Loud!, a simple token experiment, has dominated the conversation on CT last week. Powered by Kaito, Loud distributed 45% of its token supply to users who promoted Loud in a first-of-its-kind Initial Attention Offering (IAO). The novelty of the launch mechanism and upside speculation propelled Loud to 76% Kaito mindshare. After a down-only token launch, Loud, along with the broader InfoFi narrative, is being blamed for CT's downfall.
Loud’s distribution experiment indeed had a cancerous effect on CT. It failed to spark substantive dialogue, and incentive dynamics masked signal in favor of slop. Despite this, I consider Loud a poorly designed but ultimately successful experiment. Here’s why:
It is trivial to simply mute “Loudio”/”Loud”/”Stayloud” and remove oneself from the experiment.
Complaints tend to focus on the negative impact on the timeline rather than the distribution mechanism itself. As a productless token, LOUD will naturally generate the lowest-quality content.
The experiment had an unmasking effect on many KOLs.
The program’s poor design, short timeframe, and arbitrary rules (>10 smart followers) directly contributed to the quality of content.
Predictably, the token has been a miserable failure. LPs immediately undercut the 4% fee pool that was intended to fund post-TGE mindshare rewards in perpetuity. LOUD now sits at $2M. The market sent a strong signal that this wasn’t good enough.
Perhaps the most disturbing element of InfoFi is the ease with which poorly designed incentive structures can cause humans to behave like bots. It’s an unsettling realization to see that your peers have that bot in them, and that your timeline, the diligently curated lens through which you view reality, is so vulnerable to manipulation.
As the Loud experiment unfolded on X, an exciting new mini-app called Noice commanded a similar degree of mindshare on Farcaster, but for all the right reasons. This post will explore how Noice could remedy early InfoFi hiccups and accelerate the creator economy.
Noice is a Farcaster-native microtransactions app. It allows users to passively give and receive tips for every like, recast, quote, comment, or follow. Tips can be made in any token: USDC, ETH, custom tokens, etc. With the request feature, users can set prices for likes, comments, recasts, or even a call. These settings can be calibrated on the user profile.
These concepts have been implemented previously by Farcaster mini-apps Tipn and Amps. Tipn allows users to passively tip for each like, while Amps allows users to purchase likes and recasts from willing counterparties. Amps has made substantial progress with its app, screening posts for harmful content and offering filter options to catalogue the like/recast market. Noice combines Tipn, Amps, and Timedotfun into one platform. Noice’s design is made possible by an integration with Neynar.
Amps and Noice challenge the legacy social media marketing model with a peer-to-peer approach, facilitating direct payment for actions without an intermediary. Uber, for engagement.
Noice debuted two weeks ago with impressive explainer videos and a token launched via Clanker. The team has continued to ship features over the past week, bringing custom tokens, tip delegation, and keyword-triggered super tips.
Noice has completed over 870K tips worth ~$50K to over 24K unique recipients. Solana support, Monad support, and a potential X integration are upcoming growth catalysts.
There’s two angles that make Noice extremely interesting: Base and Microtransactions.
Base
The Base ecosystem has experimented with tipping to mixed results. Degen was an interesting early attempt, but it had little staying power. The traditional tipping UX is clunky, and the behavior can come across as performative or even spammy. More recently, Base x Zora’s thematically adjacent “Coin it” campaign has met substantial criticism.
Noice combines and improves upon both ideas. Support for custom creator tokens and a passive, gasless tipping experience bring a seamless, non-intrusive distribution mechanism for those who wish to experiment.
Noice has a nice tie-in with Zora, and it connects the dots on a lot of the creator economy ideas Base is pushing. Base’s recent marketing may be a bit corny, but they are clearly making an earnest effort to build something beyond meaningless coin launches. Base is making a multi-year bet on what’s next, and Noice fits hand and glove with this vision.
