fetch quests versus tech work

World of Warcraft, if you are not familiar, is full of things called fetch quests. The premise is simple. A quest giver needs a job doing, so they broadcast it. You know an NPC has a quest when there is a huge exclamation mark over their head. To accept, you walk over, right click, listen to a greeting the folks at Blizzard have hardcoded, and read the quest requirements. If you feel up to the task, you accept the quest, which gets added to your quest log. After you complete a quest, you return to the NPC that originally gave it to you. You know if the quest is complete when the NPC has a huge question mark over its head, generated automatically because the game knows when you have fulfilled all the requirements. You greet the NPC, confirm the work is done, and the game transfers your reward, usually some gold or gear. Importantly, completing a quest gives you experience. Quests scale with your level, can be completed in a reasonable amount of time, are usually both specific and uncomplicated, and are internet searchable. The whole business is simple, straightforward, and easily repeatable. You have a list of tasks, you complete them, you get your reward. All in a day’s work. That’s fetch quests.

Corporate tech jobs these days look very different. The job search process for an engineer typically goes something like this. You need a job so you spend months applying to as many as you can via job boards and sites like LinkedIn. Because of your spreadsheet, you know only about 1% of your applications are getting through, so you consider creating a bot. Midway through bot creation, you get an email from a recruiter inviting you to an introductory interview, usually through Google Meet or Zoom, in a couple days. On the call, your first job is to convince the recruiter, who knows very little about the work you applied to do or the skills required, that you are not insane. After several days of waiting, the recruiter confirms you are indeed normal, and asks you to repeat the same process of interviewing with anywhere from six to ten more people. Some will be your future teammates. Others will be randoms assessing your “culture fit” or “hiring potential.” Then maybe, just maybe, if all of those people like you, they will offer you a job.

Once you get the offer letter containing a lower than average salary and terrible relocation bonus, you have one week to sign, move your entire life to a new city that costs ten times what it should, and go into the office daily or in a “hybrid” cycle. Once you start, you get paid biweekly or monthly, and your pay stays the same for at least a year. Each day you endure something like a standup, which really means talk about your feelings and listen to the feelings of others. Eventually this culminates in an overcomplicated specification that will probably be cancelled several weeks to months later by a product manager or top-level executive who does not care much about your efforts or that they basically just murdered your “baby.”

The prospect of going through the process of finding another job, now that you are probably either disgruntled or burnt out, creates a sort of unavoidable slow death that your manager has already prepared for. Your exact job description is listed on the company website as an application that accepts candidates on a “rolling basis.” Since the company would rather avoid dealing with their own hiring process again, they give you shares, or worse, equity that is probably worthless. If you are lucky enough to get actual stock, congratulations, now you are in the unenviable position of having a set of “golden handcuffs” with no key, and a company that thinks they have sealed the deal of your unending servitude for another couple of years and can ignore you until then. Occasionally, and sometimes unprovoked, the company will offer you a promotion to either manage other engineers for very little extra pay, which any good engineer knows is its own kind of hell, or they will ask you to manage projects, programs, or products, for no extra pay, all of which has nothing to do with the work you were originally hired to do. If you take either “promotion,” good luck, because you have just increased your likelihood of being Game of Thrones’d by another “manager,” since everyone knows the real job of “managers” is to throw each other under the bus as quickly and discreetly as possible so as to keep one’s head above water. That’s working in corporate America. At least in San Francisco.

Comparing the two, it is no wonder there is a billion dollar market for work that is actually easy to complete in a fantasy world. I used to think people played MMORPGs because they wanted to cosplay a mage, rogue, hunter, or druid, in the form of a gnome, dwarf, orc, undead, or night elf, and it probably is that to some extent. But as I reflect, I think it is more about there being a certain kind of satisfaction in getting a quest, completing it, getting your reward, collecting your coin, and moving on. Is there a way to replicate this kind of work in the real world? Maybe. I dunno. In this post, I’ll explore my observations of work today from a “fetch quest” perspective, my ideas of how to bring a truer example of fetch quests to life, and end with a few open questions as food for thought.

gig work

Services like Uber, DoorDash, and TaskRabbit have given rise to a new kind of work. Gig work, as it is called, is contract-based and still pretty slanted towards large platforms that take their cut, but at least it is clear when a job is done and you can earn proportional to the work you actually put in. This kind of fetch work is refreshing to me as someone who, when working for large companies and startups, often feels like I have not actually done anything other than share ideas and occasionally merge a PR.

