Talk is cheap

Public discourse is polluted with overconfident claims that carry zero repercussions if wrong. The list of topics is endless: this country lied, that candidate cheated, this food treats a sickness, that convict is innocent.

While spurious claims are often harmless, the phenomenon becomes more concerning when wrong information causes people, organizations, and governments to support or adopt harmful policies.

The challenge

We’re following the model set by James Randi’s One Million Dollar Paranormal Challenge. James Randi (1928-2020) did the world a great service when he offered to pay anyone who could demonstrate paranormal powers in a lab setting. In doing so, he dramatically simplified public discourse on the subject. Previously, people were forced to debate each paranormal claim separately, and once one was shown fraudulent, another one would quickly replace it. Thanks to his challenge, people could quickly determine that all such claims are false, since for each paranormal claim, it is near certain there is at least one claimant who would like a million dollars, and thus the lack of applicants indicates they know the claim is false.

In the wake of James Randi’s passing in 2020, Rootclaim is launching its own challenge, betting $100,000 on the accuracy of our analyses. When interested onlookers see that one side is willing to take a risk while the other is not, they have a very strong tool to determine which side is more confident that the evidence backs their claims, and is, therefore, more likely to be correct. Furthermore, a public betting challenge implies much stronger confidence in a claim than a standard bet between two people. In a personal bet, each side only claims to understand the issue better than the other. In contrast, in a public offer, the challenger is claiming that there is not one person in the world who has a better understanding of the issue and holds the opposing opinion.

When you see a reliable public betting challenge with real stakes, you can be very confident the claim is true at a probability that is significantly above 50% (assuming 1:1 odds are being offered).

Active challenges

The following Rootclaim conclusions are currently available for the challenge. Want to debate some of our other findings? Let us know at [email protected], and we’ll consider it.  

* This analysis was released prior to new variants and Covid vaccines and is therefore not up to date. If you wish to challenge the original conclusion, please contact us to set the exact criteria.

Format

The following are the default parameters and conditions of the Rootclaim challenge, but we are flexible, so feel free to contact us with offers and modifications.

Qualifying for the challenge

The challenge, which takes the form of a debate between a Rootclaim representative and the challenger, will be based on all currently available evidence. The goal here is not to trip up or trap the opponent, but to determine which hypothesis is better supported by the evidence. If you have new evidence or evidence we overlooked (including those dealing with the priors or “Starting Point” section of the analysis), it should first be shared so that we can update the analysis. If it doesn’t significantly change the conclusion, the challenge can be accepted.

We are not claiming to have better evidence, but rather aim to demonstrate the superiority of probabilistic reasoning over human reasoning when evaluating the same evidence. 

Nonetheless and in order to encourage and show our appreciation for those who provide new evidence that changes our analysis enough to withdraw the challenge, Rootclaim will reward such contributors with a $1,000 reward (Note: if the new evidence is publicly available on the web or in a book, article, etc., Rootclaim would only pay the reward if 30 days have passed since the initial publication and if we have not already addressed it or wrote that we are aware of it and will be addressing it in the future). 

Financial and legal aspects

Both sides will deposit the amount wagered ($100,000 by each side, unless another sum is agreed upon) in an escrow account, as a precondition to starting the challenge.

The expenses for the debate (e.g., transaction fees, lawyers, Judges fees, etc.) will be taken out of the total cash pool, reducing the prize accordingly. 

All disputes shall be subject to the exclusive jurisdiction of the courts of Tel-Aviv, Israel.

Judges selection

Both sides will first agree on two judges with strong analytical skills, no previous endorsement of either side, no relevant biases, who declare they will examine both hypotheses equally, and with experience pertinent to the challenge (For example, a judge in a challenge to our analysis of the chemical attack in Ghouta, Syria in 2013 should be able to understand organic chemistry, ballistics, geolocation, video forensics, etc).

Alternatively, Rootclaim is willing to accept other fair judges selection procedures, for example:

• Both sides Agree on a consulting company and the general characteristics of the judges (e.g., experience in a certain field). 

• The consulting company selects the judges based on the characteristics and without interference from either side. 

• Each side can then eliminate one of the selected judges and the remainder of the judges are used for the challenge. 

Debate structure

• Textual debate

• The textual part of the debate will take place in a collaborative online document editor (e.g., a Google doc). 

• Using a hierarchical bullet-point structure and different text colors, each side will respond to the other in the relevant location in the document. New arguments will be added at the end of the relevant section or the end of the document if no such section exists. 

• Once the agreement between the sides is signed, Rootclaim will share the initial document with the challenger, which will initially be an updated version of the analysis he is challenging. Rootclaim will have 10 days from the signing of the agreement to provide this shared updated analysis. 

• Once Rootclaim shares the updated analysis, the judges will have 72 hours to leave comments inside the document, asking for clarifications, more information or pointing to any mistakes or problems. 

• The challenger will then have 10 days to respond in the document and add any new claims.This initial response by the challenger will be limited to the higher of 10,000 words, or the number of words in Rootclaim’s updated analysis. 

• Once the challenger’s response time is over, the judges will again have 72 hours to leave their comments on the new additions. 

• For the next 30 days, each side will have 48 hours to update the document with their responses and additions, while the other side is not allowed to make any changes. Once the 48 hours are done, the judges will have 24 hours to comment on the new additions. Once these 24 hours are done, the roles of the sides will reverse, and so on until the end of the 30 days. 

• After every two iterations (148 hours), 24 hours will be reserved as a day off for all sides, with neither side allowed to make any changes to the shared document. 

• Additional pauses can be requested by either side and approved by mutual agreement or a majority decision by the judges.

