What this covers

Stephen Grugett, cofounder of Manifold Markets, walks through how user-created prediction markets built on play money rather than real stakes can aggregate information and reshape decision-making in companies and institutions. The conversation explores why these markets work despite their simplicity, what structural barriers prevent their adoption in practice, and how the platform's design choices — particularly allowing users rather than centralized oracles to resolve outcomes — unlocked the scalability that earlier prediction-market attempts missed. Grugett situates prediction markets within debates about human motivation, institutional resistance to unwelcome feedback, and the mechanics of price discovery.

The discussion spans several distinct territories. On mechanism, Grugett argues that play money succeeds because people are driven by status and competition more than financial gain, and that user-resolved markets, though subject to fraud, are preferable to the elegant-but-impractical oracle designs others pursued. On adoption, he contends that companies abandon internal prediction markets not because the mechanism fails but because managers rationally resist markets that surface doubts about chosen missions — a structural problem no amount of technical refinement can solve. On design, the conversation addresses why crypto integration failed (wallet friction, onboarding costs, transaction delays), how to handle long-horizon questions (breaking them into shorter proxy markets), and the principle that simplicity matters more than handling every edge case. Grugett also touches on insider trading through a shareholder-harm lens, the comparison of virtual economies like World of Warcraft to prediction-market participation, and how far-future discounting is a feature of human nature no market can escape. The platform itself serves as a case study: Manifold runs internal markets on its own decisions and has seen users repurpose the mechanism for lotteries, games, and crowdsourced research — evidence that prediction markets are a general tool beyond forecasting.

Sharpest takeaway

Grugett argues that user-created, play-money prediction markets can scale information aggregation by prioritizing simplicity and status-driven motivation over financial reward, and that the main barrier to wider adoption is usability and managers' rational reluctance to surface unwanted truths rather than any flaw in the market mechanism itself.

  • Play money works because people are driven by status and competitiveness more than greed
  • User-resolved markets (despite fraud risk) are the unlock that makes prediction markets scalable
  • Companies abandon internal prediction markets because they step on management's mission and surface unwanted feedback

The claims · ranked29 claims · weighted by value

This asset isn't compiled yet

You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.

0.81

You cannot effectively bet on an apocalypse because you won't survive to collect winnings and money is worth less in that universe, so the rational move is to bet against catastrophe — illustrated by the Caplan-Yudkowsky AI bet structured as Caplan paying now and Yudkowsky repaying with interest if no catastrophe occurs by 2030.

causalhigh valueestablishednovelty 3/4durability 3/4· Stephen Grugett

“you can't like bet on an apocalypse... you're not going to be there to to collect your winnings uh the money is worth less in that universe so you should rationally bet against it”

0.81

Demurrage (negative interest on uninvested cash balances) is appealing in principle — eroding hoarded winnings to keep capital active and reflect that past good predictions shouldn't be rewarded forever — but is a non-starter in practice because users hate losing money for any reason.

causalhigh valueestablishednovelty 3/4durability 3/4· Stephen Grugett

“one idea i had is introducing demurrage or like basically negative interest rates on cash balances... in practice like users absolutely hate losing money for any reason at all so it's kind of like a non-starter”

0.80

For a usable platform, keeping the core mechanism simple and easy to understand matters far more than handling every edge case perfectly — the 'worse is better' principle — even if user-resolved markets are somewhat arbitrary, this beats an elegant but impractical oracle structure.

normativehigh valueestablishednovelty 2/4durability 4/4· Stephen Grugett

“the most important thing for creating a like a usable platform is making sure that the key mechanism is something simple and easy to understand more so than handling every possible edge case perfectly”

0.80

Discounting of the far future is a fundamental facet of human nature (reflected in interest rates), so prediction markets — like every other forecasting technique — care less about distant events; the mechanism can be tweaked at the margins but cannot fundamentally change this.

factualhigh valueestablishednovelty 2/4durability 4/4· Stephen Grugett

“the the interest rate is uh like a fact about human nature um or like discounting of the future is a like a fundamental um facet of human nature”

0.77

Companies (Google, GM, the CIA and others) have tried internal prediction markets but mostly abandoned them because managers rationally don't want negative feedback on a chosen mission — introducing a market can itself lower the perceived odds of success and undermine the mission's momentum, a Hansonian point that people don't actually want to know the answers.

causalhigh valuecontestednovelty 3/4durability 3/4· Stephen Grugett

“people literally don't want to know the answers to a lot of questions so in a corporate context that mostly manifests itself in the manager not wanting to get negative feedback about that”

0.75

The first hurdle to improving as a predictor is learning to clarify and crystallize views into numerical predictions and recording them, because without active recorded forecasts people misremember whether they were right and cannot improve.

causalhigh valueestablishednovelty 2/4durability 3/4· Stephen Grugett

“if you're not like actively putting out like numerical predictions and recording those to keep yourself to keep yourself honest it's very difficult to improve”

0.75

Markets are the most efficient mechanism for solving the 'whale problem': when a rich but uninformed trader bets heavily on the wrong side, they simply subsidize the correct traders and quickly lose their capital, analogous to liquidity providers profiting from a pension fund's forced rebalancing trades.

