Coffee Space

Inspiration

This line of enquiry started with a post over on HackerNews by thisisdallas:

Ask HN: Is Twitter Dead for Interesting Discussion?

I haven’t been on Twitter in several years because of the toxic nature. It seems like everything is now political or social commentary which usually goes, “I’m right you’re wrong”

Is that primarily the case in the startup world today? Is Twitter still a useful platform for reading about bootstrapping founders, new technologies, and interesting business related discussion?

Firstly, not being on Twitter is a great move. It is essentially an attention economy, everybody trying to get five minutes of fame, but many people ending up with two minutes of hate. Internet points have no real value and you will be a lot happier if you stop investing your time into this worthless economy. Instead, put down your phone and invest your time into people around you. Go see your parents, go for a walk with a friend - invest in tangible things.

Given the amount of political and social commentary happening, there is clearly a need for it - I just don't believe it can be done in a handful of characters in an unstructured platform. No real great truth will ever really be reached on Twitter - or at least no great truth will be reached that could not be more efficiently reached elsewhere.

Whilst this comment is more geared towards discussion, I believe ultimately discussion is just a form of debate where your opponent(s) are more generous. When I discuss debating, discussing or arguing - I group them together as an exercise of truth seeking. There is some question, there are people with some preconceived ideas about how to answer that questions and they discuss to convince one another of their truth, where hopefully the 'real' truth falls out at the end.

Let's start by addressing the issues with using social media as a debating platform:

• Random opponent: Anybody and everybody could eventually end up arguing against you, you may not even have one single opponent but multiple, all making slightly different points. This lack of argument collaboration inherently means the goalposts are forever shifting.
• Accountability: People are free to remain anonymous, and whilst this might help start some discussions, it means that the people discussing these topics are not held accountable to their own points. This isn't about cancel culture as such - it's about making sure that you can have a good faith discussion.
• Rules: There is of course no rules of debate when arguing with random people on social media - or at least none officially agreed to. This in turn gives space for the rules to change mid-debate, to the point where the discussion can longer be fruitful.
• Delay: Social media by design is made for quick replies, which are inherently less thought out. Items not replied to within some short period of time typically are lost interest in. By building in a minimum and maximum time delay between replies, it can be possible to have better thought out arguments. A person can for example consult external experts, reference their argument and generally build better arguments.
• Moderation: Of course, if you are going to have rules, you need somebody to actually enforce them. On social media even if you agree to some debate rules, there is no way to actually ensure the other person acts on them in good faith.
• Onlookers: Onlookers are often able to directly comment on or influence a debate with their immediate reactions. Often this happens before the debate has really gotten into the subject matter.
• Structure: Debates on most social media platforms are typically ill formatted (hard to read, etc), replies are often hard to find (different threading model, click to expand, etc) and fundamentally limited by the format (lack of characters, limits to links, etc).
• Freedom of speech limitations: Rightfully or wrongfully, there are limitations to speech on social media. On a different platform it could be easier to discuss more divisive issues if first the right framework is put in place to facilitate the discussion.

As I suggested, I really quite like LongBets - I believe that for making predictions about the future, it is almost unrivalled as a platform (from what I have seen). Let's start with the things I like:

• Investment: All bets require some minimum investment. This prevents spamming of proposals and also allows verification of the person by means of their banking account.
• Payment: The website requires that you help pay for the hosting and survivability of the website.
• Simplicity: The website design is super simple and accessible, and therefore has a sense of longevity about it.
• Rules: There are a series of rules to be agreed to as part of the procedure.

I will make some criticisms though:

• Charity: All bets must be made to a charity. This may be particularly difficult if a charity cannot be easily agreed - what if your argument for example is that "we should not give to charity"?
• Betting: I really don't like the encouragement of gambling in general, even if the gambling is supposed to be well thought out. It also inherently costs out some opponents from entering into the debate, in spite of potentially being better debaters/arguers.
• RSS/subscription: It used to have an RSS feed, but unfortunately that was stopped. I suspect that they had some issue with hosting costs or something.
• Sustainability: It appears that all money is invested into The Farsight Fund, so in theory it survives only by the good management of that company. If the economy was to take a downturn or the portfolio was managed badly, it is possible that LongBets would not survive, despite still having many users.
• Archiving: I believe that such a website should offer multiple forms of archiving, including physical mediums. Completed debates for example could be published in a book that could be purchased by readers and collectors. There would also be dome download-able digital mediums, such a .txt, .pdf, etc.

And so starts the blog post about this!

Proposal

Here I want to put down some ideas for the platform design itself. For me, such a system would have the following requirements:

• Plain HTML: This format appears to be one that will survive the longest and is most accessible. CSS can be used, but generally this should be very lightweight and only used to make the content more readable/presentable.
• Alternative formats: As this is long lasting content, we could offer arguments in multiple forms. This could include plain text, PDF and even a printed book.
• RSS: Allow people to receive updates about changes to debates (something that LongBets no longer offers).
• HTTP/HTTPS: Support all versions of content delivery, including secure and insecure communications - as well as modern transfer protocols.

One thing I was thinking about is how to handle login security over long periods of time. It occurred to me that actually it would be entirely possible to do this via email, entirely bypassing the need to handle user credentials. This in turn pushes the security problem onto the email servers themselves.

