Original author: Aiko, Researcher @ Folius Ventures

This is a typical Sunday for me, juggling two research projects simultaneously. One is a company research report, and the other is on AI. To find more inspiration and material, I spent about two hours browsing NetEase Treasure Room. I have some insights that I'd like to share, and since writing becomes more like an article when there's a lot of content, I decided to post it here.

This article will be divided into two parts:

1) Comparing web3 game marketplaces with Treasure Room and suggesting how game asset marketplaces should be.

2) Introducing the concept of "Metaverse Economy" and extending it to the field of AI + games.

Inspiration from Treasure Room for game marketplaces

I am particularly interested in transaction taxes, and even the previous company deck specifically explained "IAT" (In-app Taxation) as a new business model. However, I have never written about a game vertical marketplace, as I have never trusted the possibility of such a third-party created trading platform with only simple game asset aggregation functions in the past, present, or even the near future.

However, my research on Treasure Room once again confirms my previous thoughts: the means of centralized marketplaces and the pursuit of high-frequency and zero-fee decentralized transactions are fundamentally conflicting.

1) As a first-party self-operated platform, the value of all game assets in Treasure Room is supported by the underlying player DAU and transaction volume. Thus, the first step for a marketplace to have demand is to ensure that assets have a wide audience who is willing to pay a premium.

2) Customization features do not work well in the rapidly migrating web3 domain. Customized marketplaces require the accumulation of user data, player preferences, player demands, etc. It requires stable and long-term operation of games to make customized changes. However, web3 games driven by ROI may not be able to keep up with the speed of hotspot migration in terms of customization.

Treasure Room has a high degree of customization. Each game has its own market and special operations. The display interface is similar, but various lotteries/special events/appraisals/specials/servers/seasons are different. For a particular game, there will be a shopping assistant guiding you to choose a race-character-class-budget, step by step. Only projects that have been in operation for a long time can truly understand what their players need in order to customize effectively.

3) The platform's revenue and user experience have transitioned from inconvenience to convenience, from low turnover to high turnover. 0 transaction fees and high liquidity do not benefit the platform's revenue and stable asset prices.

Players familiar with NFT may understand how the emergence of Sudo Swap and Blur deeply impacted the NFT market in the past year. 0 transaction fees and batch trading have shaken the foundation of NFT as a luxury item/faithful belief, witnessing big players buying dozens of monkeys in a single stroke, causing some holders to question the meaning of community faith in such capital operations.

In our previous two decks, we talked about how to increase the slip by creating transaction friction. The treasury also has many similar designs, such as the Express Trading Service, which can reduce the originally required 4-day review time to 0 days and the secondary trading time from 8 days to 0 days. This service will charge sellers a 4% handling fee. In the treasury, both sellers listing and buyers purchasing require payment of fees. Another part of the service that facilitates buyers, such as the public notice reservation, allows buyers to subscribe to game assets in advance, and buyers need to pay a 5% booking fee based on the total price.

4) Designing game-like mechanisms that satisfy human nature is part of the delivery of a marketplace.

Many web3 NFT marketplaces focus on transaction forms and matching efficiency, and games are no exception. However, using crypto asset thinking to create a marketplace for game assets obviously ignores the strong demand for application scenarios inherent in the game itself, that is, users have a strong interest in participating in gamified selling activities.

In the treasury, some games arrange a bidding lottery system, where the chance of winning increases with the bid. It is also a game that allows users to play with numerical expectations. Presumably, due to regulation, this mechanism cannot exist openly in the game but can be naturally introduced into the trading market, thereby increasing the trading frequency of users and also improving the platform's profit margin.

In fact, there have long been suggestions for adding various gamification features to NFT trading markets, similar to the gamified features of domestic e-commerce platforms, such as Taobao Orchard or lotteries, etc. However, looking at it now, there are no major trading markets that have adopted them, except for Rollbit, which incorporates opening NFT boxes as a unique gameplay unit of crypto assets into its large casino. This precisely presents an opportunity for emerging gamified marketplaces.

5) If there is no demand, artificially increase the demand. The difference between the fifth point and the third point is that the third point can be achieved based solely on the platform's trading rules, but the fifth point emphasizes that it is a deliberate act. When there is a need, a group of people can be found to support a new trading scenario.

For example, there is an appraisal activity in the Treasure Pavilion. It is when the official invites a group of authoritative players in the game ecosystem to become appraisers. Paying the appraisers for quickly selling the appraised treasures. These appraisers can be seen as cost calculation experts, helping sellers assess asset prices and providing certification. It is presumed that after placing an order, the platform will also provide more exposure and preferential display. The fees paid by sellers to the appraisers will definitely be deducted by the platform. This money is similar to the advertising fee for content, and the platform has found a group of people called appraisers to serve as a buffer, making this fee appear more reasonable and gentle.

Before understanding "metaverse economy interaction," we need to establish a new "value measurement dimension" for game assets.

