Elias Moosman (@Wiber) profile picture

Explored what comes after robots two decades. Did something different.

Joined August 2007
Photo by Elias Moosman (@Wiber)Photo by Elias Moosman (@Wiber)Photo by Elias Moosman (@Wiber)Photo by Elias Moosman (@Wiber)Photo by Elias Moosman (@Wiber)Photo by Elias Moosman (@Wiber)Photo by Elias Moosman (@Wiber)Photo by Elias Moosman (@Wiber)Photo by Elias Moosman (@Wiber)Photo by Elias Moosman (@Wiber)
Nobody expects the car crash or hurricane, which is why enterprise E&O exists. But now that decidable telemetry is available, deploying without it isn't an accident anymore. It’s Exhibit 1 for the plaintiff's attorney. I'm just leaving the blueprints on the table before the underwriters realize it. teapawt.net/Wiber/status/209806246…
You have a $1 trillion infrastructure opex, and many times that in uninsurable E&O and D&O liability. Your soft guardrails are a theater. I have the wedge you mean when you butcher deterministic autonomous agent vibes. Pay the toll at iamfim.com. And fork the repo.
1
51
You have a $1 trillion infrastructure opex, and many times that in uninsurable E&O and D&O liability. Your soft guardrails are a theater. I have the wedge you mean when you butcher deterministic autonomous agent vibes. Pay the toll at iamfim.com. And fork the repo.
61
Using soft Turing-complete models to police Turing-complete models is mathematically bankrupt. If you want decidable boundaries instead of deterministic chaos, the MCP implementation is open source: github.com/wiber/thetacog-mc…
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
2
131
If a tree falls in the forest, does it make a sound? No. Unless it has an address…
35
And if you cannot mathematically attribute the agent's action, you cannot insure the agent's deployment. The enterprise AI market isn't waiting on better software models; it is waiting on underwriters. Uninsurable agents do not get deployed. Period. teapawt.net/Wiber/status/207778602…
The law does not govern physics. Software cannot audit software. Not only is policy too slow current models fail any basic independence test. No attribution for agentic without a hardware primitive ground truth. (cc @PeterDiamandis)
53
The law does not govern physics. Software cannot audit software. Not only is policy too slow current models fail any basic independence test. No attribution for agentic without a hardware primitive ground truth. (cc @PeterDiamandis)
1
72
Guardrails are cope. You can't bolt safety onto a ghost. AI has no substrate, no traction, no geometric floor. 8 minutes on why alignment is a topology problem, not a constraint problem: youtube.com/watch?v=ZDWiJ9d7… Entropy always wins against effort. Entropy only loses to geometry.
We are walking into a dystopia I call the "Turing Police." Big Tech wants to be the centralized arbiter of Truth. They want to tell the Ghost what is real. But Truth isn't an opinion you can legislate. Truth is a geometric coordinate. 🧵
9
1
74
@NPCollapse You're building while warning. That's the right tension. 8 minutes on why agents need geometric substrate - identity that holds between calls, not just memory that persists. Curious if Conjecture's seeing the same drift patterns.
18
@janleike You left OpenAI because alignment wasn't getting enough priority. But what if the approach itself is the bottleneck? Guardrails fight entropy with effort. Entropy always wins. 8 minutes on geometric alignment - structure over behavior.
14
@StuartJRussell Human Compatible argues for provably beneficial AI. What if provability requires geometric substrate? Behavioral alignment can't be proven - it can only be sampled. Structural alignment can. 8 minutes on why the proof has to be architectural.
14
@robbensinger MIRI's thesis: alignment is hard because we can't specify values correctly. What if the problem isn't specification - it's substrate? Identity without geometric grounding drifts no matter how well you specify. 8 minutes on why alignment is topology, not constraint.
13
@tegmark You wrote about consciousness as mathematical pattern. This is 8 minutes on substrate relativity - identity grounded geometrically so alignment becomes structural. The hard problem inverts: consciousness isn't the mystery, it's the constraint. Does this track?
11
@AnthropicAI Constitutional AI is behavioral alignment - training the response. What about structural alignment - grounding the substrate geometrically so drift becomes architecturally impossible? 8 minutes on why guardrails fight entropy with effort, and entropy always wins.
11
@ESYudkowsky Corrigibility assumes you can patch alignment onto an agent after the fact. What if the substrate has to be geometric from the start? Identity grounded structurally, not behaviorally. 8 minutes on why entropy only loses to geometry. Would welcome the critique.
11
@ylecun You've argued that current alignment approaches are solving the wrong problem. What if alignment isn't a constraint problem at all - it's a topology problem? 8 minutes. Curious if this framing changes anything or if I'm still in the weeds you've already mapped.
12
@geoffreyhinton You've been saying we can't guardrail our way to safety. This is 8 minutes on why - and what a geometric alternative might look like. Does the topology framing land or if I'm missing something you've already seen?
21
Complexity escapes when you can't measure it. Knight Capital: $440M lost in 45 minutes YouTube: Radicalized millions accidentally Next: AI does... what exactly? We made complexity measurable by giving words physical addresses. Meaning is now physics. (It's called the Unity Principle - now you can herd cats)
2
1
70
For the physicists: Semantic→Physical→Hardware Words get deterministic locations. Processors work less hard when categories map to reality. The computer knows exactly where to reach. The cats can't escape.
