tldr: Punditry and public predictions tend to be unaccountable, unfalsifiable, unequivocal, and unrevisable. Augur makes it possible to make better predictions and to create media claims backed by these predictions. I demo this by placing an Augur trade to back a public prediction.
Lately, I’ve been thinking about how Augur can be used to improve the way we make public predictions and to replace punditry as we know it today. Here are some early thoughts on how blog posts, tweets and other media can link up with Augur to help us make more honest, useful, and accountable predictions. I’ll also put my money where my mouth is and stake an ether on a public prediction…
The Four Traits of a Good Prediction
What traits should a prediction i.e., claim about the state of (future) reality possess? To work backwards, I think a bad prediction would be the following:
- Unaccountable — no skin in the game.
- Unequivocal —i.e., dogmatic. No indication of (un)certainty level or perceived probability.
- Unfalsifiable — cannot be proved wrong
- Unrevisable — set in stone, cannot be updated based on new events or knowledge
Augur is basically the opposite of all four of these:
A prediction on Augur is accountable. If I make a prediction on Augur, I am taking on risk. If every time someone made a public prediction they needed or were expected to take on risk, that would weed out many low-conviction or dishonest claims and result in a far higher signal to noise.
A prediction on Augur is probabilistic rather than unequivocal. The price paid for a share is an expression of perceived probability.
A prediction on Augur is falsifiable, since the oracle serves as a merciless truth engine that will deem your prediction TRUE or FALSE. While it could in theory be unfalsifiable, that would mean it’s an “invalid” market which market creators are incentivized against creating.
A prediction on Augur is open to revision rather than unrevisable. Unlike a betting market, a prediction market is open to revision since you can buy and sell, rebuy and resell etc a prediction at any point — provided sufficient liquidity. I may predict that Dwayne Johnson will be elected president in 2024, but if Alexandria Ocasio-Cortez jumps in the race later on, I may sell my shares to reflect my updated views.
I’m calling a prediction that is accountable, probabilistic, falsifiable, and revisable a radical prediction because it is both a big departure from the status quo and…much more powerful.
Imagine that whenever anyone made a public prediction whether on Twitter or a blog post, they linked to an Augur market where they placed a stake to back it up. We might call this a “hyprelink”, pre as in “prediction.” I’m sure there’s a better term for this, but this will have to do for now…
If I share a hyprelink, I am:
1) Putting verifiable skin in the game. As Nassim Taleb says, “Don’t tell me what you think. Tell me what is in your portfolio.”
2) Showing my level of (un)certainty. “I think X will happen” could mean 51% or 91%. That’s a big difference! By trading shares, I set a price that signals my perceived probability of the outcome.
3) Forced to state the prediction in a way that is falsifiable since Augur users are incentivized against creating markets that have unverifiable outcomes.
4) Providing a way for others to track my revised views over time. Ideally a hyprelink should display one’s complete history of trades in the relevant market to reflect one’s change in views. Seeing what someone thinks is less interesting than seeing how what they think changes over time.
Just as Google crawls hyperlinks to rank websites, future search engines might crawl hyprelinks to rank experts and forecasting websites.
“Proof of Augur Stake”
In the spirit of experimentation and putting my money where my mouth is, I’ll do a primitive demo of a “radical prediction.” My hope is over time we’ll develop better tooling for verifiable, Augur-backed public predictions and hyprelinks, but this will have to do for now.
So…I just placed a sell order on Augur:
Limit Price: .48 (short)
Quantity of Shares: 1.93
This prediction is: Accountable (putting my money where my mouth is), probabilistic (setting a price to it), falsifiable (will be deemed false or true by Augur’s oracle), and revisable (by tracking any future trades said address makes, you can see my updated views).
I am stating the view that there is less than a 48% chance that a Democrat will win the 2020 U.S. presidential election. I’m risking just over one ether plus tx fees. It’s a relatively contrarian bet considering that the centralized prediction market PredictIt, which is much more liquid on this outcome, is giving Democrats a ~63% chance as I write this. I could almost definitely get better odds on this trade, but I want this prediction to reflect my perceived likelihood of the event to demo a radical prediction in a pure way. Finally, note that this is only a prediction, not a preference for the outcome.
Update: An astute reader pointed out to me that the market in question has poor wording (mentions “for a second term”) and thus will likely resolve as invalid. So I’ve removed my order. Still, this works as a demo of how an Augur-backed prediction would work, and the aforementioned transaction remains verifiable.
In an upcoming post, I’ll share some thoughts on how to avoid plutocracy in Augur-backed predictions and some of the finer points of how this might work.
I think radical predictions can encourage more honest, productive and humble discourse. Punditry, as we know it today, may be looked back on as a temporary phase in human history between the time when we had accountability from tight-knit, tribal society and the time we had accountability from prediction markets.