• To avoid flooding the document as a means to win the debate, each iteration will have a 30% reduction in the maximum number of words each side is allowed to add (e.g., 10,000 words, then 7,000, 4,900, etc.). Exceeding the limitation is allowed only by mutual agreement of the sides or by a majority decision by the judges if they find the addition is indeed needed and legitimate. Judges may not base their decisions, directly or indirectly, on information appearing in rejected additions. Any addition allowed for one side will be granted automatically to the other side and can be used at any of their remaining writing-privileged sessions. If an addition is granted at the final session of the textual debate, the other side will be granted another 48-hours session to utilize his equal addition. 

• Links, references, and images

• Links and references are unlimited and are not counted as words. It is the other side’s responsibility to check the sources support the claims based on them. The judges will refrain from checking sources directly and if needed, will request clarifications regarding these sources via comments in the shared document. Each side’s responses will be deducted from its word allowance. 

• Links and references must also include the specific locations (page, section, etc.) relevant to the claim. If the relevant information continues beyond the starting location (e.g., multiple pages), the reference will include both starting and ending locations. 

• An embedded image (including graphs) is considered 50 words plus any word in the image.

• Video and audio can be quoted, described, or screenshot into the document (counted as 50 words per screenshot) but not embedded. Links to the source media must be added in the references and just like other references, it is the other side’s responsibility to check the content and judges shall refrain from viewing or listening to the material directly.

• Video or audio might be embedded by mutual agreement of the sides or by a unanimous decision of the judges, if the content of the media is deemed important and too complicated or lengthy to be accurately described in text. 

• The last side to hold writing privileges over the document, is limited in this last iteration to responses and may not add new claims or information not required for his response. If the other side believes new claims or information was added, the judges will rule on the matter. Judges may not base their decisions, directly or indirectly, on information appearing in rejected additions and the total number of words in the stricken material will be forfeited from the word limitation allowance of that side.  

• Once the textual debate’s deadline is reached, changes and additions can only be made by mutual agreement.

• Throughout the textual debate, the judges would have access to the debate document, with commenting privileges only. Judges can ask questions in the comments, to which both sides can respond. 

• Live debate

• Information presented in the live debate is limited to information presented in the textual debate.

• If either side believes new information was or is being presented without permission, he may ask the judges to examine his claim by comparing the information to the textual debate document. If the judges agree, the information will be disregarded and will not be considered by the judges in their rulings, either directly or indirectly. 

• Information not included in the textual debate document may only be introduced if judges ask for it or by mutual agreement by both sides.

• Unless otherwise agreed upon, the debate will be held in online video format.

• The first part of the live debate will be a presentation of each side’s case in full, intended to give the judges and the viewers an overview summary of the hypothesis. 

• The winner of a coin flip toss decides who will present their case first. Each side’s presentation is limited to 45 minutes. 

• Prior to the live debate, the judges may comment in the document on issues they would like a side to elaborate on in the live debate. The second part of the live debate will be devoted to the sides explaining these issues. Once one side responds to a comment by the judges the other side is allowed equal time for rebuttal, if one is needed.

• Following this section, judges can freely ask questions of both sides. To allow equal talking time to both sides, once one side responds to a question by the judges the other side is allowed equal time for rebuttal, if one is needed. This step repeats itself until the judges are satisfied and have no more questions.

• Lastly, each side is allowed to give up to 5 minutes of closing arguments. The order of speaking will be reversed from the order decided by the winner of the coin toss. 

Winning

• Each judge will decide which of the two hypotheses is more likely

• Each judge will write down his decision, without consulting or sharing his decision with the other judge or any other party. 

• If both Judges agree, the prize pool, minus the debate expenses, is paid to the winner. Otherwise, it is split equally.

Apply Now: [email protected]

Q&A

If this were real you would've lost the money by now. What's the catch?

There is no catch. Though we disagreed with the judges' decision, we already had our first debate, lost it, and paid 100,000 dollars to the other side. So you can rest assured that this challenge is real. 

Whose money is it?

The funds from our side are pledged by our founder, Saar Wilf, and both sides will put the money in escrow before the debate begins, as was done in our first debate. 

I know I’m right but I can’t risk $100,000.

People routinely make overconfident claims that fit their political and personal biases without any repercussions. When a cost is introduced (“skin in the game”) people are forced to honestly consider their positions, their biases, the reliability of the sources they use, etc., and that is our true goal.

A challenger who is truly confident in their conclusion should expect a far better than 50% chance of winning a debate in front of competent judges. Applicants who can’t afford to risk $100,000 are encouraged to pool funds together or even crowdfund it. We are also willing to reduce the stakes as low as $10,000 for applicants already involved in public debate on the issue. Winning such a debate would achieve our goals of promoting rationality regardless of the money won.

If you are so sure in your conclusion, why not allow a risk-free challenge?

We would love to be able to copy James Randi's challenge and just offer the prize to anyone who could show a Rootclaim conclusion is wrong. But unlike paranormal claims that can be tested in a lab, there is no way to prove who is right (if there was, we would just use this method instead of probabilistic inference). The best we could come up with is a debate in front of unbiased professional judges, which leaves us open to human error, and a risk of loss even if we’re right. In fact, this is exactly what happened in the first Rootclaim challenge, where mistakes we made in structuring the debate led the judges to vote for the less likely hypothesis. 

Rootclaim conclusions are expressed as probabilities. If a hypothesis is calculated as being 90% likely, there's still a 10% chance it's wrong, which translates to a risk of losing the debate.

If only one side takes a risk, the other side is incentivized to participate without thoroughly studying the issue, just for this low chance of winning the prize due to a mistake.