The Web 10 Vision
The Web 10 Vision
An advanced, societally beneficial future of the internet, software, and knowledge.
The Present and Future
It’s 2023. Your alarm interrupts your slumber and you groggily roll out of bed and check your devices: a $1,000 smartphone and a $2,000 laptop that need to be thrown out and replaced every few years. You need to write a report on historical trends to increase corporate profits. You manually review information scattered everywhere: files, documents, spreadsheets, websites, paid subscription apps. It takes the whole day to do the tedious process of finding all of the information and drafting the report. After work, you want to watch a few video tutorials to learn a new skill. Instead, your social media app uses eye-grabbing recommendations and infinite scrolling to show you hours of distracting videos. It makes $1 in ad revenue by wasting hours of your time. Before you know it, the day is over.
It’s 2025. You wake up naturally and open up your laptop, a sleek $300 device with access to limitless speed and storage so it never needs to be replaced. Your team needs information to enhance a product that improves people’s lives. You spend a minute describing it, and your laptop pulls up the info in a few seconds. What used to take hours now takes minutes! You decide to spend the rest of your day doing creative work on the product. You explore a few design approaches, using AI to quickly visualize what your design ideas look like and validate the feasibility of certain approaches. After a great day at work, you want to learn a new skill. Your laptop finds the videos that can best help you with that based on what you already know and the types of videos you’ve enjoyed and found valuable in the past. You’re able to learn the basics in a few hours, and you spend the rest of your day trying it out with your family.
This vision for 2025 is achievable with present-day technology, and it’s coming soon. Let me tell you all about it.
Historical Trends
We need to look at historical trends to see where the future will be headed. It’s been a significant last few years for tech. There are three major tech trends I’d like to highlight: artificial intelligence, decentralization, and all-in-one apps.
On the AI front, we’re on a cusp of having capable artificial intelligence that can do a significant amount of human knowledge work. The release of ChatGPT (a large language model, or LLM for short) in November 2022 demonstrated this to the world by being capable of writing some usable (and a lot of near-usable) code, essays, emails, and more. I think it’s likely that capable AI will be popularized later this year with AI models that can base their thoughts on real-world information and take real-world actions.
On the decentralization front, Web3 has been very trendy for the past few years, albeit not so much this past year. Web3 gained popularity in part due to dissatisfaction with the current state of the internet. It popularized a lot of interesting and compelling concepts like shared ownership of the web and trustless coordination made possible by its version of decentralization technology. Despite the fall in crypto market values, decentralized autonomous organizations (DAOs), an aspect of Web3, still control billions of dollars in funding and are governed by millions of people.
And in the realm of traditional software, a new type of app has been emerging. Within the last 10 years, the first generation of multi-purpose “all-in-one SaaS'' apps like Notion, ClickUp, and Airtable were founded. These apps can handle multiple types of data—everything from tasks to information about products. This was so compelling to businesses that these apps all became unicorns within several years, with Notion and Airtable both being valued at $10 billion+.
Future Trends
These major trends—AI, decentralization, and all-in-one apps—are being held back from their true potential. The key to unlocking those barriers can give us a more nuanced view of the future.
Right now, LLMs are generating text based on historical training data. This produces output that is somewhat useful, but not that usable. In order for LLMs to use real-world information and take real-world actions, they need to be able to access that information and take those actions. Much of the information and actions needed to get practical work done aren’t on the internet (like your current state of knowledge about various skills), and they’re not machine understandable (they’re in text, which is human readable, rather than in a database, which is machine readable). A new app that can ingest private and public information in a machine-understandable way will be required to fully take advantage of the power of LLMs, and AI in general.
Web3 highlights the need for a better version of the internet, one that is built to increase trust, give people more ownership and control, and enable humanity to benefit from public information and public goods. There is massive interest in Web3, as well as associated movements like responsible tech. However, Web3 has major issues, including fundamentally higher time and monetary costs for developing and operating Web3 services, and low user accessibility and adoption. This will significantly hinder the impact of Web3. Alternative movements also have major issues, namely that they are not able to effectively influence tech companies in the Web2 (traditional internet) space, which continue to be focused on maximizing profit (as companies are legally required to do), often at the expense of users and society. A new type of technology, and a new type of organization, will be required to achieve the societal goals of Web3 and responsible tech in a viable manner.
All-in-one SaaS apps are indicative of the demand for the centralization of data and apps, but they have a problem—none of the currently available solutions on the market can represent the full extent of knowledge and workflows that people and organizations work with. Therefore, they can’t be true “all-in-one” apps, and now they’re just new apps that many people and businesses need to learn to work with in addition to all of their specialized apps which they can’t get rid of!
