Philosophy, information science, law, security, cryptography, formal methods, math, systems. he/him/his 🧩
Baltimore, MD
Joined August 2021
- Tweets10.3K
- Following6.8K
- Followers768
- Likes61.9K
Pinned Tweet
"The creation of something new is not accomplished by the intellect but by the play instinct." - Carl Jung
Ned Whitman retweeted
“The core algorithm was found on October 16 by an internal Claude research model with 14M output tokens and no human input. Anthropic researchers worked with Claude to improve the presentation, resulting in this paper.”
github.com/anthropics/formal…
Ned Whitman retweeted
A good day for democracy, with big wins in Serbia and Brazil. Israel could be next on Tuesday, then the US midterms the following week. The best case against the autocrats is the autocrats (and how they govern with corrupt self-promotion). trib.al/l5x6pNO
Hyperlanguage takes this to another level. You can have typed object described with English glyphnumerics and controlled in a language using Chinese gylphnumerics and vice-versa. This is the shared authorization semantics that gives rise to internet infrastructure and making it possible for there to be one “whole” for the world.
Formal Meaning is a language with a context-free grammar for describing data, and a control language for governing what can be done with it.
It binds descriptive predicates to authorization predicates. For example:
“This image is intimate.”
→ “This image cannot be shared without valid consent from the people depicted.”
A policy connects these statements: classifying the image as intimate makes consent a required condition of sharing. A system enforcing that policy must verify evidence of consent covering the recipient and intended use before permitting disclosure.
Meaning becomes operational. The same formal structure describes information and expresses the conditions, obligations, and proofs governing its use.
Formal Meaning is a language with a context-free grammar for describing data, and a control language for governing what can be done with it.
It binds descriptive predicates to authorization predicates. For example:
“This image is intimate.”
→ “This image cannot be shared without valid consent from the people depicted.”
A policy connects these statements: classifying the image as intimate makes consent a required condition of sharing. A system enforcing that policy must verify evidence of consent covering the recipient and intended use before permitting disclosure.
Meaning becomes operational. The same formal structure describes information and expresses the conditions, obligations, and proofs governing its use.
I’m developing Formal Meaning as a hyperlanguage with one structural grammar, versioned concept IDs, and interchangeable Hebrew, Chinese, and Latin glyphnumeric tokens.
Within the shared vocabulary:
ה101 ≡ 许101 ≡ G101 → issue a grant.
Each spelling resolves to the same typed meaning object. That object can then generate natural-language descriptions using each language’s own grammar.
Example:
English: “The data steward grants Noa permission to read the data, provided her identity is verified for this request.”
עברית: ״נאמן המאגר מעניק לנועה הרשאה לקרוא את הנתונים, בתנאי שזהותה אומתה לצורך בקשה זו.״
中文：“数据管理方授权诺娅读取数据，前提是她的身份已针对本次请求通过验证。”
You can mix token families within an expression while preserving its formal structure. The natural-language descriptions make that structure readable across languages.
For authorization, the system checks the issuer’s authority, the requested operation, and its conditions against the canonical meaning object.
The goal: preserve exactly who authorizes whom to do what with which data, under which conditions—across languages and machines.
AOC! If the base can’t see her promise as a presidential candidate when it’s time, it’d be a loss. She has mass appeal.
Rep. Alexandria Ocasio-Cortez, aka AOC, has softened her tactics and pursued voters and lawmakers in both parties. To reach her biggest goals, she'll need the far-left base—and everybody else. on.wsj.com/4zWD0Xe
Sex and power meet wherever society decides what kinds of persons, relationships, families, desires, and futures are legitimate. Sexuality is not merely private behavior; it is a field through which institutions classify populations, distribute status, organize kinship, regulate reproduction, define normality, and produce identities. Religion moralized it, law codified it, medicine diagnosed it, markets commodified it, and digital systems now render it computable. Power does not only repress sex—it produces categories, norms, incentives, and social realities around it. The political question is therefore not simply who controls bodies, but who has the authority to define persons and organize the social world through sexuality.
The crowning treatise on sex and power needs to be understood, it’s the golden thread. For those who “know”, it brings great shame, but the only way is through.
I’m developing Formal Meaning as a hyperlanguage with one structural grammar, versioned concept IDs, and interchangeable Hebrew, Chinese, and Latin glyphnumeric tokens.
Within the shared vocabulary:
ה101 ≡ 许101 ≡ G101 → issue a grant.
Each spelling resolves to the same typed meaning object. That object can then generate natural-language descriptions using each language’s own grammar.
Example:
English: “The data steward grants Noa permission to read the data, provided her identity is verified for this request.”
עברית: ״נאמן המאגר מעניק לנועה הרשאה לקרוא את הנתונים, בתנאי שזהותה אומתה לצורך בקשה זו.״
中文：“数据管理方授权诺娅读取数据，前提是她的身份已针对本次请求通过验证。”
You can mix token families within an expression while preserving its formal structure. The natural-language descriptions make that structure readable across languages.
