Arbital's next pillar: Discussion

Arbital has a prototype discussion platform. If you're excited about this and want to participate, you can start by voting on claims and proposing comments.

[toc:]

State of the conversation

Anna Salamon recently sparked a conversation about the way Less Wrong's decline has left the community without a locus for discourse, leaving important conversations fragmented and hindering our collective truth-seeking and future-steering abilities. Since then the community has been abuzz with discussion about how to improve the state of online discussion. %%note: Notable posts include Sarah Constantin's A Return to Discussion, Ben Hoffman's Canons (What are they good for?) and Improve comments by tagging claims, Eliezer Yudkowsky's post on Building an intellectual edifice, and several threads about moderation.%%

We think Arbital is well-placed to develop into an awesome platform for the kind of high-value conversations being discussed. %%note: We have three excellent full time programmers, strong connections with a community of dedicated truth-seekers, and a working site with many useful features.%% Arbital is a grand experiment aiming to improve the flow of information, especially important information. We have already built a functioning set of wiki and explanation features which will support discussion by enabling close integration with a structured body of collaboratively improved knowledge. %%note: There are still lots of known avenues for improvement in this area, but the wider vision requires many pillars and, given limited dev time, it seems better to have a set of workable parts covering all the core functions than to have one highly polished part unsupported by the rest of the system.%%

Discussion: requisites, routes, risks

We've been meeting with many people and putting a lot of brain cycles into trying to figure out what's needed for valuable discussion to thrive and scale, while avoiding the many pitfalls. %note: A thriving discussion contains truth-tracking exchange of ideas (e.g. Double Cruxing, making explicit claims), produces artifacts which new arrivals can use to catch up, rewards insightful content with higher visibility, and is resistant to Eternal September.% After also looking at existing platforms, we think we have a good shot at building something dramatically better for our purposes than anything currently publicly available. If you're interested in a sample of our thought here are a few of the things which seem important to us. %%hidden(Parts of this pillar):

Good epistemic norms

Fostering healthy norms of discourse is essential. Fortunately, we have a seed community with an unusually high standard of truth-seeking debate, a genuine interest in trying to understand the world, and an awareness of importance of good norms. The initial version of this will be developed with the early adopters, but it's likely to include things like using probabilities, factoring out [arbital_claim claims] %%note: E.g. Explicitly tagging the core claims of a post will make people substantially more likely to respond to these claims., and inline claims on Improve comments by tagging claims%%, stating cruxes, and encouraging respectful disagreement.

Giving good content more attention (and visa versa)

Time and attention are scarce resources, especially for people most likely to be advancing the frontier of knowledge. We expect a system which reliably promotes excellent content to the view of many people will incentivise the creation of more excellent content.

There is a tricky tradeoff between being open to authors and being reliably interesting for readers, failing to adequately navigate this has caused many communities to fall to Eternal September. We recognize this challenge and have several ideas about how to get the best of both worlds %%note: Including ideas around manual moderation / review / promotion with a well-organized team, Stackexchange-inspired trust metrics, reaction-based promotion (weighted by trust), and more ambitious plans%%, but until we implement a system which we think has a good chance of working we will be growing slowly and carefully via invites to preserve quality.

Creating legible settled knowledge

In order to make it practical to join complex ongoing conversations, the settled knowledge and shared terminology must be accessible and structured.

We aim to tackle this in two ways:

• Making the original discussions filterable and navigable in ways which makes for useful reading (e.g. differential post visibility to view only the core of a conversation, tagging, making it easy to browse what's been said on any topic).
• Using our wiki features to allow easy creation of more polished explanations. We have features which make it easier to navigate and learn from than a normal wiki, including ones to cater for people with different backgrounds, make content have clearer dependencies and more modularity, so people can be linked to exactly what they need to learn a topic (and explore from there, if they want).

Engaging with ideas from cradle to grave (or canonization)

In order to enable collaborative thought throughout the whole idea lifecycle there has to be space for bouncing and fleshing out early stage ideas as well as nitpicking and pinning down details of carefully thought out theories. We like [epistemic status:] tags as a response much more than fragmenting between different platforms with different expectations, and will be thinking about how to support different sets of discussion norms in parallel on Arbital. %%

The (current) plan

Our long term goal is to have an open platform with all sorts of communities and topics, but for now we will be focusing on Arbital Labs. We'd love to be able to let everyone join in comments and submitting content, but we want to steer the development based how people interact with features and explore norms of discourse in a manageable environment, rather than trying to maximize early growth.

In the meantime, you're welcome to create pages and claims on Arbital anyway (think of it like a wiki/blog hybrid with explanation and claims features), but there's no guarantee they'll end up listed in the feed. We're also happy to host pages for cool projects, talk to us on Slack if you're not sure.

Is this a replacement for Less Wrong?

Replacing Less Wrong is not an explicit goal. We are focusing on building a platform which enables people to share information and collaboratively develop an accurate understanding of the world, not trying to replicate everything LWers are used to.

However, our aims have significant overlap with those of Less Wrong, our ideal feature-set includes many of the things that made LW work well, we expect many of the early adopters to come from the LW community, and we will be in close contact with LW stakeholders. It is likely that, if the LW community decides they would like to move to a future version of Arbital en-mass, we would support them by importing the archives.

Want to be an early adopter?

If you’d like to participate in this experiment try voting on a claim on the home page or suggesting comments.