On top of this, if we require that debates are in English only (the most commonly spoken language), then it should be possible to handle all debates in 7 bit ASCII only. This will incredibly simplify parsing and storage of debates.

To this end, I also believe that all moderation could also be achieved via email, allowing moderators to either use their email client or some external client as they wish. This means that moderators themselves would not need to have a login either, and could moderate debates also via email.

Infrastructure

The plan for infrastructure would be as follows:

+-------+  +-------+
|Browser|  | Email |
|       |  |       |
|Client |  |Client |
+---+---+  +---+---+
    |          |
====|==========|====
    v          v
+-------+  +-------+
|Static |  | Email |
|online |  |online |
|Server |  |Server |
+-------+  +---+---+
    ^          |
    |     +----+
====|=====|=========
    |     v
    | +-------+
    | |Process|
    +-+offline|
      |Server |
      +-------+

Hosting the static content separately would allow for a highly optimized server to be used (with high traffic capabilities), with emails being processed and static content being generated on a dedicated 'offline' server (not outwardly accessible). This should greatly reduce associated costs with running the website.

The email server itself could simply be a custom domain for an existing email service, meaning that spam filtering, etc, could be handled more easily.

Generally the overall process of the implementation would be:

1. Fetch: Fetch the latest updates from the email server.
2. Process: Process the latest updates offline.
3. Publish: Publish the results of the processing online.

Protocol

Like LongBets, it is important to define a procedure. This isn't fully complete, but I imagine it may look something like this:

1. Debate proposal: To start off with, somebody would need to create an argument for a particular case.
2. Proposal validation: Proposals are given feedback to ensure they meet the rules required for being valid.
3. Payment: Once the proposal is validated, a person must then pay to have their debate listed.
4. Debate acceptance: Another party accepts the debate conditions within some time period.
5. Moderator allocation: One two parties are willing to debate on the given conditions, a moderator is allocated.
6. Proposer starts: The person who proposed the debate is then able to more comprehensively give their initial argument.
7. Response cycle: Both parties are allowed to enter a response cycle, until a given number of cycles are reached or one of the exit conditions is met. These responses are first checked by the moderator and then published.
8. Closing remarks: Each party is then able to give closing remarks to conclude their arguments in good faith and to thank their opposing debater for their time.
9. Jury allocation: To decide a victor, a random jury is allocated from a pool. They must prior declare any conflicts of interest regarding the subject matter. The moderator is to ensure there is a roughly even bias within the jury if possible. (Some subjects may be too controversial to achieve this.)
10. Jury vote: The jury is then able to vote on the winner.
11. Winner declared: The winner is then declared on the site if there is an overwhelming victory, otherwise the vote can be split if it is too close to call.
12. Debate review: Everybody involved is then able to comment on the debate itself. This is then added to the 'profile' of each person (moderator, debaters and jury members). This is to allow for better selection in the future.

I believe that each person involved should be offered a token in exchange for their work. Moderators should likely be offered more than the jury team. This is used as a partial incentive for each person to maintain a high standard or work.

Rules

This is just some rough set of initial rules for the website - it would definitely need to be thought out better:

1. Real names: One of the biggest issues with social media is that people can make arguments in bad faith from anonymous accounts. It is very difficult to have an argument in good faith when your opponent is arguing in bad faith, and worse still cannot be held accountable for doing so. What's more is that future people who interact with this person have no idea that this person is arguing in bad faith.
2. Win condition: This would be when the 'jury' have voted. Both parties would agree to accept their conclusion.
3. Allocated moderator: Some person to oversee and ensure the debate/discussion is held correctly.
4. Collaboration: Responses could be drafted collaboratively in private with a small group before being shared.
5. Timeout: Make an agreed timeout window for responses to be made.
6. Payment: There should be some payment made to the website to ensure its survival and to also pay the moderators for their role.
7. Retraction: Debaters enter an agreement that they are signing over the rights to the content they are writing, although it will be attributed to them.
8. Moderation log: All comments/changes made by the moderator are logged publicly for the purpose of transparency.
9. Maximum counts: There should be some agreed number of maximums, for example the length of the response, the number of references for each response, the number of responses that can be made in total, etc.

There are of course many edge cases that this doesn't even begin to consider, for example:

• Unhandled case: There could be some unhandled case in the logic, which in such case the moderator could perhaps be allowed to make a good faith effort to handle the situation. This is how judges and referees are often allowed to operate.
• Debater selection: It might be that the proposer wants to debate a specific person, or wishes to select the person with whom they debate with - rather than allow some random person to debate with them. This should likely be declared in the proposal process.
• Member withdraws: Perhaps a debater or moderator fails to reply, or requests to withdraw. This should be handled more elegantly. It could be possible that somebody else chooses to enter into the debate in their place if all persons agree.
• Rules broken: There should be some way of handling broken rules and the severity to which they are broken. Perhaps there is some negative scoring towards the person debating and some consequences as a result.

Next Steps

In the future, perhaps after the PhD, I will consider throwing a mock together. The initial version could be done entirely manually, with a more automated version being implemented once the idea generates interest.

The only real investment to get this project off the ground would be a decent domain name and some time. I imagine it would be possible to perform some initial implementation tests by allowing friends to have a debate and generally ironing out the kinks of the system before going fully public with the idea.