For authorization, the system checks the issuer’s authority, the requested operation, and its conditions against the canonical meaning object.
The goal: preserve exactly who authorizes whom to do what with which data, under which conditions—across languages and machines.
Hebrew glyphnumeric form
ה101{
א201: @a1;
פ202: @p1;
ע203: ק301{מ205: @d5};
ת204: [ז401{פ202: @p1}]
}
Chinese glyphnumeric form
许101{
发201: @a1;
主202: @p1;
操203: 读301{对205: @d5};
条204: [证401{主202: @p1}]
}
English glyphnumeric form
G101{
I201: @a1;
A202: @p1;
X203: R301{O205: @d5};
C204: [V401{A202: @p1}]
}
Mixed form, with reordered fields
ה101{
C204: [证401{A202: @p1}];
发201: @a1;
A202: @p1;
ע203: 读301{O205: @d5}
}
This just scratches the surface of what data brokerages can and should become
Data brokerages could become institutions of collective power in the information economy: organizations through which people negotiate the uses of their contributions, establish conditions for access, verify what happens, and share in the resulting value.
Participants would delegate limited authority to a brokerage while retaining rights over how it operates: its mandate, management, fees, distribution rules, and participation in larger networks. The brokerage could represent creators, researchers, curators, workers, communities, or organizations contributing useful information.
The economic questions begin with the properties of that information. Where did it originate? Who contributed to it? What permissions apply? How was it transformed? Which contributions improved its usefulness? What obligations follow when it becomes part of a dataset, model, analysis, or other derivative product?
Originality, provenance, usefulness, and entitlement to payment would be separate claims, each requiring appropriate evidence. A recorded timestamp can help establish when a contribution entered the system. Authorship, novelty, lawful permissions, and economic value require further evaluation. Value also depends on the task: a contribution can be essential in one context and readily replaceable in another.
Derivatives make this especially interesting. A dataset may combine original records, annotations, corrections, and successive transformations. A model may depend on contributions from many organizations. A provenance system could preserve those recorded relationships and connect them to explicit agreements about attribution, permitted use, and compensation. Those agreements would define which obligations continue downstream, their duration, and how they share a finite revenue budget. Recorded ancestry alone would not establish perpetual ownership over everything later influenced by it.
Security by construction would give these arrangements an operational foundation. Within a defined execution environment and security model, valid authorization could become a precondition for computation. An accepted operation would carry evidence connecting its inputs, permissions, transformation, and resulting obligations. Participants and auditors could inspect that evidence when checking compliance or resolving a payment dispute.
The ambition is to make rights usable throughout the system. A brokerage authorizing an evaluation should be able to specify that scope, enforce it within the controlled workflow, and obtain evidence of the covered operation. Information already disclosed outside that boundary still requires contractual and institutional protections.
Tokenomics would organize the economic relationships: payments, contribution claims, shared infrastructure funding, and settlement between brokerages. Access credentials, membership votes, revenue entitlements, and payment instruments have different functions and should have distinct rules. Buying a financial claim should not automatically confer authority to expand the use of another participant’s information.
Rewards should support useful contributions and stewardship: creation, verification, curation, maintenance, expert evaluation, and secure computation. Paying primarily for collection volume would encourage more collection. Sustainable value must come from useful services and funded demand; issuing tokens cannot establish either.
Collective bargaining gives participants a practical way to exercise their rights. A brokerage can pool technical expertise, negotiate conditions, fund audits, challenge underpayment, and refuse unacceptable offers. Members also need powers to inspect their representative, contest decisions, replace leadership, and leave. Shared benefits could include income, better tools, research, and resources for future work.
Music streaming could run on a federation of rights brokerages: labels and artist cooperatives license their catalogs to competing music apps, while secure playback records support transparent royalty accounting.
Apps could compete on discovery, recommendations, sound quality, community, and privacy. A shared licensing and settlement network would let participating catalogs reach listeners through many different services, with consistent records of permissions and payment obligations.
A subscriber’s funded royalty budget could be allocated according to qualifying playback time. Each recording’s proceeds would then be divided under the applicable agreements. Tokens could record bounded claims on those proceeds and coordinate settlement across brokerages. More playback would redistribute the available budget; sustainable income would still depend on paying listeners and useful services.
Artist empowerment would need to be built into the contracts and governance.
Artists should be able to inspect which uses generated revenue, how their shares were calculated, what fees were deducted, and how advances were recouped. They should have practical ways to dispute underpayment, obtain independent audits, and change representation under clear exit and portability terms.
They should also retain negotiated authority over how their work is licensed. Streaming, sampling, remixing, advertising, and model training would have separately defined permissions. A brokerage’s authority would be limited to the rights it actually controls or has been authorized to administer.
Independent musicians could organize their own cooperatives and participate in the same infrastructure. By pooling catalog administration, negotiation, technical expertise, and enforcement, they could gain capabilities that would be expensive to maintain individually. Members could choose to fund shared promotion, production tools, or legal support alongside individual payouts.