Firstly, it is the assets traded in the Treasure Pavilion that have triggered my thinking:

The commodities traded in the Treasure Pavilion are usually "big items," often involving the sale of an account, which contains multiple characters and character skins.

It can be observed that the assets traded in the Treasure Pavilion are no longer just a single item, but rather a segment of experience. For games with rich hierarchical experiences, a weapon or an equipment piece may enhance strength but cannot fundamentally change the game experience. Why do some players choose different characters to replay the same game over and over again? It's because the most valuable aspect lies in the characters and accounts that can fundamentally provide new experiences. Purchasing an account means being able to experience the progress in that account, which is like a life restart simulator, experiencing what the world of a pro player is like.

In the deck we released in February, the importance of personal progression was also mentioned. In April of this year, we happened to talk about a team that allows players to package the progression of game characters into NFTs for sale. The character's progression includes unlocking skills and skins. Unfortunately, it has been a long time since we've seen such an interesting design.

How can someone who is still addicted to asset units (such as 1155, 6551, etc.) know the key to it? It doesn't matter how you divide the assets or whether they can be packaged or whether the traded sword is worn or a grain of sand, these are not important; what is important is to identify the "datasets" that can generate value and recognition for players and select these "datasets" as assets to be put on the chain.

Furthermore, why are "large-scale" assets more important than "fragmented" assets? Because they contain multiple dimensions of value measurement.

Asset trading in games can also be seen as an experience. Many web3 games now emphasize the freedom of asset trading in the game. For example, popular ones are similar to certain minerals or military resources on a planet, etc., which I mentioned in a previous blog on game AMM design. Not to mention how many DAUs are needed to sustain the operation of such a trading system, it is possible to focus on this direction and lose sight of the bigger picture.

In the subscription era, we used time (or pow) as a measure and focused on the stability of in-game virtual currencies. For example, when we used to argue that the economic system of Dreaming Westward Journey was more stable than that of World of Warcraft, we talked about Dreaming using a point card system while Warcraft used a monthly card system, so Dreaming could better stabilize the value of every asset in the game, and also effectively control the inflation rate of gold coins (of course, there are many mechanisms worth learning from, I recommend everyone to read Frost's article, salute).

And now, there are not many open-ended economic games left, and there are not many people paying attention to the specific prices of each item, or it can be said that players who used to pay attention to the stability of item prices are completely different from players who "whale spent a few hundred thousand dollars". F2P and Gacha have become mainstream, so a single value dimension of time (pow) can no longer support transactions, but rather proof of capital + proof of luck + proof of work + proof of skill. In plain language, it means: I have money to gamble, I can still win, and I can cultivate and maintain this account with correct and efficient methods to achieve multidimensional value.

In this complex value measurement dimension, players can roughly calculate the cost of cultivating such an account, but bringing this account back into the game only results in basic resources and a period of gameplay experience. Even if they want to sell it, the price is basically equal to the original price + the value converted from the amount the new owner spent on it, and what players earn is saved cultivation time + a completely new experience, not any direct economic return. It is truly a very good commercialization method.

And, this is actually a value measurement dimension that gradually emerged after F2P. When the game becomes "no longer fair," when players can spend money to buy values, these values that include economic value, through randomness and strategy effects, are ultimately applied to accounts, forming a new asset that is more suitable for F2P players to trade. However, F2P Gocha made too much money, and everyone went to roll art and industrial production capacity. They never thought that the secondary market could be commercialized for a second time, and it could provide an outlet for the accumulated value of characters.

So, what is the interactive economy?

First, let's assume a scenario: which is better suited as an experimental field for AI strategic games, Rise of Nations or Civilization6? (Without considering API interfaces and implementation difficulty, etc.)

First, the conclusion: Rise of Nations is better.

The resources in Civilization 6 are obtained through the player's time and technology. In multiplayer mode, due to the limitations of LAN, participants are often real-life friends, and entertainment and time polishing are the main purposes. Therefore, there are many verbal negotiations/intimidations, such as "let me borrow this," "give me this" and other practices of exchanging resources with emotions. Secondly, the claimable diplomatic relations are limited and have a sense of distance. Countries can friendly ally with each other, but cannot involve more complex and deep class relations, such as submission and paying taxes.

And there are two main features in RateTu: 1) General Gacha System - A significant part of the military power and skills in RateTu depends on the generals, which are obtained through paid gacha. This creates strong economic interactions within the game. For example, giving red envelopes to region leaders every month can be automatically calculated as "the cost of drawing a general = the cost of paying wages to 5 counties." Therefore, there is more economic interaction, and frequently activities like sending red envelopes or paying salaries occur. 2) Alliance Relations - RateTu, as a multiplayer online game with thousands of players on the same server, has more intricate social designs compared to games like Civilization, which focus on single-player or local area network multiplayer. The seamless large map in RateTu requires joint military expeditions, and building roads and bridges among alliance allies is particularly important. Alliances provide resource allocation buffs, and the siege gameplay requires the cooperation of players with different economic strengths and social classes. Consequently, strong interdependencies between economy and territory are formed.

