Crypto-Collective Knowledge Management

🤝

The term collective knowledge management (CKM) is derived from personal knowledge management (PKM). PKM tools assist with the use and organization of personally-relevant knowledge in digital form, including user interface affordances such as outlining, links, embedding, and search. CKM is the idea that such tools are also useful for managing knowledge relevant to multiple people. Although the term hasn’t been used widely until recently, many examples of CKM exist: Google search allows query-based retrieval of any public text on the web, Google Docs is the foremost example of a real-time collaborative writing tool, and Wikipedia orchestrates the writing of many contributors into canonical topic-based articles.

What we have versus what we need

Google search and Wikipedia in particular are great examples of how the internet has enabled immense scale and speed: every crawlable page on the web, every sufficiently noteworthy topic, etc, returned to you within a second. But choosing the extreme end of scale is at odds with the key property of personal knowledge management: relevance. A scientist or other knowledge worker--we’ll use the general term researcher--who wants to know what knowledge is emerging at the moment will not find the answer on Wikipedia or at the top of Google.

Today we’re seeing the cracks of scale and speed. In the continuing crisis of COVID-19, dissenting opinions from the state authorities, CDC and WHO, are shut down because misinformation can spread rapidly and be deadly. But the strategy of tech companies to defer to these authorities is backfiring as large swathes of the public increasingly mistrust the agencies, and those working on the frontlines of the conversation are denied access to the full range of information.

The standard answer for researchers is conferences and journals. However, the scale and latency (for ~1 publication per researcher per journal per year) of this answer is disappointing in the internet era. Instead, researchers increasingly turn toward Twitter and pre-publication archives to get up-to-speed information.

Twitter has many drawbacks. It isn’t good for readers to build knowledge: tweets aren’t organized and the content is difficult to work with. Generally people copy Twitter content somewhere else to work with it. It also isn’t good for writers to build their reputation: tweets are listed chronologically, interleaving personal and professional, rather than presenting any indication of your most knowledgeable tweeting.

Pre-publication archives make it difficult to figure out the most relevant information. In just one category, machine learning, arxiv.org lists 101 entries uploaded on a single day.

Three approaches to crypto-assets for knowledge

Block-based

Block-based knowledge management has been popularized by Roam Research and similar software in the last couple years. A block is a paragraph of text. This text may include special codes for markup or embedded data. In Roam, each block has a unique id such that it can be accessed, referenced, or embedded throughout a Roam graph. (And, presumably, across Roam graphs in the future.)

For block-based CCKM, we introduce RefGold. The core idea is simple: the reuse of a block pays a microtransaction to the block owner and records a transaction of its use on a public ledger. The term “ref” comes from Roam, where the reuse of a block is called a reference and the original block maintains a back-reference to the reuse (AKA backlink).

If a block is typically just text, why would anyone pay RefGold instead of just copying the text? A few reasons: 1) voluntarily, as a small tip to appreciate the block owner, 2) because the reference is recorded, the block owner may choose to share backlinks to public references, so this serves as an advertisement for the payer, 3) new types of journals may require that cited information use RefGold, and they may pay the RefGold bill for accepted articles.

Page-based

Wikipedia is the best known example of a page-based approach to knowledge management. Each topic has a dedicated title and page, accessible on English Wikipedia at en.wikipedia.org/{title}. A given topic has, at any given moment in time, a consensus about what is important and truthful. The barrier to truth is established by written norms--for example, Identifying reliable sources (medicine).

Personal knowledge management also typically works with pages. (In Roam this is just a special kind of block.)

For page-based CCKM, we introduce TokenPage, a page-based tokenized knowledge service. Imagine that we have the domain token.page. The owner of a page has rights to set the content of that page at token.page/{title} (in accordance with the overall governance of TokenPage, which would enforce rules like legality of content). A simple implementation of this model is page ownership corresponds to an NFT, which also allows ownership trading.

The benefit of Wikipedia’s page-based organization is that there’s little ambiguity about where to find information about a topic. Likewise, TokenPage offers owners a strong power: determining the canonical knowledge of a topic.

The owner can grow the intrinsic value of a page by making it contain broad and accurate knowledge. The page also gives value to the owner because their name is featured prominently on the page, and each user has a list of owned pages, which serves as a portfolio of their knowledge management expertise.

A major drawback of TokenPage is that a user can attack a page just by spending enough money. For example, a user could own the “Yankees” page and simply write “The Yankees suck”. Despite the low quality content, the page may retain its value to the malignant owner because others, like Yankees fans, would be willing to pay to “fix” the page.

Community-based

I’ll restate the problem with TokenPage: there is a tension between groups such as Yankees fans, Red Sox fans (AKA Yankees haters), and those who just want to maintain accurate knowledge, which bids up the price of a TokenPage orthogonally to the page’s intrinsic value, and may lead to wild swings in the page content when ownership changes hands. This does seem fun, to be honest. On the other hand, stability seems like a core attribute of a system that people will use for building knowledge. As mentioned, Wikipedia has achieved something like this by maintaining strong norms, but it leaves out emerging knowledge that hasn’t yet met that criteria.

Another approach would be to let a thousand communities bloom and govern their own idea of accurate knowledge. Hence our third and final CCKM approach, TribePages. In the world of TribePages, we’ve accepted that different communities may have differences in accepted knowledge about a topic.

Suppose now that we use the domain tribe.page. Each community can have a subdomain, such as yankees.tribe.page. Now the same idea of TokenPage applies where a given topic has the community-approved knowledge for that page at {community}.tribe.page/{title}.

Additional affordances are added at the user level. A user can subscribe to many communities. Then, when browsing tribe.page/{title}, they would see a result among any subscribed communities. If communities have conflicting page content, it can show a diff view. Likewise, a user could specifically seek the diffs between two communities to compare their views on a topic.

Community subdomains can be governed as a DAO. The DAO can receive money by membership fees, RefGold, or by using TokenPage-like NFTs to award ownership of each page to an individual user.

Who would create such a community? Yankees fans are an example with a clear demarcation: what baseball team do you like? However, other communities may be more nebulous. For example, a task force may form to understand the pros and cons of a COVID vaccine booster shot. If this community faces an impasse on their foundational knowledge assumptions, they may choose to fork into new communities. (Pages of text with unchanging assumptions are easy enough to copy over.)

Next steps

I know very little about crypto, and I have an unrelated full-time job, so I’m not the best person to make these things happen. Please reach out to @cicatriz on Twitter if you’d like to discuss more.

Acknowledgements

This is mostly just patching together ideas from others:

Balaji tweeted about bidirectional links eventually being on-chain.
Jessie (JESSCATE93@) has been writing about the Roam/Crypto space for a very long time.
Emad (EMostaque@) has a system for rewarding climate change knowledge contributors.
Uberfact