The Future
In order to overcome the obstacles of centralizing information and making it machine understandable, as well as aligning technology with human values, I believe we need to create a new type of architecture for technology, which can be described as:
A centralized app for all computing, which is “better, faster, and cheaper” compared to both the traditional internet and Web3
…that can represent any type of knowledge, which will empower people and organizations to enhance and automate most of their knowledge work
…which is governed by an organization that is legally required to act in the best interests of its users and society, which is nearly unheard of in the corporate world and in DAOs.
I’m going to name this vision Web 10.
“10” denotes a feeling of advancement and permanence, much like Windows 10, iPhone X, and so forth. This is aligned with Web 10’s design; it’s a future-proof concept for what the web would look like in an advanced society, not dissimilar from planetary computers in sci-fi. In a world where we can trust, there is no need for decentralization. In a world of efficiency, there is no need for duplication and fragmentation. In a world aligned with our values, the web is optimized for societal well-being, not for special interests.
The Core Design of Web 10
How did I arrive at this design for Web 10? I’ll go into the details of why these design principles are the future and how they work.
Centralizing Data: Documents, Spreadsheets, Databases, and More
The key to Web 10 is data. Data powers apps, AI, and increasingly organizational and societal decision making. The first step in enabling data to be centralized and used for the public good is the ability to represent it. Web 10 will have a simple model that is able to represent all information: it will have types of objects with particular properties that interact in particular ways with other objects. This model can represent anything from a spreadsheet of financial transactions (each transaction is an object with properties), to someone’s Facebook friends (two people are represented as objects that are friends with each other), to how atoms and celestial bodies interact in the cosmos (everything in our universe is a type of object with particular properties and allowable interactions).
This concept has been in development for decades. Its general name is “semantics.” In the early 21st century, the “Semantic Web” was imagined to be the future of the internet. It was clearly better than the current form of the internet, and endorsed by many, but failed because it needed to get large numbers of people and companies to change how they operated (adding organized data and relationships to all websites on the internet) at little benefit to themselves. However, using semantic data with technology is a thriving industry. It’s the present and future of business data and intelligence (a16z certainly thinks so, and people are nerding out over semantic advancements in BI tools), and it’s flourishing in centralized havens of the internet like the Google Knowledge Graph.
These concepts are beginning to make their way to all-in-one apps. They’re responsible for the success of the next-generation of such apps like Roam Research (one of the fastest startups to $1 million in annual revenue) and Tana (the latest hot app in the knowledge management space). These apps each have a part of the puzzle (Roam has generic bidirectional relationships and Tana has metadata), but far from the entire set. Apps like Anytype are more advanced when it comes to semantic features. Web 10 will finally deliver on a fully functional semantic model—it contains cutting-edge features that are not yet even in the database systems that powered Roam Research’s rise, like combining object relationships with object types—features that are required for any system to be able to model all of the data in existence.
With this model, Web10 will be able to represent the same data contained in any format of knowledge, from files on your computer, to documents and spreadsheets, to information contained in database systems like relational and graph databases. This model, combined with the proven strategy of implementing semantic data in centralized apps instead of decentralized ways, will be the reason why Web 10 will become the future of the internet.
Centralizing Software and Software Builders
Just like we’re a stone’s throw away from centralizing all data, we’re also a stone’s throw away from centralizing software creation. All-in-one apps are part of the growing nocode movement, an industry worth tens of billions of dollars that allows users to build apps like Instagram, Yelp, and Duolingo without any code.
There’s just one thing missing—proper semantic data. Without it, users need to design custom visual components, data structures, data integrations, etc. from scratch. Bubble, one of the leading nocode app builders, lists all of these out as requirements in its documentation. Unfortunately, this makes it so that building apps with existing nocode tools still requires expertise and many steps, so it’s more like a 10x reduction in development time (years to months) rather than a 10000x reduction (months to minutes).
Web 10 will make it possible to build apps in minutes (or auto-generated ones in seconds), rather than weeks or months. Semantic data is data with a particular meaning and purpose. It can be connected with other pre-built (or AI generated) parts of an app with a similar meaning or purpose, like visual components (e.g. semantic blocks). For example, if you have information about restaurants and want to build an app like Yelp, you can pair your information like the address of a restaurant and the restaurant’s menu with a semantic block for a map (which might require two addresses to display a route) and semantic blocks for food items (which might require a picture, title, and ingredients, and automatically display information for allergies and dietary restrictions based on the ingredients).