Microtransactions: an alternative approach to InfoFi
Microtransactions are an established but under-explored primitive. They have existed in the real world for a long time, but blockchains are a natural fit to more fully realize their potential. Blockchain infrastructure limitations have prevented experimentation here until recently.
Microtransactions have a massive scope in the context of Web3 social and InfoFi.
For Web3 social, Noice offers a seamless way to populate social graphs with a variety of identity and reputation-related data. Passive tipping is a less arduous method of bootstrapping the network with arbitrary data, allowing Farcaster to finally showcase the appeal of an open, programmable social graph. Mini-apps will become more composable, and each new mini-app will impact the meta. A cool example of this is demonstrated by Amps <> Noice arbitrage.
For InfoFi, Noice offers a substrate for trustless, accountable commerce in the engagement units of InfoFi. Noice explores InfoFi concepts from a very different angle than Kaito and Cookie.
Kaito is a top-down approach measuring relative mindshare. It is a simple, intuitive tool ideally positioned for both Web2 and Web3 brands. Yapper leaderboards allow businesses to seamlessly employ social media users as marketing contractors by incentivizing mindshare.
Noice is a bottom-up approach that deals in the granular actions that Kaito ultimately seeks to measure. A possible critique of tipping is that in the long run, tipping flows are a wash. A viral post may bring in a lot of money, while the average user is roughly break-even. This could prevent users from adopting a tipping lifestyle. Noice’s unlock, however, is reducing the friction of tipping to a degree that there’s little reason not to participate. Noice flips the script from “What’s the point?” to “Why not?”. So what if it’s mostly a wash? Noice makes every interaction mean something. It penalizes slop. Tipping is proof of stake in the social graph, table stakes for quality discussion.
Kaito’s competition-inducing design carries a publicity boost at the cost of quality and control. With the cost of its interactions, Noice may bring less of a publicity boost from algorithm gamesmanship, but it makes no sacrifice on quality. This could make Noice complementary to Kaito. Noice could be incorporated into the mindshare metric, or could be used in tandem by brands who wish to better curate and control the meta.
Noice has done a great job pushing out an MVP and generating buzz. It’s a refreshing implementation of tipping that improves upon prior attempts. But tipping, as it’s currently conceived, may be limited in scope. The main reason I’m intrigued by Noice is its potential to evolve into a foundational element of InfoFi and Web3 Social. So I’m going to disregard execution risk and discuss what I hope to see in a hypothetical Noice v5.
There are 3 major features I believe are necessary to get there:
Robust ‘request’ feature - Noice has a firm grip on the tipping side, but it’s request feature is currently disabled. It is unclear what the request feature will look like, but Amps has a formidable product. Noice needs a robust request feature that includes its familiar passive feel along with single-post targeting. Perhaps this could come from a merger or some sort of collaboration with Amps. Noice's impressive usage metrics suggest tipping has stronger appeal in the short term. In the long run, the request side may have a higher ceiling.
Extending programmability - Widening the logic available to tippers could be a breakthrough in finding PMF. The ability for users to say “Don’t merely interact with my post, do ____,” is a powerful unlock. Noice appears to be working toward this with a criteria feature.
New actions: impressions, reads - Impressions measure the number of times content appears on a user’s screen. It is a popular metric in market research, appearing on X and playing a significant role in Kaito’s mindshare metric. Measuring the extent to which a user engages with or reads a post is far more difficult. Reading insights can be estimated with metrics such as dwell time and scroll depth. There are technical hurdles here, but adding impressions and reads as transact-able actions on Noice would be a major breakthrough.
These features would nicely round out the feature set and blow open the design space for things to build on top.
Launchpad
LOUD offered a proof of concept for Kaito-assisted TGEs. While Loud! used Kaito’s mindshare metric to allocate incentives, PROCOIN and KART attempted to drive mindshare by directly incentivizing interactions.
