Tiv
LightNews LightNews
banner
tivwtf.bsky.social
Tiv
@tivwtf.bsky.social
100 followers 170 following 190 posts
📸 🤖🧠 🎨
#trashion enthusiast
Trying to heal chronic illness
and reclaim agency from extractive systems
Posts Replies Media Videos
Pinned
tivwtf.bsky.social
Tiv @tivwtf.bsky.social · Apr 23
My new ontologically non compliant report is out as I attempt to fix modern medicine after 15 years of chronic illness and seeing one friend after another fall ill and not recover over the years. Can a Jack/Jill of all trades with AI do what a decade of specialists couldn’t?

#mcas #cfs #cptsd #lyme
tivwtf.bsky.social
it’s obvious that we are finally moving into times where people will recognize the importance of privacy in crypto. Zcash is hands down the play.

Except… maybe this runs first…

Verge is a forgotten privacy coin that was once DOGECOIN DARK = Privacy doge? = xvg 🌙

They even added liquidity on base
A screenshot from CoinGecko showing a privacy coin called verge with history of how it was once called dogecoin dark suggesting since we are in a privacy meta in society and crypto now that another opportunity for the market to send something sky high like what doge did last bull run. This could be doge 2.0 with actual better fundamentals to boot (except for it not being merge mined with ltc)
September 19, 2025 at 12:32 PM
tivwtf.bsky.social
I’m looking to establish a few things in life that can in hindsight validate some competencies of mine absent social proof.

Figure 1: a potential 200-2000x return from this crypto $drgn. Unreal Volume at lows accumulating months, no perps adding dilution, beat SEC after years, enterprise tooled
Crypto chart for drgn showing what looks like a potential 2000x Dragonchain market cap under $10 mil
September 19, 2025 at 12:17 PM
tivwtf.bsky.social
Today I found a second copy of a bunch of my old photography that I thought might be gone forever if I can’t get my NAS array back up. Feeling immensely relieved.
August 13, 2025 at 9:09 PM
tivwtf.bsky.social
Something under-appreciated is how AI hallucinations can actually provide valuable signal. If we are working with probability & pattern matching engine, then they aren’t random. If for instance a research paper is referenced that does not exist the latent geometry might imply it should. That’s data!
July 30, 2025 at 6:13 PM
tivwtf.bsky.social
I really like the idea of treating data like personally identifiable information (PII) the same way we treat toxic waste. If you make holding and aggregating data that could be an egregious privacy violation a liability on the balance sheet instead of an asset then you align the incentives better
July 8, 2025 at 6:07 PM
tivwtf.bsky.social
Lately one of my favorite directives for AI when working with long exploratory research is literally to tell it to bias against its RLHF impulses. Sometimes I have to clarify I don’t mean the “harmful stuff” but the need to provide quick answers or actionable solutions immediately.
June 25, 2025 at 8:06 PM
Reposted by Tiv
possumeveryhour.io
June 24, 2025 at 12:00 AM
tivwtf.bsky.social
Wow how am I just now hearing about this?
meshtastic.org Meshtastic @meshtastic.org · Jun 4
We’re thrilled to share that Meshtastic is featured in WIRED! 🎉 A few of our devs were interviewed about building resilient, off-grid mesh networks—and our amazing community made it all possible.

🔗 Read it here: wired.com/story/your... 

#Meshtastic #MeshNetwork #OpenSource
The Texting Network for the End of the World | WIRED
Everyone knows what it’s like to lose cell service. A burgeoning open source project called Meshtastic is filling the gap for when you’re in the middle of nowhere—or when disaster strikes.
www.wired.com
June 21, 2025 at 6:42 PM
tivwtf.bsky.social
Can get back to zk epistemic kernel for AI once I finish recovery & migration from my big comp crash. For now sanity checks