causalhigh valueestablishednovelty 2/4durability 3/4· Stephen Grugett

“if you think about what's happening when a whale comes in and buys up a huge amount of money and bets it on the wrong side what they're essentially doing is just like providing a huge subsidy and bonus to the people who are actually correct”

0.75

People will invest enormous time and effort into virtual economies regardless of financial gain — as shown by World of Warcraft and Second Life virtual real estate where people invested the equivalent of hundreds of thousands or millions of dollars — making non-monetary motivation a powerful force that can sustain prediction-market participation.

factualhigh valueestablishednovelty 2/4durability 3/4· Stephen Grugett

“people are willing to put a lot of time and effort into virtual economies irrespective of financial gains... like world of warcraft or earlier like second life which has like virtual real estate people are willing to invest the equivalent of hundreds of thousands or millions of dollars”

0.73

Prediction markets can isolate a specific risk better than financial instruments because asset prices are affected by too many factors at once, so picking multiple well-targeted proxy-variable markets yields a clearer understanding than trading a noisy instrument like oil prices.

causalhigh valuecontestednovelty 3/4durability 3/4· Stephen Grugett

“the nice feature of prediction markets is that you really can isolate the particular risk that you care about”

0.71

Play-money prediction markets can succeed because people are driven more by status and competitiveness than by greed, treating the platform as a social game where users hone predicting skills and prove their knowledge with an objective track record.

causalhigh valuecontestednovelty 2/4durability 3/4· Stephen Grugett

“the reason why we think that uh play money can work is that people are driven more by uh status and competitiveness um than greed you know we we kind of see ourselves as a social game”

0.68

Insider trading is best framed (following Matt Levine) as a taking from shareholders rather than a fairness issue; congressional insider trading is illegitimate because representatives should be compensated transparently via salary, since trading on insider information is a roundabout form of corruption.

normativehigh valuecontestednovelty 2/4durability 3/4· Stephen Grugett

“i basically have matt levine's view is that it should be viewed as a like taking from shareholders rather than a like fairness or equity uh type of discussion”

0.68

Prediction markets add clearest corporate value not in mission-evaluating questions but in outward-facing informational ones — market research, consumer behavior, competitor behavior — because those don't step on management's mission or trigger the same downsides.

normativehigh valuecontestednovelty 2/4durability 3/4· Stephen Grugett

“the areas where prediction markets more clearly add value without distracting from the mission or hurting management are where they're used for further informational aspects or um to do like market research”

0.63

Manifold pivoted from a crypto/web3 design back to web2 because crypto imposes severe usability frictions: most people lack crypto wallets, onboarding and transferring tokens is cumbersome, transactions are slow to confirm, each transaction must be signed, and transaction costs reduce liquidity.

causalhigh valuecontestednovelty 2/4durability 2/4· Stephen Grugett

“we're one of the first firms to pivot from web3 back to web 2 for like usability usability um reasons”

0.63

There is significant overlap between forecasting-community hobbyists and financial professionals, and prediction-market platforms can identify untapped top predictors, contradicting Tyler Cowen's claim that the absence of hedge funds hiring forecasters proves prediction is just a hobby.

factualhigh valuecontestednovelty 2/4durability 2/4· Stephen Grugett

“there actually is like as you would expect um a fair amount of overlap between the people who who are using these services and financial professionals who actually do manage money or trade”

0.61

Using prediction markets to fund investigative journalism (e.g., shorting a government's GDP before releasing damaging information) is a clear pro-social use of the market mechanism that provides a public good society otherwise wouldn't produce.

normativehigh valuefringenovelty 3/4durability 2/4· Stephen Grugett

“that's a very obvious case of uh pro-social benefit of markets using the the market mechanism to or in some ways is kind of like providing a public good”

0.57

Long-horizon and speculative questions are best handled by breaking them into shorter-term proxy variables (e.g., GPT-4 benchmark scores instead of 'AI apocalypse by 2050', or favorability ratings instead of 'will the Rock be president'), because these short-term markets resolve sooner and let speculators correct prices without tying up capital for years.

normativehigh valuespeaker onlynovelty 3/4durability 3/4· Stephen Grugett

“the best way to tackle these long-term more speculative questions is to try to break them down and address the proxy variables that are most relevant”

0.57

Real-world decisions often resist prediction markets because the success criterion cannot be operationalized into a single measurable metric; markets work best when many markets on proxy metrics inform human intuition rather than dictating decisions directly.

causalhigh valuespeaker onlynovelty 3/4durability 3/4· Stephen Grugett

“it's often very difficult to operationalize what the success criterion is down into one thing that the market can then measure and be used to act upon definitively”

0.53

Manifold dogfoods its own prediction markets to inform company decisions, including a conditional market on whether they should monetize by selling play money — phrased as whether they'd keep the feature if introduced — which the market favored and which substantively influenced their decision to proceed.

factualhigh valuespeaker onlynovelty 3/4durability 2/4· Stephen Grugett

“the first big case where the market actually substantively informed our decision about how to act as a company was an early market we created on whether we should try to monetize by selling the play money”