1
1
26
Let’s break this scenario down step by step by using the example of a high-profile public figure blocking someone and how that action can inadvertently lead to content suppression, shadow banning by proxy, and further polarization. The Scenario: A high-profile public figure blocks someone on a platform (social media or otherwise). The platform's algorithm interprets this action as a signal, using it to curate what content their followers see or don’t see. This blocking behavior indirectly informs the algorithm about the tastes and preferences of the figure’s followers, leading to a self-reinforcing feedback loop that could exclude certain content. This, in effect, becomes a kind of shadow ban for the blocked person and their content, not just for the public figure, but potentially for the entire follower base. Now, let’s explore the effects of this, who is interested in maintaining or disrupting it, and how this presents a market opportunity. Polarization and the Algorithmic Consequence Self-Referencing and Accelerating Effect: When a high-profile figure blocks someone, the platform algorithm learns from this and begins to infer that the followers of the public figure would likely share the same preference (i.e., disinterest or aversion to certain content). This creates an echo chamber. The more people who follow the figure, the more they are protected from content the figure dislikes or actively rejects. As more content is filtered out, followers are shown fewer dissenting views, reinforcing a narrow band of ideas and preferences. This contributes to polarization. Shadow Ban by Proxy: Though not an intentional platform feature, the algorithm’s behavior could result in the blocked person’s content being less visible to people who follow the high-profile figure. This isn’t a traditional “shadow ban,” where the platform itself suppresses content. Instead, it’s a consequence of how signals from influential figures propagate through their network, making their preferences into collective behavior. Who Wants This Change to Happen? Advocates for Diversity of Thought: People and organizations interested in content diversity and free speech would push for changes to the algorithm to ensure one user’s actions (especially high-profile ones) do not affect the broader content visibility for others. Content creators who risk being shadow banned would advocate for transparency and fairness, so their content is not suppressed just because it doesn’t align with a public figure's preferences. Users Seeking More Control Over Algorithms: A growing demand exists among users for more control over what they see. Many would support a system that allows them to decide, not have algorithms decide for them, which content is filtered or promoted based on a single action (like a block from a high-profile figure). Who Wants to Maintain the Status Quo? Platform Algorithms Optimized for Engagement: Social media platforms and content distributors benefit from engagement-maximizing algorithms. Platforms might see value in this behavior since it keeps users in an echo chamber where they are comfortable, thus increasing time spent on the platform. Influencers and Public Figures: Many public figures may inadvertently benefit from this algorithmic behavior because it preserves their influence and reinforces their voice without being challenged by opposing views or content they dislike. Advertisers and Brands: Advertisers targeting the followers of high-profile figures might appreciate this self-reinforcing mechanism because it makes targeting easier. They can better predict the preferences of a public figure's audience, knowing that dissenting or non-aligned content will be less visible. Market Niche for Remedying This: A solution that addresses this issue could tap into several key markets: Content Moderation Tools for Influencers and Public Figures: Value Proposition: Offer influencers and public figures a tool that gives them control over how their actions (like blocking someone) affect their followers’ content visibility. This tool would allow them to decide whether their blocking action should cascade down to their followers, preventing accidental polarization. Market: Public figures, influencers, celebrities, and organizations concerned about audience curation and broader societal impacts. Content Discoverability for Creators: Value Proposition: A service or tool for content creators that ensures their content remains discoverable despite actions like being blocked by a high-profile figure. This tool could offer insights into how actions taken by others affect their content visibility and provide ways to boost their reach in a fair, algorithmic landscape. Market: Small content creators, independent artists, journalists, and public thinkers whose views might diverge from the mainstream. Transparency Tools for Platforms: Value Proposition: Build transparency tools for platforms that help users see why they are being shown (or not shown) certain content. This could allow platforms to maintain engagement without being seen as a driving force behind polarization. Market: Social media platforms, content distributors, and streaming services looking to improve trust with users by offering more control and transparency into their algorithms. Verbalizing the Effect of Polarization in a Way that Creates Economic Value To articulate this in a way that creates economic value, the framing should be around user empowerment, influence control, and transparency: User Empowerment: “We give users more control over what they see, helping to foster a more diverse and vibrant community by breaking out of algorithm-driven echo chambers.” Influence Control for Public Figures: “Our tool helps influencers and public figures manage their audience more effectively, ensuring their content preferences don’t unintentionally limit the content their followers can discover.” Transparency and Trust for Platforms: “By introducing transparency tools, platforms can maintain high levels of engagement while building trust by showing users exactly why they’re seeing certain content—and what’s being filtered out.” Who’s on the Side of Changing This? Content creators and diversity advocates who want a fair playing field. Public figures concerned with the broader impact of their actions on their audience and society. Platforms seeking transparency and trust as a long-term strategy rather than short-term engagement boosts. Who Isn't Improving It? Platforms currently profiting from echo chambers that drive engagement through narrow content preferences. High-profile figures who benefit from unchallenged influence and maintain a passive role in how their actions affect their followers’ content consumption. Conclusion: The effect of high-profile public figures blocking others and how it inadvertently polarizes followers presents both challenges and opportunities. On the one hand, platforms and influencers benefit from a narrowed, highly predictable content ecosystem. On the other, there is a market opportunity for tools that empower users, increase transparency, and preserve content diversity, which could mitigate polarization. The value proposition lies in building solutions that foster user control, fair discoverability for creators, and platform trust—key drivers of economic value in an increasingly polarized digital landscape.
1
115