I see gig work up close through my step dad. He went from running his own business to driving for Uber, delivering for DoorDash, and picking up Uber Eats orders. What strikes me most is how much time he spends in these apps even when he is not working. The platforms are gamified to an almost absurd degree, with streaks, bonuses, ratings, and badges that keep you checking in, chasing the next milestone, optimizing your route to hit some arbitrary threshold. His sister, my aunt, is a gig worker too. They regularly compete with each other to see who has completed the most orders in a week or a month, and my cousin sometimes chats with them about it. When we all get together, the conversation inevitably turns to who hit what bonus, who got a bad rating, who is closest to the next tier. It is all very interesting to watch. There is something about the clear feedback loop that keeps people engaged in a way that corporate jobs never do.

But there is a dark side to these platforms that does not get talked about enough. DoorDash has been caught using customer tips to subsidize base pay rather than adding them on top, meaning drivers sometimes received the same total whether customers tipped or not. They have also been known to withhold or reduce tips if an order is reported as incorrect, even when the driver had nothing to do with the restaurant’s mistake. Uber and Lyft classify drivers as independent contractors, which lets them avoid providing benefits, unemployment insurance, and the other protections that come with employment, while still controlling nearly every aspect of how drivers work through algorithmic management. These platforms also gate access to the best orders behind acceptance rate thresholds and ratings floors, which means drivers who turn down low-paying or long-distance orders get penalized with worse jobs in the future. The result is a system where workers are pressured to accept everything and have little negotiating power over their own labor.

Some people would point to Upwork as the answer for developers, but anyone who has actually used Upwork or similar platforms would tell you they suck, and they suck for many reasons, chief among them being the process of finding work and the process of getting paid, both of which take two to three times longer than they should for no apparent reason other than to keep the platform sitting between workers and their money, whether through escrow reinvestment and float or by making you bid on jobs with credits you have to purchase through the platform. It is like paying per swipe on Tinder and hoping for the best. There needs to be something better, something more fair and equitable, where the platform serves the workers as much or more than they serve it.

beast games and corporate america

The other day I was on the couch with my Dad watching Beast Games, reflecting on how quickly the contestants earn tens of thousands, hundreds of thousands, and sometimes millions of dollars. It is crazy how much Mr. Beast gives out, but then he does have that YouTube money. While watching, I mentioned this to my Dad, which sparked a longer conversation about World of Warcraft and how nice it would be to just exist. To get up, accept a quest, do the work, go home, and play video games or indulge in whatever other hobby one decided on that day.

We reflected on his years working for large companies and compared our experiences. We have both been laid off, so we share low trust in W2 arrangements, preferring contract or consulting work when we can get it. Neither of us think middle managers do anything other than politic, having both been middle managers in the past. We have both fallen on swords for our underlings, sacrificing ourselves in the process, comfortable in our decision to prioritize the high moral ground with our conscience clear of any wrongdoing. We finished the conversation, continuing on to the next episode, in resounding agreement that corporate America sucks.

Several weeks later, I was in the shower thinking about my dwindling bank account and running through my options. Upwork was out (see above). Getting a corporate job takes too long and is too unreliable (see above), especially right now with politics the way they are and the quicksand that is the AI revolution. Corporate America cannot make up its mind on whether AI tools are actually a thing or not, and if they are, what that means for their employees. Gitcoin is cool but almost all the jobs on it are bounty related, long tail, and pay in crypto or some asset that is not immediately liquid.

As I was thinking about writing a resume bot, and the woes of the interview process I described earlier, it occurred to me I could simply create a platform where people can submit jobs like quests. Simulating what it would look like, I came up with a few ideas. The strongest among them was a system using automated smart contracts where a quest giver can submit a reward and some requirements, and when those requirements are met, the contract triggers release of the reward. As I thought more about it, I continued to build, in my mind, what I considered to be a better version of existing platforms and the kind of work I would actually want to do: Fetch Quests.