0.53

Users have hacked the prediction-market mechanism to do non-prediction tasks — running lotteries, playing games (Manifold plays Wordle), crowdsourcing research via free-response markets where bettors propose and back answers — revealing the market mechanism as a general-purpose tool beyond forecasting.

factualhigh valuespeaker onlynovelty 3/4durability 2/4· Stephen Grugett

“users have come up with all sorts of like exotic ways to use prediction markets that we hadn't previously considered... getting users to help research topics for you”

0.52

The missing piece preventing prior prediction-market platforms from scaling was the idea of user-resolved markets; while allowing the market creator to also be the judge opens the door to fraud, fear of that fraud pushed others toward centralized oracles that severely limit scalability, whereas in practice the fraud is small and manageable because users self-select into trustworthy markets.

causalhigh valuespeaker onlynovelty 3/4durability 3/4· Stephen Grugett

“the other key piece of the puzzle that other people were missing is the idea of user resolved markets”

0.49

Market creator reputation will follow a bimodal distribution: a few high-trust, high-volume market creators alongside a long tail of smaller individuals whose markets are participated in by their social-graph connections (friends and friends of friends).

forecasthigh valuespeaker onlynovelty 2/4durability 2/4· Stephen Grugett

“i think it'll be kind of a bimodal distribution well you'll see a few huge market creators that are high trust... and then there'll be a very long tale of smaller individuals”

0.44

A world where prediction markets are liquid and widely understood would be better because grounding people in well-calibrated predictions instead of ideologically biased speculation enables more productive conversations; one could imagine news articles or CNN segments with embedded markets that instantly ground opinions in fact.

forecasthigh valuespeaker onlynovelty 2/4durability 2/4· Stephen Grugett

“it's better in nearly but not all cases when people are grounded in terms of facts and like well-calibrated predictions instead of like pure speculation or like ideologically biased thinking”

0.44

A global leaderboard becomes less important as the platform grows; domain-specific or community-scoped leaderboards (e.g., hand-picking a few Russia-Ukraine markets to find the best geopolitical predictor) give a more meaningful assessment of skill than a global ranking that mixes serious and trivial markets.

normativehigh valuespeaker onlynovelty 2/4durability 2/4· Stephen Grugett

“the concept of a global leader board becomes less important over time the larger our site gets... in terms of ranking who the top traders or predictors are really is not all of the markets that we have but the subset of markets that are of interest to you”

0.34

Only a portion of news consumption is driven by a desire for accurate understanding of the world, so prediction markets can help that minority even if much consumption is driven by other motives; assuming most political news consumption seeks accuracy would be naive.

factualcontestednovelty 1/4durability 2/4· Stephen Grugett

“i'm not implying that a large part or even like a majority of the consumption of political news is based on a desire for accurate information... but at least a small part of that is people's desire to consume news actually is to legitimately understand the world”

0.25

Manifold's market mechanism is a dynamic parimutuel system that the founders derived from first principles before discovering it had a name in the literature, later finding a more elegant implementation based on a quadratic cost function.

factualspeaker onlynovelty 2/4durability 1/4· Stephen Grugett

“originally we actually came up with the idea for dynamic paramutual buy or sell thinking about it from first principles and then later i went back through the literature and read a bunch of papers and realized that it was called dynamic pair mutual”

0.23

Manifold's free signup bonus of 1000 manifold dollars is a growth tactic and not committed to indefinitely; giving away free money is not economically sound, and the company uses undisclosed anti-abuse measures to prevent bots and multi-account exploitation.

factualspeaker onlynovelty 1/4durability 1/4· Stephen Grugett

“if you're giving away free money that's not a you know economically sound mechanism we currently do do some things to prevent abuse and like bot behavior i'm not gonna tell you what those things are”

0.23

Manifold uses fees (a creator commission of about 4% to incentivize market creation and resolution, plus a liquidity fee that subsidizes the entire market) even though fees impede market efficiency by deterring marginal traders, accepting reduced liquidity to encourage desired behaviors.

factualspeaker onlynovelty 1/4durability 1/4· Stephen Grugett

“fees do impede the efficiency of the market in some sense... but part of the reason why we have fees is to encourage things that we like you know we have a creator fee the creator earns a commission on trades that incentivizes them to create and resolve more markets”

0.20

Manifold's previous startup Throne (a subscription group-chat app for online creators) failed to get growth because the creator market was already saturated with monetization tools and creators were not keen to switch platforms.

causalspeaker onlynovelty 1/4durability 1/4· Stephen Grugett

“the ultimate cause is it's kind of too late like the creator market is already saturated with all different sorts of tools most creators are already monetizing one way or another and aren't keen on moving monetizing using a new platform”

0.18

Manifold was founded in December 2021, ditched crypto after about a week of research due to regulatory and usability concerns, received the ACX grant from Scott Alexander, and launched a working prototype timed to the grant announcement to onboard the ACX community as its initial user base.

factualspeaker onlynovelty 1/4durability 1/4· Stephen Grugett

“basically decided to ditch crypto after like a week's worth of speculation and research... partially based off reading about the regulatory nightmare that is prediction markets and crypto”