Semantic blocks, workflows, etc. will make it possible for users to create their own apps for any purpose. In addition, users can create their own personalized interfaces on top of any existing data or app in a Web3-esque manner. This will free us from the monopolies and oligopolies that have emerged around particular types of apps, like Yelp for restaurants, Goodreads for books, and yes, Facebook for friends. Users can store their data in one place and use a single app to get things done, not juggle hundreds of apps for slightly different purposes. Since all apps draw on the same data, users won’t need to enter duplicate data in multiple apps, be unable to integrate certain apps with each other, or have data stuck in one app that cannot be exported or used for other purposes. Users can customize apps exactly the way they like them, increasing productivity and eliminating the need to relearn continual UI changes in every app. People, and especially organizations, will collectively save billions of dollars that were previously spent on specialized, single-purpose apps like CRMs and ATSs.
Making Data Public
Moving to the collective benefits of Web 10, in a manner that’s somewhat similar to Web3 and the Semantic Web, Web 10 will create enormous value by making important information public and free to use for all.
The value Web 10 provides for users is greatly magnified when data is public. Therefore, Web 10 as a whole has an incentive to make as much important, non-harmful data public as possible, an incentive that does not exist for the traditional internet. The internet today is decentralized, with any person or organization able to launch a website and monetize that website, and no centralized system able to gain resources from the overall value the internet provides. Thus, even though sharing public data about restaurants or books would be of great value, internet companies like Yelp and Goodreads have every incentive not to. Web 10 can monetize from access to public data as a whole (either by charging money or using non-problematic monetization methods like ethical ads that benefit users and respect user privacy and preferences), and leverage that funding to fund the creation of more public data. It can also track how useful each piece of data is, for instance by measuring how often it is used by users, and offer market-rate compensation for users to add additional data to Web 10 (The Graph is an example of this model in Web3).
Secondarily, the upside of keeping data and apps private is not very high. If anyone else can create a similar app in a few minutes, or add similar data to Web 10, the value add of keeping data and apps private becomes so low it doesn’t make sense to do. Similarly to how companies are incentivized to contribute to open source software that serves their internal needs, it makes more sense for people and organizations to collaborate on making data public, rather than risk spending lots of resources to create a data store or app that ultimately becomes worthless when another user or coalition creates a public version of that. The impulse for altruism, recognition, learning, etc. are additional factors that will spur users to contribute data to Web 10, similar to the reasons behind why people contribute to Wikipedia and open source software.
Additional Designs for Web 10
While replacing the internet, including all data and software, is the biggest facet of Web 10, there are many additional ways that Web 10 can transform society. Here are a few ways. I believe it’s critical to ensure that whatever organization introduces Web 10 moves in this direction.
Powerful, Fast, and Cheap Computing
Web 10 involves the creation of a centralized “app” for all data and software. While it will be possible to store data and run programs locally on one’s own computer, there is no need to do so. Indeed, it would be massively more efficient to have everything done in the cloud and stream information down to devices as needed. What used to require $1,000 phones and $2,000 laptops will only require $100–$300 devices that don’t need to be thrown out every few years.
If a user shares data with another user, instead of (1) downloading a picture from iCloud (2) onto their iPhone and then (3) sharing the full picture with someone else via AirDrop and (4) having the recipient store their own duplicate copy of that information on their iPhone that may be backed up to (5) their own iCloud and possibly (6) on an external drive, Web 10 can simply update the access permissions for the picture in the cloud so that a new user can access it instantly. Storage will become limitless, cheap (especially with techniques like cross-user data deduplication), and more reliable.
Software will become much more efficient as well. People can access a limitless number and variety of cloud processors to run software, which are available for anyone to “rent,” rather than sitting idle most of the time. This will radically increase the speed of computer usage, since cloud processors can be much more powerful than the processors on consumer devices, and greatly decrease cost.
This model is similar to the concept of “thin clients,” which is what early computing pioneers like Douglas Engelbart envisioned everyone would use. The popularity of web apps has made this model more appealing using present-day technology, with the emergence and rapid growth of Chromebooks a testament to its viability. Chromebooks are much cheaper than traditional computers and faster for certain use cases. However, since they operate with older technological paradigms, they don’t work with or can struggle to meet the performance demands of certain apps and websites. With Web 10, all apps and data are cloud based, so all processing and storage can also be fully cloud based. Users will not experience any downsides of using a cloud-based model, only the upsides of limitless speed and storage. The explosive success of cloud gaming, in which all game processing is performed in the cloud and streamed down in a manner that requires minimal processing power, is a closer example of this approach.
