the MDM controlled devices back into the researcher account. 

Scripts can be written by anyone and future integrations to popular code repositories such as GIT enables developers to continue scripting in the tools they prefer. For now, developers can write or paste their scripts into our online code editor that supports a number of code editing features to help the developers.

There are two main sides to our equation, those who need compute power and those that can contribute power. Reducing the barriers for both research teams with compute needs and device owners to pool their resources is our primary objective. We are removing the amount of technical knowledge required to set up the projects and making it easier for device owners to contribute without any technical or scientific know how. 

MeaCORA takes away the complicated setup processes so that projects are simpler and faster to start. Costs can be further reduced by relying on existing devices teams have access to. The attractiveness of this solution for low funded teams to generate and consume their own computational needs acts as an incentive to self-promote their team for the public to join. A device owner who joins a team increases that teams computational capabilities. 

The scripts required for MeaCORA are based on JavaScript as it is one of the most used languages and often a starting language for new software engineers. Adapting existing scripts or creating new scripts for use in MeaCORA is supported by an in-browser editor and integrations with popular repositories like GIT are planned for early 2020. Scripts can be shared in the community that will help accelerate other projects. Effort can then be focused on the new problems to be solved leveraging existing efforts from other contributors.

The shaping of the solution has been informed from interviews with university faculty members and industry teams involved in computational problems. 

We are aimed at helping teams and organisations that need computational power for the projects they are working on. Within 1 year we expect the number of organisations that will be using the MeaCORA platform to be 10. To support this number of teams and the anticipated compute power they will require, we will have approx 100 teraflops of compute power available through the connected devices on the network at any given time. 

In 5 years we aim to have the necessary compute power to support 200 organisations, which we estimate to be 2,000 teraflops.

Our goal is to provide tools to solve today's hardest problems. The path to those solutions are going to come from the minds that think up the future and what we will need to know to get there. Providing the computational power to support their work, we want to encourage more experimentation, more successes and more people contributing their thoughts on how these problems can be solved. 

Much like Kickstarter provided a channel for startups to attract funding from sources other than VCs and corporations, we want to provide passionate teams a channel to access compute power that has traditionally not been available to them.

Within the next year and the next 5 years we want to:

• Adapt the admin interface so that it supports the majority of projects requirements
• Projects to have benefited from MeaCORA (financial & data)
• Adapt MeaCORA to the feedback received from community members
• Grow corporate involvement via CSR
• Grow research institution (inc Uni) involvement 
• Grow public awareness and downloads of the app - more discussion happening
• Introduce developer SDK and integrations
• Repository integration
• More scripting languages supported
• AI based decisions for energy efficient selection of compute power
• ML based assessments on scripts for identifying optimal compute distribution
• Create a draw for any device manufacturers to directly embed MeaCORA as part of the hardware distribution
• In time we are looking to create our own Cryptocurrency for the creation and trading of credits

With lower barriers to access compute resources, we are expecting new approaches and new teams to experiment and develop theories that are in some cases outside of academia.

• Refining performance optimisation for the most complex computational problems across devices at scale. 
• Organisations and the general public may be wary about an application processing someone else's data on their devices. This may deter people from contributing to the network.
• Challenges on developing trust that MeaCORA can perform the computations needed and maintain the security of data is important.
• Researchers
• Any existing projects already past the planning stage are unlikely to change approach and technology selection.
• Alterations needed to adapt existing scripts that are not in JS. There may be assumptions or misunderstandings of the effort required to use MeaCORA which may hamper the adoption of the platform.
• Existing partnerships between hardware manufacturers or cloud providers may detract from the desire to move to an alternative solution even with lower overall costs.
• In many cases sensitive data cannot be permitted to exist on devices not owned by the organisation. 
• Installing third-party software on an organisations devices (such as servers or staff PCs) would need to be approved by that organisation's security team. Organisation specific compliance and policy checks may need to be tailored before adoption.