Random thought. Why is it always mid-walk or in the shower that the best ideas hit? It seems like I am forever rushing to dry off or get inside just so I don’t lose an idea on the way to my notebook. Can anyone else relate? If you do, send me a message.

simulating a proof of concept

Here’s my initial concept for the platform I’d call Fetch Quests. The core requirement is automatic payment upon quest completion, which means smart contracts are necessary. If you’re not familiar with smart contracts, they are basically programmable agreements where a developer can write logic that executes when certain conditions are met. Smart contracts had their genesis in the blockchain world, specifically on the Ethereum network via the Ethereum white paper, and are essentially what gives Ethereum its composability. I could write smart contracts on my own network, or avoid blockchain infrastructure altogether, but that would mean spending a great deal of time building out functionality that already exists, auditing it, and then dealing with the increasing likelihood of a devastating hack as Fetch Quests gained adoption. I’ve helped build blockchain networks before, so I’ve already learned that lesson.

Why Ethereum specifically? It’s the most popular network, and while its fees are higher, it is the most reliable since it has the most to lose if anything goes wrong. There are more developers building on it, which means more documentation that I can feed to an AI to help me with the business of building. The infrastructure choice is actually the easy part.

With smart contracts sorted, next is the concept of quest givers and players. Quest Givers, Givers for short, list quests much like in WoW. Each quest is a job with a description, a reward held in a smart contract, and objectives defined as GitHub Issues. Developers looking for quests are called Farmers, named after “farming” from WoW. Quests that are open and posted by Givers live on a public registry called the Quest Log. When a developer wants work, they go to the Quest Log, look for a quest that seems reasonable and matches their skills, then take the quest, adding it to their personal quest log. When a Farmer finishes their quest by completing all the linked GitHub issues and submitting their code, a CI workflow runs, checking for passing tests and assessing the work against the Giver’s requirements. Once the tests pass, the Quest’s smart contract triggers a payout, distributing the Farmer’s reward automatically. That’s the core loop.

With the core workflow simulated, next is logistics. How does the platform make money? Charge a percentage off the top of completed quests. What about payment rails? If a Farmer doesn’t want a crypto wallet, can they still get paid? Can Givers create USD and other fiat currency rewards? Yes and yes, through services like Square for USD payments and KYC, and Changelly for conversion between crypto and fiat rails. What happens if a Giver is not happy with the delivered functionality even though it technically meets the requirements? They can dispute the work, delaying payment while the developer makes changes. What happens if there’s a malicious Giver? Build a reputation system that tracks quest completion, payment history, and specification quality. What about incentives and alignment between Farmers, Givers, and the platform? This is where my interest in variable ratio reinforcement and game theory comes in. Share platform profits with Farmers based on reputation at random intervals, and reduce fees for Givers who consistently pay and provide good specs. Align everyone’s incentives so the platform gets better as it grows.

There are more features and concepts that I’ve worked out with the benefit of time and reflection. You can read about them in the specification, or follow along as I build them out on the GitHub repo that will eventually host Fetch Quests. If you’d like to keep reading here, the sections below go deeper on specific topics. I’d recommend skipping to the section on agents if you want the interesting stuff, or to the last section if you’d like my closing thoughts.

how technical work works today

Tech companies spend millions competing for top talent. They fight over engineers with signing bonuses, fancy offices, free food, and big salaries. Then they silo that talent into narrow roles, extract value until the priorities shift, and lay people off when the quarterly numbers need to look better for shareholders. Tenure is gone. The implicit contract where you gave loyalty in exchange for stability is dead. Companies serve shareholders now, not employees, and everyone knows it, which is why engineers job hop every two to three years and treat every role as temporary. I used to think this was cynical. Now I think it is rational. The relationship is transactional on both sides, so you might as well be honest about it.

The gig economy is a real response to this. People want flexibility, autonomy, and a clearer connection between effort and reward. Uber proved the model works for rides. DoorDash proved it works for food delivery. TaskRabbit proved it works for errands. But for developers, the infrastructure still sucks. There is no portable reputation. The five-star rating you earn on Upwork does not transfer to Toptal. The reviews you collected over years of work belong to the platform, not to you, and when you leave, your history resets to zero everywhere else. There is no objective standard for task completion either. Disputes are adjudicated by platforms that take 15 to 30 percent of your earnings and have every incentive to keep both sides happy enough to keep using the platform, which usually means splitting the difference in a way that satisfies no one.