Only days after Farcaster sold 10K Pro membership NFTs, a community governance takeover quickly took shape. The token, $PROCOIN, was distributed to NFT holders who quotecasted the announcement post with “ratifying @procoin governance.” Procoin didn’t use Noice for its launch, but logic and targeting support would have made it feasible.
Ciniz (Henlo Kart, Onchain Gaias) demonstrated Noice’s token distribution utility with KART tokens. The Noice team is aware of this use case (“We aren’t just a launchpad”), so additional features enabling this seem likely.
Doomscroll-to-Earn
The Noice and Kaito TGE comparison begs the question: why pay to drive a proprietary mindshare metric (Kaito) and forfeit control over emergent dynamics? Further, why reverse-engineer engagement by trying to calculate the number of impressions a marginal like is worth (Noice, Amps)? Paying for impressions or reads directly without Google, X, Meta as middlemen would be ideal.
Noice could achieve this by letting users set a Request price for both impressions and reads. The Noice mini-app could act as an alternative Farcaster client where users get paid to scroll. The timeline would be composed of their original following feed, with additional content from Requesters, which could be individual users or established brands.
An impression puts the post on a user’s timeline, while a read puts the post in a separate inbox. Users can set different prices for different entity types (mutuals, extended circle, randoms, businesses).
The lack of a reliable and practical read measurement is a bit of an issue here, but reputation could play an important role, giving us a fair amount of wiggle room for a small-scale implementation.
For example, instead of finishing analysis threads with “Tagging DeFi chads who may be interested!” users could tip these individuals via the request read feature. As reputable KOLs who opted-in by setting a price, they are likely to honor it or at least give a refund.
Further, imperfect reading metrics such as scroll depth and dwell time could potentially suffice. A user whose time commands a price that makes gaming these metrics lucrative likely isn’t concerned with this, or the potential reputational hit would outweigh the benefit. Alternatively, users who cannot command high prices and/or high volume are less incentivized to try and exploit it. Eventually, we could try some sort of TEE-zkML eyeball tracker setup if we want to solidify the design and onboard advertisers.
Curation
Much of the backlash around InfoFi at the moment is concerned with the negative impact on the timeline. Farcaster could allow Noice-powered channels with channel-mandated tip settings. A channel could require a tip to cast, with the pot being distributed among top posts via quadratic voting (liking). The end result is a variety of semi-public private groups with high signal-to-noise ratio.
Channels could distribute a channel-specific valueless UBI token to users based on their consumption of the channel. The token is tipped with likes, creating a activity weighted republic of algorithm curators. The end result is a fresh take on the FML/TIFU/MLIB forums of the early 2010s, and a much wider design space for channels.
More generally, tips could act as a complimentary input in the ‘For you’ algorithm, offering a familiar Reddit-like experience for those who desire it.
Spam filter
Physical mail used to be fun, because the costs associated acted as a natural filter for intentional, necessary content. Today, the USPS is a federally funded slop courier. We’ve now grown accustomed to dodging relentless spam across most communication methods: mail, email, robo-receptionists, inbound phone calls, and social media bots.
Noice users could require tips to receive messages or comments. If the comment is liked or the DM is read, the sender could receive a refund. The system would effectively act as slashing for spammers. This concept was recently introduced by Telegram. Premium users can set a price per message in their privacy settings, payable in Telegram Stars. If this was somehow possible for email, mailboxes, and phone calls, it would be an overnight success.
To extend this concept to voice calls, Noice could be integrated with a VoIP app on Base (Voice over IP; internet-based calls e.g., FaceTime, Telegram). Users could require non-contacts pay a fee to call them, perhaps receiving a refund if the user doesn’t hang up with a spam button.
Extensible logic
The recently released super-tip feature enables users to set a tip multiplier any time they comment “ok banger.” Customizable trigger phrases are coming soon. Noice’s intention of extending the tipping logic is encouraging. There are many existing social media interactions that are poorly executed or unavailable due to inadequate infrastructure:
“Giving away 1 ETH to someone who likes this tweet!”