• We are working on our distribution of computation to devices available on the network both from the analysis of code and AI used to select which devices should pick up specific problems optimally. 
• Independently verified penetration tests will be published with open communication channels for FAQs. Simple explanations about our architecture and how the security of both the transmitted data and the users device is preserved. 
• Benchmarked performance tests will be available to the public on a permanent basis. 
• We are currently talking with a number of universities and research programs so that we can adapt the platform to suit their needs while creating awareness of how the platform can be of benefit to future projects. 
• As part of our education plan we are developing use case examples as well as robust documentation on how to use and write scripts for MeaCORA.
• With the development of the commercial model and use cases for a number of research setups, the cost benefits will be more readily understood as well as longer term benefits that will benefit a future research organisation's projects. 
• A number of options will be rolled out for private cloud, public cloud and internal installations of the platform. There are a number of security options for very sensitive information that requires deeper encryption than our standard packages. 
• Each organisation is slightly different and while we discuss with corporate entities on how a MeaCORA solution would benefit them, individual POCs can compliment industry standard practices. 

2 Full-time - Founders

We have been involved with digital product development for over 18 years each, helping startups and multinational global brands develop their solutions from concept to launch across a range of industries. We have known and worked with each other for years, running our respective regional offices of multidisciplinary teams and managing industry partnerships with some of the biggest names. 

With our hands on involvement in products playing to our respective strengths, Ken in engineering and Ronan in design, our combined vision of releasing the stored potential of existing minds and technology is what led us to this point. 

Ronan Donohoe (CEO / Founder)

• MSc in Multimedia from the University of South Wales

Ken McHugh (CTO / Founder)

• MSc in Software Engineering from the University of Oxford

Core competencies are: 

• CTO positions within multinationals
• UX & Product development - Director level within multinationals
• Proven background in full life cycle development of multiple complex digital products
• Successfully raised funds for past start-up ventures
• Managed design and development teams
• Hands-on skills in both design and development
• Developed and managed partnerships with large organisations
• High business acumen
• Innovation collaboration with start-ups, SMEs and multinationals
• Jury member of annual European university innovation competitions 
• Financial and accounting
• Digital marketing

Currently working with FiftyZoo in Australia on developing partnerships with enterprise organisations. 

There are three main user segments for MeaCORA that can be defined as: those that have a compute problem, those that provide computing power and developers that integrate MeaCORA to their applications. 

The projects will typically come from academia, NGOs and GOs working in various aspects of science that require large computational power for their projects. 

Scientists upload computational scripts to MeaCORA. They purchase compute credits and deploy to the network. MeaCORA takes a service fee for each credit purchased and distributes the rest to the compute providers.

For all compute power performed by a user's device, credit is earned that can either be donated to a research group or exchanged in our marketplace. Native applications for smartphones, desktops or servers can be installed by anyone with a device. Idle compute power is then used to work on a script and generate credits. 

By licensing the platform to research bodies (primarily academia) to create their own MeaCORA network, both the generation and consumption of compute power can be self contained in the private network. This also helps address additional restrictions about what devices may be acceptable to comply with an organisations’ security policies. 

Application developers can also integrate MeaCORA into their apps so that credits can be earned for the developers, providing an alternative ‘freemium’ model for their development. 

Current funding is coming from our consultancy work in product development. Much of the consultancy work has a link with issues we need to solve for MeaCORA. Specific work on MeaCORA is driven by custom PoCs/PoTs for enterprise clients interested in utilising the MeaCORA platform. With additional support in the form of grants and awards, we can concentrate more time on R&D. We are exploring partnership deals with telcos and how specific solutions benefit them and their customers (enterprise, research and public customers).

We have three primary models that will contribute revenue to further development:

• Credits: A percentage from compute credits purchased
• Licensing: On-site licensing of the platform, partnership licensing 
• Custom work/consultancy: PoC/ToC builds, custom script development

Our main outgoings will be payouts for compute contributions to the network, R&D and marketing.