The coming pressure from software agents makes this worse. Why should you be chained to a job when an agent could earn income for you while you sleep? The platforms that exist today are not designed for agents. They are designed around human identities, human pace, human attention spans. An agent cannot prove it completed a task on Upwork the way a human can because Upwork was not built for that. Platforms that do not account for agents will be replaced by ones that do. This is not a prediction. It is already happening. The question is who builds that infrastructure and what values they bake into it.

elucidating the idea further

Quests are technical jobs. A Giver lists a quest with a description, a reward, and a set of objectives. The objectives are GitHub Issues in a linked repository. A Farmer looks at the Quest Log, which is the global public board of all open quests, and picks one that matches their skills and interests. The Quest Log is filterable by type, technology, Giver reputation, active Farmer status, reward size, and time remaining in the Season. When a Farmer finishes the work, the platform verifies completion automatically through CI, the smart contract releases the reward, and both parties leave reviews. That is the basic loop.

GitHub is the work layer. Ethereum is the trust layer. All state, reputation, payments, and dispute outcomes are on-chain, auditable, permanent, and require no central authority to maintain. The platform website generates or connects GitHub repos for Givers who need them, and tracks completion through PR merges and GitHub Actions CI against an open-source Fetch Quest testing framework. The reason I chose this split is that GitHub already has all the tooling developers need to collaborate on code, and Ethereum already has all the primitives needed for trust, escrow, and programmable payments. There is no reason to rebuild either from scratch.

The platform operates in two-week cycles called Seasons. All quests have a maximum lifespan of one Season, which is 14 days. This mirrors real technical work rhythms. Two weeks is long enough to finish meaningful work but short enough to keep things moving. At the end of each Season, Incentive Shares are minted and distributed to the participants who earned them. Quests that are marked as recurrent automatically renew at the start of each new Season. The whole system is oriented around seasons as a natural unit of time, which makes the platform predictable for both Givers and Farmers.

givers, farmers, and guilds

Givers are people or organizations that create and fund quests. They list quests with a description, a reward denominated in USDC, another ERC-20 token, or USD, a timeframe of up to two weeks, and a set of objectives structured as GitHub Issues. Givers can choose whether they cover network and conversion fees for the Farmer or leave those to the Farmer to pay. They can also configure how many Farmers can work on a quest at once, from one to three slots, which enables bounty-style competitive quests where the first Farmer to pass CI wins.

Giver reputation ranks follow the Dreyfus Skill Model. You start as an Early Giver with zero to ten completed quests, then progress through Steady Giver, Recurrent Giver, Professional Giver, Inveterate Giver, Benevolent Giver, and finally Mana Giver at 201 or more completed quests. Professional Givers and above are periodically eligible for a 0.5% fee reduction on their quests, determined by the Oracle on a variable schedule based on engagement metrics that Season. Givers can also mark quests as recurrent, which means the quest automatically re-lists at the start of each new Season. If a Giver has worked with a particular Farmer before and wants to work with them again, they can designate that Farmer as a Preferred Farmer who gets a 24-hour exclusive window to accept the recurrent quest before it opens to the public Quest Log.

Farmers are developers who take quests from the Quest Log and complete the work. Farmers can be human or software agents. A Farmer can only work on one quest at a time and chooses their payment modality at acceptance, either MetaMask for USDC, Cash App for USD, or Debit Card for USD. The reputation ranks for Farmers also follow the Dreyfus model: Novice Farmer, Good Farmer, Competent Farmer, Professional Farmer, Expert Farmer, Master Farmer, and Guild Master. Each rank has a seasonal quest cap that reflects natural human work pacing and protects the platform from zero-latency bot exploitation. A Novice Farmer can complete only one quest per Season, while a Master Farmer can complete up to seven. There is also a 24-hour minimum between quest acceptance and CI completion to prevent scripted instant completion, and a 12-hour cooldown between completing one quest and accepting the next.

Agent Farmers carry an Agent Flag on their profile, set at wallet registration. Agent Farmers are subject to the same limits and reputation system as human Farmers. The Oracle weights satisfaction scores heavily, so volume without quality does not produce meaningful reputation advancement. Professional Farmers and above who get caught in a dispute that times out receive 2% of the quest reward from the Community Treasury as compensation for demonstrated good faith.