“Giving away 1000 USDC to the best investment theses in the comments!”
“Follow for follow!”
The ability to pledge an action to a post would allow these interactions to be carried out in a trustless way. Pledge 200 USDC in tips on a post, if it is not awarded, the 200 USDC defaults to a lottery.
Sure, some of these ideas are a bit out there, but Noice offers a lot of room for experimentation and it should be celebrated.
Here are a few concerns that could inhibit Noice’s traction:
Product market fit for tipping - Noice has been well-received early, but the path to PMF remains unclear. Do we really think the tipping meta has long term legs? It likely needs to evolve into a grander vision.
Lack of a moat - Noice has hit the ground running with its initial rollout, but it lacks a moat and there are existing competitors such as Tipn and Amps.
Business model and token - There is a high degree of uncertainty regarding the revenue model and tokenomics. It will likely be an extremely low-margin business, but it could benefit from the the unknown ceiling element that benefits the L1 trade.
InfoFi will likely remain a polarizing narrative. There is a visceral backlash against adding what seems like an unnecessary layer to everyday interactions. No one wants to live in a world of Loudio slop or calculated small talk like Nosedive.
Nonetheless, InfoFi is unique, controversial, and powerful enough to backbone a bull market. It is absolutely a DeFi summer, NFT, AI agent-caliber narrative. After many years of circular ideas and oracle-perp DEXs that fail to impact the real world, we now have something Web2 will love, yet CT ridicules it. Kaito is now expanding to Instagram and TikTok. A Loud! experiment on TikTok could be a watershed moment for Web2 marketing companies.
In 2017, Sunny Co Clothing’s viral red swimsuit marketing stunt promised a free $65 bathing suit to users who reposted a model image. The clothing startup went from 7K to 753K followers overnight, collecting over 50K emails and selling out of stock immediately. This campaign was viewed similarly to Loud! at the time, with users unhappy about the deluge of identical photos on their feed, and participants unsure about the integrity of the deal.
The lesson here is that the negative externalities we saw with Loud! are inevitable, with or without InfoFi. Businesses will always seek an edge in reaching the customer, and InfoFi offers one. Kaito leverages social media data to build an arena of attention. Noice brings flexibility, control, and trust to small-stakes interactions, allowing for better execution of creative Web2 marketing initiatives.
When we invented cars, people traveled through space much faster than before. A lot of them crashed. Then we installed airbags and learned how to drive, and society was better off. Experimentation—both good and bad—is an important part of growth. Noice unapologetically explores the transactional element of social interactions. It brings a new method of control that has been missing from the InfoFi stack. With proper execution, Noice could help repair the social fabric of the digital world.
Noice is Degen + Tipn + Amps + Timedotfun. Microtransactions is a powerful concept bringing a fresh angle on several narratives, including InfoFi and Web3 Social. Doxxed, 5-person team with KOL energy sitting comfortably in the “In” crowd on Base. Production value of the promo videos is striking, and the project generally appears to have the “it” factor. An upcoming X integration and support for Solana and Monad could be major catalysts for growth. Base is very likely to push Noice. It has a nice tie-in with Zora and could be the missing puzzle piece to the creator economy narrative Base is committed to.
PMF is no guarantee, and there is very little clarity on the revenue model and token economics. But this is a new, disruptive, intuitive concept that brings a lot of whitespace to work within. It seems obvious to me that Noice belongs in the blue-chip Base builder basket alongside Clanker, Bankr, and Zora. There is a fair amount of execution risk at this stage, but from a market cap to future mindshare perspective, Noice is a compelling play at $12M FDV.
This article was originally published on Delphi Digital’s Alpha Feed on Jun 6, 2025.
Disclosure: NOICE holder
Jordan
807
Ok. Now I'm really grasping the coolness of this
I'm really excited about @noiceapp and all of the experimentation it enables. Tap in to brainstorm doomscroll-to-earn with me!