Our go-to-market is built around the development of live use cases for early adopters. This is currently underway with a research program in a US university analysing contaminants in public drinking water. We are also in talks with enterprise companies in Australia to gauge their interest and needs while testing with small user groups.

The Solve mission and values resonate strongly with what we are looking to do with MeaCORA in helping solve some of the hardest challenges in the world, in our case via computing resources. 

We are hoping that through the Solve network we can be introduced to individuals and organisations that compliment each other. In particular we are seeking those that have computational problems from academia and other Solve teams that our platform can contribute to helping. With more perspectives on what is needed from the platform we can respond and cater for those needs. 

Really, all aspects of the business we are looking to learn and get input, from design, strategy, legal, marketing and technology. Getting that input and mentorship for 12 months would be invaluable.

Should we win a grant or award, the funding would allow us to dedicate more time to develop the solution.  

• Partnering with a telecom company would provide us access to both public users and networks of unused compute power. Potential for a telecom or their enterprise clients to be involved with their CSR programs can provide an opportunity for them to both give back to the communities that form their subscribers and also be an attractor for more subscribers.
• Universities or other academic institutions that are working with computationally heavy problems such as climate modeling, engineering, genetics, physics, energy utilisation models and other research fields.
• Charities investigating new treatment methods in cancer, malaria or other diseases. 
• Wikipedia or any online publishing company that is looking to reduce the dependence on banner ads, paywalls or donation based revenue models.
• App developers including games and productivity tools that could use background processing to help solve some problems while their users get to enjoy their apps and do good for the world.

• Energy source assessment for more environmentally friendly providers of compute power, whether through low usage times in a particular area or through renewable sources such as wind and solar power that is powering the devices used (assessed through energy certificates) 
• Code analysis for better distribution. Scripts are analysed before distribution and appropriate devices for the most optimal performance on that particular computation are selected to receive that problem part.

We believe that the MeaCORA platform will open up computational research to a much larger audience than just academic institutions. By removing much of the complexity and costs involved with other forms of computational research and processing, small teams and even individuals will be able to access tools that can further develop ideas and experiments that would otherwise be out of reach. 

The simple way anyone can be involved in issues that matter to them, by downloading our app, opens up the door for conversations and thinking that we hope inspires others to develop their own unique thoughts on what can be possible for the future.

Winning this award would allow us to develop tutorials and examples on how to apply STEM related problems through MeaCORA scripts and deploy across the network. Some initial conversations have started on developing this in a primary and secondary education environment via a third party.

Utilising AI and ML, we want to improve our solution both in the computational efficiency and where the energy comes from to power those devices.

The scripts we are developing for some of our use cases rely on AI. Specifically we are using AI to analyse images to detect contaminants in water to indicate how safe it is to consume. Any third party can develop their own AI scripts for use in their research as well as commercial organisations looking to improve their products and services.

On the platform itself we are looking to develop algorithms that will review the scripts developed to determine the best distribution method. There are two primary steps in this which involve scanning the scripts and pairing parts of those scripts with available devices for the most efficient distribution (some machines are better at processing certain types of functions than others). This is intended to produce faster processing times with better energy consumption ratios. 

A second major undertaking we are looking at is using open source data on energy production sources. Here we are looking to give preferential selection of devices used that use green energy. As the production of energy throughout the day changes depending on demand, we will be utilising AI to determine what devices (and where) would be best for our compute needs.

We believe that the MeaCORA platform will open up computational research to a much larger audience than just academic institutions. By removing much of the complexity and costs involved with other forms of computational research and processing, small teams and even individuals will be able to access tools that can further develop ideas and experiments that would otherwise be out of reach. 

The simple way anyone can be involved in issues that matter to them, by downloading our app, opens up the door for conversations and thinking that we hope inspires others to develop their own unique thoughts on what can be possible for the future.

Winning this award would allow us to develop tutorials and examples on how to apply STEM related problems through MeaCORA scripts and deploy across the network. Some initial conversations have started on developing this in a primary and secondary education environment via a third party.