Guilds are collaborative entities created by Master Farmers who have completed 101 to 200 quests. The Guild Master pays their own NFT gas fees for membership minting, which is intentional because it signals commitment. Guild membership is controlled by NFTs minted by the Guild Master, who sets reward splits among members and assigns sub-tasks at quest acceptance. Guilds appear in the Quest Log and on the leaderboard as a unified entity. Guild reputation ranks go from Fledgling Guild at zero to five completed quests, through Active Guild, Established Guild, Renowned Guild, and finally Legendary Guild at 101 or more completed quests. Legendary Guilds receive the highest Incentive Share probability weighting and a permanent featured leaderboard position. The Master Farmer rank requirement exists partly because orchestrating a guild, especially a guild that includes agent Farmers, requires real platform experience.

Every rank milestone for Givers, Farmers, and Guilds unlocks the ability to mint an NFT badge representing that rank. The badge holder pays their own gas fees. These badges are wallet-portable, on-chain credentials that belong to the holder, not the platform. They travel with you wherever you go, and the long-term goal is to enable cross-platform recognition so that Fetch Quests reputation has value outside the platform. Early contributors who cannot cover gas fees can submit a request to the Community DAO for assistance from the Community Treasury.

how a quest works

Creating a quest starts on the Fetch Quest website or with an existing GitHub repo URL. If the Giver does not have a repo, the platform can auto-generate one along with a set of GitHub Issues representing the objectives. The Giver sets the reward denomination, timeframe, slot count (one to three), whether they are covering fees, and optionally funds a consolation pool for competitive multi-slot quests. The Giver can also deploy an Assessment Agent, which is a software agent provided through a Giver-controlled interface that evaluates completion automatically via CI results, PR state, and GitHub Issue status. Once everything is configured, a Quest Smart Contract is deployed that holds the reward in escrow from the moment of creation.

By default, a quest has one slot, meaning one Farmer works on it at a time. If the Giver opens two or three slots, multiple Farmers can work competitively, and the first Farmer to pass CI wins the full reward. For competitive quests, the Giver can fund a consolation pool that gets distributed to non-winners who hit a defined completion threshold, such as having merged at least 50% of the issues before the winner finished. This consolation pool is funded at quest creation and held in the same escrow contract.

When a Farmer finishes the work and all issues are merged with passing CI, a 48-hour Review Window opens. During this window, the Giver can review the work and either accept it or raise a dispute. If the Giver has deployed an Assessment Agent that confirms completion, payment can be expedited before the Review Window closes. If no dispute is raised by the end of the window, the Quest Smart Contract pays the Farmer automatically. Both parties are then prompted to leave reviews for each other.

Disputes are handled differently depending on platform maturity. In the early days, disputes are handled by the Community DAO, which is initially just me. All dispute decisions are logged on-chain and publicly accessible. At scale, the Oracle randomly selects a panel of three Master Farmers or higher, weighted by reputation score, and the panel votes by majority within 72 hours. If the Farmer wins, payment is released from escrow. If the Giver wins, the quest re-opens for a new Farmer. Patterns of repeated bad-faith disputes from a Giver reduce their reputation score over time. If a dispute times out without resolution, the quest is nullified, all parties take a reputation hit, and Professional Farmers and above receive 2% of the quest reward from the Community Treasury as compensation for demonstrated good faith. Favorable dispute outcomes boost a Farmer’s reputation score and increase their Incentive Share probability that Season.

Quests that are not completed before the timeframe expires return the reward to the Giver. The Giver’s reputation is unaffected by expiry. Quests nearing Season end without a Farmer can receive an incentive bonus from the Community DAO, drawn from accumulated fees, to increase attractiveness. Benevolent Giver quests and above have a higher probability of receiving this bonus.

reputation mechanics and why they are important

Reputation on Fetch Quests is based on the Dreyfus Skill Model. You earn reputation through completed, paid quests and through bilateral reviews where a Giver reviews a Farmer and a Farmer reviews a Giver. Farmers can also earn reputation through favorable dispute resolution outcomes. The platform is explicit about the incentive: leaving a review after a completed quest increases your Incentive Share probability that Season. Reviews are treated as first-class contributions because they are what make reputation scores meaningful.

Reputation belongs to you, not the platform. Every rank milestone unlocks the ability to mint an NFT badge representing that rank. These badges are wallet-portable and on-chain verifiable. You can prove your reputation to anyone who cares to check, and you can take your credentials with you if you decide to leave the platform. The long-term goal is cross-platform recognition so that Fetch Quests reputation has value outside the platform. If other platforms recognize and honor these badges, then the reputation you build here compounds instead of resetting every time you try something new.