As we can see, player payments have a significant impact on the dynamics and overall atmosphere of a multiplayer game. Regardless of whether it is good or bad, it is a completely different experience.

Paying unlocks the channel connecting the game to reality, and a player's wealth and luck in the real world greatly influence the game. In addition, outside of in-game confrontations, there may be numerous economic exchanges and human games that are difficult to quantify. Since assets are purchased by players, in-game values can reflect a player's real-world economic strength. This economic strength can be used as bargaining chips to negotiate with other players, transferred to their real lives, or re-introduced into the game as values through other players.

Now, let's imagine a scenario that differentiates game interaction from meta-economic interaction.

Furthermore, what interesting use cases does meta-interaction have?

- AI agent meta-interaction

Now, let's consider which aspects can involve AI based on the value measurement standards we just defined.

• Capital (Optimizing capital utilization & out-of-game strategies)

• Time (Saving player time)

• Skill (In-game strategies)

Now, AI already has the ability to bid/negotiate, so if AI takes over the in-game interactions as well as interactions with other players outside the game in the entire economic interaction system; and each player has an agent, these agents can communicate with other agents and negotiate conditions/bargaining, wouldn't that be fun?

Imagine the following scenarios:

1) Each player has an initial agent with consistent game strategies and bargaining abilities;

2) Players need to allocate funds to the agent to carry out operations such as bribery/purchasing other agents;

3) The agent returns optional strategies to the player every 6 hours, and after selecting the optimal strategy, the player lets the agent execute it;

4) The agent can learn the player's risk preferences from each strategy selection, and optimize the next strategy (e.g. whether to continue bidding to buy other agents, whether to increase troops and attack cities with a certain amount of resource reserves/reinforcements, etc.);

5) Of course, a player can have more than one agent and independently train multiple agents to cooperate, such as a diplomatic agent and a tactical agent, which may result in a say agent that can handle foreign affairs of other countries without spending a penny on soldiers and resources.

6) At this point, the agent can become a "personality" with appearance, voice, and character, and the player who trains this agent can benefit from the commercialization of the agent.

So, this way, the original game experience with thousands of people playing together, 24 hours non-stop and heavily relying on socializing outside the game, can turn into a passive socializing (similar to single-player) experience, and the agent can also exist as a "mascot" similar to characters in 2D games, becoming the player's main interaction object, and even be commercialized as new game content.

Finally, why do I find AI games involving meta-economic interactions interesting?

Perhaps it's because I've been researching incentive mechanisms in the crypto field for too long, or maybe it's influenced by zkml's belief that future AI natives can manage users' assets for financial transactions, or maybe it's because I'm not interested in generative/passive forms of interaction between intelligent agents and humans after "Dwarf Fortress," or maybe it's because of my obsession with viewing games as interactive art, but I always feel that the current AI + game interactivity is far from enough.

It cannot be denied that almost all AI + games at present are just testing how well AI imitates human behavior. Dota is a training ground for AI micro-operations, Minecraft measures AI's understanding of the physical world, Stanford Town simulates AI's understanding of human behavior and emotions. So, why hasn't there been an AI simulation of human use of economic means placed in a game experiment yet?

No matter the era, games need to cater to human nature. Can observing AI's daily behavior be considered a satisfaction of voyeuristic desires? How long can this kind of satisfaction last, and how entertaining is it? Can it compare to the visual and auditory experiences brought by graphics cards and VR? Can it compare to the adrenaline and dopamine released when opening a loot box? Can it compare to the satisfaction of vanity when looking down on all beings in the entire server? If not, how can we use AI to create games that better cater to human nature and stimulate their pleasure points? That's what I find interesting.

Furthermore, I have also been thinking about and studying the role of AI in narrative games. These are some of my reflections and insights after completing "The Wheel of the Cosmos Sisterhood" this week. I believe storytelling is one of the aspects where AI can greatly enhance the experience/differential experience.

Conclusion

In writing this article, I actually have three purposes:

1) Explore the interesting mechanisms in the Treasure Pavilion that could be borrowed by teams designing marketplaces. Also highlight the differences in feasibility and potential between the Treasure Pavilion and web3 gaming marketplaces.

2) Explain the new dimensions for measuring the value of game assets. With the development of F2P, players' entertainment and consumption habits have undergone significant changes. Instead of sticking to the logic of subscription-based gaming transactions, it would be better to shift perspective and consider other forms of asset transactions, identifying the main value carriers and transaction forms, such as trading a segment of a gaming experience (including multiple valuable character accounts).

3) If you are interested in the intersection of "metaverse economy interaction" AI and gaming, and are also engaged in AI and game-related research, please feel free to contact me. We have more research to do.

Original article link