Augmenting Individual and Collective Knowledge and Thought
A key vision of early computer pioneers was to have computers augment human capabilities. A key vision of the Semantic Web was enabling computers to act as “intelligent agents” and automate a significant fraction of knowledge work for humanity. Web 10’s knowledge structure makes it possible to realize this vision. Semantic data will let humans share information with computers/AI that they can understand. This will enable computers and AI to help people do a significantly higher fraction of knowledge work while requiring vastly less human effort. For example, in order to get an AI to draft an email for you, it would benefit from knowing everything about yourself, the person you are emailing, the situation, etc. Instead of feeding this into an existing system like ChatGPT by typing several paragraphs of background information, an AI system connected to semantic data could learn all of this information instantly. It is also possible to automatically train AI on any use case, no matter how niche, if the data is stored semantically. This would otherwise require significant effort to organize, format, label, etc. the data so it can be used for training.
Semantic data will also make it possible to significantly increase humanity’s collective intelligence, which is humanity’s ability to solve any problem whether that’s "developing a new product or service... seeking a cure for cancer, or improving conditions in underserved communities." Right now, we don’t really think collectively beyond polls, and we only store knowledge collectively in unstructured forms like Wikipedia. Imagine if we could structure thoughts themselves and see everyone’s reasoning, beliefs, sources, estimated likelihoods, etc. on an individual basis as well as in an aggregated manner, like a highly enhanced version of argument mapping. Now imagine if algorithms and AI can further help with that process. Web 10 will revolutionize how collective knowledge and thought works, hopefully revolutionizing humanity’s ability to address the greatest challenges of our era.
Social Impact and Better Governance
Unfortunately, mainstream implementations of organizations that can sustainably scale their revenue in service of human values are vanishingly rare. Companies, of course, are required by law to maximize profits, not improve lives. Public benefit corporations (PBCs) are a significant improvement to corporations in that social benefit activities seem permissible, but the execution of those activities is highly uncertain, given issues like the fact that only shareholders can sue to enforce a PBC’s mission and there is no relevant case law for that yet. Nonprofits have a mandated social mission, but given the lack of financial upside, they often cannot effectively raise capital for impactful activities. Decentralized autonomous organizations (DAOs) are perhaps the best bet for alternative governance. They can raise money as well as enshrine alternative governance methods like futarchy in code. Unfortunately, most DAOs are governed with tokens that bear a striking resemblance to how shareholders control corporations, rather than being at the forefront of socially impactful governance innovation.
While DAOs are interesting, they don’t eliminate the relevance or implications of real-world laws. Initial coin offerings are now regulated in the US and other countries, and DAOs have a limited ability to interface with the real world without a “legal wrapper” which makes them very similar to regular companies once a wrapper is created. Fortunately, it is also possible to use traditional legal structures to enable alternative governance structures. Movements like Zebas Unite are harnessing traditional corporate structures to enable people to create organizations with shared ownership and control, as well as socially beneficial goals beyond profit maximization. I believe that democratizing organizational control using traditional legal structures, such as by assigning voting shares of a company to a charity but not necessarily ownership shares, is one of the best ways to ensure an organization can pursue social impact activities while still being able to raise for-profit capital. This is the type of organization that I hope launches Web 10 in order to gain the benefits of for-profit fundraising (e.g. venture capital, equity crowdfunding, etc.) while ensuring Web 10 is required to maximize human values, not profit. If there is competition among Web 10 organizations, I hope users will choose the version of Web 10 that is in their best interests.
Regardless of whether a DAO or hybrid organizational structure is used to enforce a governance structure, perhaps the larger challenge is how to best govern an organization in the first place. Web3 provides the ability to encode alternative governance methods, but it has not yet unearthed a governance structure that seems better than traditional voting. Web 10 will enable significant improvements in collective intelligence which will hopefully unlock the next advancement of governance beyond representative democracy, which is currently in use in most companies and countries. My hope is that Web 10 transitions to a collectively intelligent governance scheme once it becomes known.
Conclusion
Thanks for reading! I’m excited to hear your thoughts about Web 10 and the future. I’m developing an app called Cosmic that’s the first full-scope implementation of Web 10. To stay tuned, sign up for this newsletter! If you’re interested in funding Cosmic, working on Cosmic, using Cosmic, or anything else, feel free to get in touch at team@cosmickb.com.