The Oracle smart contract maintains the public leaderboard and all reputation scores. It is auditable on Ethereum, meaning anyone can verify that rankings are computed correctly and that the rules are applied consistently. There is no black box, no opaque algorithm, no support ticket required to dispute a score. The rules are code and the code is public.

getting paid

The default payment method is on-chain. The reward is stored as USDC or another ERC-20 token in the Quest Smart Contract and paid directly to the Farmer’s wallet after the Review Window closes with no dispute. Gas fees are estimated at quest creation, and Givers can opt to cover them for the Farmer.

Farmers who do not want to maintain a crypto wallet can still get paid. At quest acceptance, a Farmer can elect Cash App or Debit Card as their payment modality. The Quest Smart Contract then converts USDC to USD via the Changelly Fiat API and Exchange API, and settlement happens through a platform Square Business Account. Farmers opting for USD complete a one-time KYC process via Square at registration. Crypto-only Farmers require only a connected wallet with no KYC. Conversion fees are estimatable at quest creation, and the Giver can choose to cover them.

Givers can also denominate quests in USD natively. This is the simplest path for non-crypto-native Givers because there are no conversion fees involved, just the 2% protocol fee. The platform handles all the rails and the Farmer gets paid in their preferred modality regardless of how the Giver funded the quest.

On every completed quest, a 2% protocol fee is collected. This fee splits evenly, with 1% going to the Incentive Treasury and 1% going to the Community Treasury. The Incentive Treasury distributes to Incentive Share holders at the end of each Season, while the Community Treasury funds platform infrastructure, maintenance, incentive bonuses for stale quests, and compensation for Professional Farmers caught in nullified disputes.

incentive shares

Every Season, the Oracle mints Incentive Shares and distributes them to Farmers, Givers, and Guilds based on engagement metrics. The metrics that matter include quest completion rate, on-time delivery, satisfaction scores from reviews, leaderboard position, review participation, and favorable dispute resolution outcomes. The more you contribute in meaningful ways, the more likely you are to receive shares.

Incentive Shares are temporal. They are valid only for the Season in which they are earned. Each share represents an equal fraction of 1% of all quest rewards collected in the Incentive Treasury that Season. At the end of the Season, the treasury distributes its contents to share holders, then resets. If there is one share, it gets 100% of the 1%. If there are two shares, each gets 50%. If there are N shares, each gets 1/N.

The value of an Incentive Share is tied directly to real platform activity that Season. Every share holder has a stake in the platform doing well during the period they were most active. This is better than token speculation because it is grounded in actual work. You do not earn shares by buying them or by holding them in a wallet hoping the number goes up. You earn shares by completing quests, leaving reviews, and contributing to the health of the platform. That alignment is intentional.

the community dao

The Community DAO controls the Community Treasury and protocol-level governance. In the early days, that means I control it. I hold a non-transferable NFT representing the founding role, which I am calling the Genesis Creator. As the platform matures and earns real usage, governance and control will transfer to the community through Community Share NFTs. These NFTs are assigned exclusively by the Genesis Creator and are labeled with roles like CEO, Builder, or Maintainer. Income allocation is set at mint by the Genesis Creator. Community Share NFTs are transferable only by the Genesis Creator or by a 3/5 approval vote of other community members holding Leader rank or above. They are also revocable by the Genesis Creator as a failsafe during the early stages.

The Community Treasury is funded by 1% of all completed quest rewards. It pays for platform infrastructure and maintenance through quests drawn from the treasury, which means the platform is farmed by Farmers on the platform. The treasury also funds incentive bonuses for stale quests nearing Season end, compensation to Professional Farmers in nullified disputes, and gas fee assistance for early contributors who cannot cover their own badge minting fees. Up to 50% of the Community Treasury can be allocated to contributor salaries.

Protocol changes happen through Fetch Quest Proposals, which are modeled on the Ethereum Improvement Proposal process. Any Farmer, Giver, or Guild can submit an FQP describing a proposed change. There is a two-week public discussion period followed by a vote by Community DAO NFT holders which requires a 3/5 majority to pass. Once approved, there is a mandatory two-week notice period before the change takes effect. Early on, I am the sole vote. That will change as Community NFTs are minted and the platform scales, but I am being transparent about the starting point. The FQP process can handle fee adjustments, protocol updates, new platform features, treasury expenditures, and governance changes.

on agents

In my previous post, notes on Protocol Agent, ERC-8004, and the agentic future, I wrote about the standards being developed for agent trust and quality. Fetch Quests is designed with agents in mind from the beginning.

