Sybl is a collaborative and modular programming environment that connects previously isolated computational models in one simulation space

Sybl extends popular programming and simulation engines by introducing a modular ecosystem where simulation models can be combined, increasing their resolution, accuracy and dynamic complexity. Designed around an intermediate format, Sybl displays commonalities between packages, establishing a link between different engines — augmenting the standalone software architecture paradigm. This enables real-time investigation of events within a simulation while also producing high precision datasets for downstream applications — such as machine learning and computer vision.

By introducing an open format, the platform links data markets across different disciplines with simulation tools and engines already available on the market. As different applications are chained together, changes are registered into a common reference model, converging into the same simulation space.

Sybl is designed to simplify the modeling of complex systems, facilitate the integration of different simulation engines, and enable interactions across different scales and domains.

The Evolving Role of Simulation

High performance computing is the engine enabling almost all modern science, technology, and consumer product breakthroughs. Simulations enable us to extend beyond our basic working memory, to design and build systems whose complexity would be otherwise unfathomable.

Today simulations are trusted as ‘digital twins’ to directly manage the operation of complex mechanical systems. A digital twin is a virtual representation of the elements and dynamics of an IoT device and the environment around it. A twin becomes intelligent not just by capturing the data, but by processing it through powerful algorithms that learn and grow in tandem with the physical entity. Our dependence on computer modeling only increases as real-time data demands reach new heights with 5G data networks, so the esoteric language of scientific computing needs to be transformed. It needs to come out of university basements, government agencies and corporate R&D labs, into global networks - where we can observe, evolve and govern them in the future.

Democratising Digital Twins

Despite major advances in digital modeling, render engines, and AI methods, the current architecture of most simulation isn’t able to incorporate the dynamics of an increasingly complex world. The limitations lie both in the lack of common formats that support the integration of simulation modules across different platforms, as well as the computing power necessary for large scale simulations with overlapping domains. There exists a virtuous cycle in AI research in which the development of novel environments can enable the development of novel algorithms, and vice-versa. This effect can be seen in how games and AI have been intertwined. In a similar way, the ability of researchers to create complex environments that mimic the conditions found in the real world can lead to physical spaces through with new simulation techniques can emerge. By applying the same thought process to simulation spaces, we can begin to build a picture of a metamodeling environment, where engines and algorithms can be combined, multiplied, and averaged.

The use of game engines as simulation platforms for autonomous vehicles points to the necessity of a networked platform that is able to support the instantiation of complex systems with a real time data, such as real time traffic data and pedestrian activity in the city. Game engines offer a high level of modularity, but their architecture can’t scale to the growing demand for complex simulations because of their core role as general simulation environment that only runs on one computer. The creation of a open simulation platform could rapidly increase the potential that simulations could play in the near future. These environments need to be much more sophisticated than what you would built in a game. They need to be able to ingest the real-time data for entire supply chains and ecosystems at ever decreasing timescales.

features

Collaboration platform for simulations

Simulations are run in an open environment where multiple organisations can observe and collaborate to adjust their parameters.

Visual metaprogramming

Sybl bridges visual programming tools from existing modeling and rendering packages (Maya, Unreal etc) to bridge software ecosystems across disciplines

Real-time data markets

Core to Sybl's integration methodology is the use of markets for sharing and licensing models, datasets, real-time feeds visualizations through a common data format capable of converging results across multiple scales and disciplines.

Overlapping address spaces

Simulations in Sybl operate within shared addresses, enabling them to directly interact with each other, sharing data and adjusting to predicted events.

Distributed governance tools

Sybl includes tools for communities and organisations to manage lifecycles of digital twin components.

concluding remarks

When the control and reasoning over the functioning of infrastructures move to a simulation environment, the distinction between the physical and virtual changes to a new configuration, one that prioritizes the digital as a constant object, while the physical is in a perpetual state of update and optimization.

By creating a space for the integration of digital twins of complex real world systems, their digital models can begin to co-evolve, reshaping how they are conceived. In this model, a digital twin is able to integrate and optimize for interactions, actors and environments that run beyond its scale and domain.

As advanced as technology is today, and as powerful a tool as computer modeling behind it has been, it's no replacement for the real world. We need to constantly reconcile our models with reality, before our models redefine us. For that, we need an open platform for simulations.

the project

Sybl is a speculative design project produced as part of Strelka Institute's 'The New Normal' programme.

The project explores the metastructure of infrastructure and the workflows for scientific computing to suggest how recent advances in open source collaboration and cloud computing can the shift constraints that traditionally restricted the role of design in industries with high technical complexity.

Inspired by the role that platforms like GitHub and Wikipedia are playing in software development, data science and media production, Sybl proposes the emergence of a new platform for collaborative simulation built around emerging data standards like Pixar's USD (Universal Scene Description) and next-generation neural network approaches like Capsule Networks.

Project Team

Ricardo Saavedra, Mariia Fedorova, Grigory Chernomordic, Mark Wilcox

Music by

Ollie Zhang, Julius Holtz, Ricardo Saavedra

Thanks to

Strelka Institute (Benjamin Bratton, Nicolay Boyadjiev, Olga Tenisheva, Liza Dorrer, Elena Mozgovaya, Vlad Ilkevich), Metahaven, Liam Young, Natalie Afordia - Uber ATG