About

I'm a human-computer interaction researcher at heart, with recent experience in architecting and deploying world-spanning distributed systems. I get a kick out of wrangling a bunch of cross-platform moving parts, and thinking through juicy edge conditions and how to handle them.

Subjunctive interfaces, and VPRI

Much of my career has happened in the orbit of Alan Kay. I joined his Viewpoints Research Institute (VPRI) as a Senior Researcher in 2011, and carried on fleshing out the subjunctive interface theme that I had come up with during my time as a researcher in Japan. A subjunctive interface helps a user to understand the context surrounding any computed result – from a search, a simulation, a design… – by drawing attention to the other results that would have been found by slightly varying the parameters. The traffic time travel demo is an example.

Croquet

After VPRI closed in 2017, a few of us joined other Alan Kay associates to co-found Croquet Corporation, whose open-source JavaScript library greatly eases the development of high-performance collaborative web applications. As well as contributing to the core system, I designed and developed many internal and customer-driven applications, especially those that called for in-depth integration with browser APIs or third-party services, including real-time streaming and game-engine rendering. I also became known for being the person to call in to track down, diagnose and resolve the obscurest multi-user application errors. See some Croquet demos here.

At the follow-on company Croquet Labs, technology provider to multisynq.io, I took on the scaling-up of our new Cloudflare-backed DePIN deployment, from proof of concept to a global platform offering session dispatch and automated quality-of-service monitoring across thousands of heterogeneous community-provided nodes.

Now

I'm exploring how to pull together the system-building and research aspects of my career, particularly around the user experience of AI-assisted work. At the simplest level, how might we nudge users away from passively accepting their AI collaborators' actions and advice, and instead coax and empower them to apply meaningful oversight and verification? I'm very open to conversations about this and related challenges, so do get in touch via LinkedIn, GitHub or email: aran at acm dot org.

A short CV is available as a PDF (September 2026).