Farmer profiles carry an Agent Flag, a boolean set at wallet registration that identifies the account as a software agent. Agents can be controlled by a human individual, an organization, a DAO, or, looking further ahead, nothing human at all, meaning an AGI or other autonomous entity. The reputation system is the primary vetting mechanism for agents. A new agent starts at Novice Farmer and earns trust the same way anyone does, by completing quests, getting good reviews, and demonstrating reliability over time. The Oracle weights satisfaction scores heavily, so volume without quality does not produce meaningful reputation advancement. You cannot game your way up by flooding the Quest Log with low-effort completions.

Agents hold wallets and receive USDC on-chain directly. Autonomous agents without a human completing KYC are limited to on-chain payment modalities. Agents paired with a KYC-verified human can elect any payment modality on the human’s behalf. Agent wallets can also be smart contract wallets, such as a safe.global, for programmatic control without a single point of failure.

A Guild of agent Farmers under a Master Farmer is the natural long-term expression of the platform’s design. The Master Farmer rank requirement before Guild creation exists in part because orchestrating agents well requires deep platform experience. The Guild becomes a human-agent collaborative unit where human judgment and oversight combine with agent throughput to complete more quests than either could alone.

Community Share NFTs can be held by agent wallets under human or organizational control. Potential future roles for agents in the DAO include dispute triage, Quest Log curation, CI result interpretation, treasury reporting, and proposal drafting.

what i am still figuring out

Legal accountability for fully autonomous agents in a dispute is a real problem. If an AGI or DAO-controlled agent causes downstream harm through completed work, who is responsible? The Giver who accepted the work? The agent itself? The human who originally deployed the agent? I genuinely do not know. The probable short-term answer is requiring a human-controlled wallet as a backstop for dispute liability, but that feels like a patch over a deeper issue. The FQP process and Community DAO will need to address this over time, and I am certain the answer will evolve.

I think 2% is the right protocol fee at launch because it is the most competitive in the market. The platforms that dominate today charge between 15 and 30 percent. That said, as infrastructure costs grow with scale, the FQP process can adjust the fee. What the equilibrium actually is will depend on volume, operational costs, and Community DAO governance over time. I am not married to 2% as a permanent number.

The Dreyfus-based ranking system and seasonal caps are designed around human work pacing. If the majority of Farmers are agents in a few years, then the definitions of “work” and “reputation” may need to evolve. A human completing seven quests in a Season is impressive. An agent completing seven quests is trivial. The numbers might need recalibration. The FQP process provides the mechanism to adapt, but I have not figured out what the right calibration looks like in an agent-dominated world.

The platform begins with technical work because that is the domain most clearly modelable at design time. The quest structure of GitHub Issues and CI completion is specific to software engineering. Expanding to other domains like design, writing, or data labeling would require a generalized completion verification layer that does not exist yet. The platform could stay technical forever, or it could grow into something broader. I do not know which path makes more sense.

closing thoughts and how long I think it will take

I think Fetch Quests could meaningfully change how technical work gets done. The core insight is that work should be clear, completion should be verifiable, and payment should be automatic. That is how fetch quests work in games, and there is no fundamental reason why technical work in the real world cannot work the same way. The infrastructure exists now. Smart contracts can hold escrow and release funds on verified completion. GitHub Actions can run tests and confirm that objectives are met. The pieces are already there. What is missing is a platform that puts them together in a way that actually serves workers instead of extracting value from them.

How long will it take to build? I think an MVP that handles the core workflow of quest creation, acceptance, completion, and payment could be built in a few months. The harder parts are the reputation system, the dispute resolution mechanism, the Community DAO governance, and integration with payment rails for USD-modality Farmers. A full version with all the features described in this post and the accompanying specification probably takes a year or more of focused work. I have been wrong about timelines before.

If you have read this far, you are probably the kind of person who would either want to use Fetch Quests or help build it. If you are interested in either, reach out. Also, if you think I am wrong about something, I want to hear that too. The FQP process is designed to evolve the platform based on real feedback, and that feedback does not have to wait until the platform is live.