tldr: LETS is looking for a web developer to help create a prototype of Profound, a system for collective dialogue at scale. If interested, send a message to Humphrey Obuobi ([email protected]).

Profound is a tool that enables constructive, large-scale conversations about complex public issues. In short, the system will:

• Elicit responses from individual participants (written or vocalized) to an open-ended question
• Extract short claims, beliefs, and positions from their response
• Allow participants to respond to each other’s claims
• Summarize where people agree and disagree, as well as themes of the conversation.

Profound belongs to existing class of digital tools called “collective dialogue systems” (CDS) (source 1, source 2) that are built to understand the interests, values, and positions of a given populace. Going beyond traditional polling (which suffer from bias in question design, shallow insight, etc), these tools are “deliberative” in that they 1) support people weighing in with their full thoughts, 2) allow people to directly engage with others’ responses, and in some cases 3) summarize themes from the full conversation. Well-used examples include Pol.is, Remesh, and Ethelo; each of these tools have been deployed by governments and civil society organizations of varying sizes to navigate controversial public issues like land use issues, rideshare regulations, and the behaviors of AI models.

This tool uses the same basic interaction that these tools rely on (agree/disagree to statements), but builds on them in the following ways:

• Voice-based elicitation. Most of the common tools rely on text-based elicitation of short, specific statements; this makes engaging with others’ statements through the CDS easy. At the same time, spoken natural language is the most natural and efficient way to elicit people’s thoughts on a given matter, thus reducing the barrier for people to contribute their thoughts to a public conversation. Profound’s approach strives to achieve the best of both worlds by identifying meaningful claims from spoken contributions – a task that large-language models are uniquely capable of.
• Elicitation-first. The design focuses on initial elicitation first and foremost, thus prioritizing public voice. The system will be designed to infer as much as possible about people’s alignments based on the semantic distance between their claims, reducing the amount of explicit voting/evaluation required. Directly exchanging perspectives still is an important part of any deliberative tool; this is meant to reduce the burden of parsing through others’ opinions, rather than completely replacing that step.
• Public-oriented, accessible design. Many tools in this space (e.g., Pol.is, Remesh, etc) are built for researchers, by researchers. On the contrary, Profound is intended to feel significantly more accessible than the average survey, CDS, or otherwise (mostly through simple, focused, and playful interface – think something like Duolingo). In practice, this tool will be deployed by institutions like government departments, public radio stations, nonprofit organizations, and others who are dedicated to convening the public around shared experiences and challenges.

By building with these principles in mind, Profound aims to become a social utility for civil society. At a time when trust in democratic institutions, building systems to help foster mutual understanding and coordinate the public will are more important than ever before. Enabling more constructive conversations (by more organizations, works towards a vision of democracy where everyone not only feels that they have a voice, but where people and public institutions are able to easily make sense of their shared challenges.

We’re looking for a software engineer with experience building mobile-first web applications who can help build a prototype of Profound.

Priorities/responsibilities include:

• Developing a claim extraction system to elicit spoken words, identify short statements, and store them
• Developing a voting system that enables unique participants to weigh in on the approved statements