Provenance would connect recordings, contributors, and authorized derivatives to their supporting evidence and agreements. A licensed remix could carry explicit obligations to the relevant contributors, making those shares easier to account for as the remix circulates across participating apps. Recording ancestry would support those claims; authorship, permissions, and payment entitlements would still require appropriate evidence.
Security by construction would aim to make valid authorization a condition of covered playback and preserve evidence for settlement. Those records could establish specified playback events without pretending to prove human attention. Fraud controls and dispute procedures would remain necessary.
Listening histories would also need protection. Artists and labels should receive the evidence required to verify their income, with controlled disclosure of sensitive information. Accountability should not require publishing everyone’s musical life.
Governance rights, access permissions, and financial claims would have distinct rules. Buying a revenue token should not automatically give someone power to broaden the use of an artist’s work. Artist representation and enforceable rights would determine whether a label-operated brokerage serves its contributors well.
The opportunity is a music economy where artists have stronger representation and inspectable accounts, listeners have more choice, and developers can build new experiences around interoperable catalogs. Labels could continue financing and promoting music while earning through agreements whose operation artists can understand and challenge.
Data brokerages could become institutions of collective power in the information economy: organizations through which people negotiate the uses of their contributions, establish conditions for access, verify what happens, and share in the resulting value.
Participants would delegate limited authority to a brokerage while retaining rights over how it operates: its mandate, management, fees, distribution rules, and participation in larger networks. The brokerage could represent creators, researchers, curators, workers, communities, or organizations contributing useful information.
The economic questions begin with the properties of that information. Where did it originate? Who contributed to it? What permissions apply? How was it transformed? Which contributions improved its usefulness? What obligations follow when it becomes part of a dataset, model, analysis, or other derivative product?
Originality, provenance, usefulness, and entitlement to payment would be separate claims, each requiring appropriate evidence. A recorded timestamp can help establish when a contribution entered the system. Authorship, novelty, lawful permissions, and economic value require further evaluation. Value also depends on the task: a contribution can be essential in one context and readily replaceable in another.
Derivatives make this especially interesting. A dataset may combine original records, annotations, corrections, and successive transformations. A model may depend on contributions from many organizations. A provenance system could preserve those recorded relationships and connect them to explicit agreements about attribution, permitted use, and compensation. Those agreements would define which obligations continue downstream, their duration, and how they share a finite revenue budget. Recorded ancestry alone would not establish perpetual ownership over everything later influenced by it.
Security by construction would give these arrangements an operational foundation. Within a defined execution environment and security model, valid authorization could become a precondition for computation. An accepted operation would carry evidence connecting its inputs, permissions, transformation, and resulting obligations. Participants and auditors could inspect that evidence when checking compliance or resolving a payment dispute.
The ambition is to make rights usable throughout the system. A brokerage authorizing an evaluation should be able to specify that scope, enforce it within the controlled workflow, and obtain evidence of the covered operation. Information already disclosed outside that boundary still requires contractual and institutional protections.
Tokenomics would organize the economic relationships: payments, contribution claims, shared infrastructure funding, and settlement between brokerages. Access credentials, membership votes, revenue entitlements, and payment instruments have different functions and should have distinct rules. Buying a financial claim should not automatically confer authority to expand the use of another participant’s information.
Rewards should support useful contributions and stewardship: creation, verification, curation, maintenance, expert evaluation, and secure computation. Paying primarily for collection volume would encourage more collection. Sustainable value must come from useful services and funded demand; issuing tokens cannot establish either.
Collective bargaining gives participants a practical way to exercise their rights. A brokerage can pool technical expertise, negotiate conditions, fund audits, challenge underpayment, and refuse unacceptable offers. Members also need powers to inspect their representative, contest decisions, replace leadership, and leave. Shared benefits could include income, better tools, research, and resources for future work.
Multiple brokerages could federate through an architecture resembling multi-cloud coordination. Each would retain its own mandates, data controls, and internal economic rules while sharing discovery, contract negotiation, approved computation, provenance, and settlement. Together they could deliver services that no single brokerage could provide alone.
At the scale of billions of transactions, the goal would be to preserve verifiable permissions and obligations through participating systems without making everyone’s activity publicly observable. Contributors could verify their entitlements, buyers could check relevant permissions, and authorized auditors could examine compliance. Sensitive information and transaction histories would need their own protections.
This is the case for an alternative to surveillance capitalism: people gain institutions that can act on their instructions, negotiate collectively, and hold information services accountable. Businesses can earn by delivering authorized, useful results under terms participants can influence.
That alternative also needs protected spaces for refusal, public knowledge, and uses that communities choose to support without payment. Information about one person can affect others, so member approval cannot settle every affected interest.
The goal is an information economy where people can contribute, collaborate, and benefit while retaining meaningful rights as their information moves through larger systems. Security supports enforceable boundaries. Provenance supplies evidence. Tokenomics allocates agreed returns. Participant governance gives those capabilities their purpose.