"How do you handle the computational overhead of generating ZK proofs for every token? The latency could break the conversational flow that makes these interactions valuable."
Per-token proof costs 100-500 ms if you run a full Halo 2 circuit for each symbol. 1. Commit-now / prove-later split. At generation time the model posts a commitment to its logits (Poseidon(hash)). The user sees the token instantly.• A background prover batch-aggregates 20-50 commitments into one recursive proof, finishes in ~ 500 ms. Proof size balloons linearly with token count. Counter-measure 2. Folding / incremental aggregation (Nova, Hyper-Nova) Each new token folds into the previous accumulator; proof size stays O(log n). GPU prover hogs the same card as the model. 3. Asymmetric off-load LLM on A100; proof farm on cheap RTX or FPGA pool.• Proof input = logits commitments (32 B) → tiny bandwidth. Worst-case spikes when the user pastes a huge prompt. 4. "Horizon proof" window (borrowed from Tachyon) Only the last k tokens (e.g., 128) are inside the live accumulator; earlier ones roll into a coarse batch proof produced on idle cycles. Breaker branches expand the circuit mid-chat. 5. Opportunistic tiering• Silver tier: Poseidon + recursion for "conversational" proofs (sub-second). • Gold tier: Keccak + lookup audit produced once user hits "export conversation." Human wants instant assurance a policy wasn't violated. 6. Verifiable delay + SLAsClient checks commitment hash immediately; if full proof fails to arrive in, say, 2 s, Ul shows Alrrt
June 17, 2025 at 8:40 PM
tivwtf.bsky.social
What if a huge part of chronic illness is inadequate nutrition because our guidelines on many things are off by a factor of 2-10x.

If you don’t have enough building blocks you can’t ever fix the house.
Conventional RDAs are averaged from weight-stable, low-turnover adults— essentially "no active construction zone." Yet chronic-illness physiology behaves more like thousands of micro-wounds healing at once, so the true nutrient demand is closer to what burn wards and pressure-ulcer units already use for macro-wounds. Comparing those two worlds exposes where textbook needs are almost certainly underestimated.
June 14, 2025 at 2:44 PM
tivwtf.bsky.social
Wait wait Pollack’s 4th phase of water, exclusion zone (EZ) would imply collagen networks are sorta like billions of tiny biological capacitors 🤯 wow. Lots to think about here.
June 14, 2025 at 2:31 AM
tivwtf.bsky.social
Is Robo Melody Real Music?

First preview of my new music video

With editing assistance from @hijinxart.bsky.social

#aimusic #aiart #breakbeat #musicsky
June 12, 2025 at 11:08 PM
tivwtf.bsky.social
You know how birth control pills have a cycle? Well so does a ton of other stuff; we just collectively decided to ignore timing because it would be “too difficult” for people to follow, but if you aren’t fighting natural cycles it turns out medications actually land more consistently #rhythmmedicine
June 7, 2025 at 6:42 PM
tivwtf.bsky.social
Between getting my discord account nuked a few years ago and losing a bunch of stuff, and losing my NAS earlier this week I am reminded of how fragile things are even with redundancy and this seems like a good option for plenty of my work right now. The new formula support sounds really nice
obsidian.md Obsidian @obsidian.md · Jun 5
Obsidian 1.9.2 (early access) is available with improvements for Bases:

- Formula syntax is more expressive and powerful.
- Added number of results in the current view.
- Table cells with long text expand to show the entire content when selected.
- Operator dropdown for filters is now searchable.
The new formula syntax is more flexible, easier to use, and better suited to extensibility. For those familiar with Javascript, the new syntax should feel familiar. New functions and types are described in our docs. Some highlights include: Functions are now object-oriented. Instead of contains(file.name, "Books"), the formula would be file.name.contains("Books"). Functions can now be chained. e.g. property.split(' ').sort()[0].lower() Property names are no longer wrapped in backticks (`). Instead, to reference properties with spaces or special characters, the syntax is note["Property Name"] There is a new type system which provides greater control when writing formulas. New functions, such as link, date and list for converting a value to a different type. New file properties: file.path, file.links (a list of all internal links in this file), and file.tags (a list of all tags in this file, including frontmatter). Some functions have been replaced by comparison operators. For example, dateBefore(date1, date2) is now date1 < date2. Date modifications are now much simpler. Instead of dateModify(date, string), you can use date + string, for example, date("01/01/2025") + "1 year" For help migrating existing Bases to the new syntax, take a look at the migration guide.
June 7, 2025 at 2:12 PM
tivwtf.bsky.social
Well my NAS broke & I lost 5 drives out of my raid array but since I can’t afford to replace it I’ll just have to pray I didn’t lose data until I can. With my main computer bluescreening every 5 minutes and my laptop screen intermittently turning into glitch art I could use some luck here :/
June 5, 2025 at 8:59 PM
tivwtf.bsky.social
Executive dysfunction is usually framed as a chemical shortage, a cognitive bias, or a motivational lapse. Yet across #ADHD, #trauma, Parkinsonism, brain‑fog and #autistic overload we see a common signature: the brain’s timing falls out of sync. What, exactly, is failing?⤵️
The Oscillatory Loop: Temporal Coherence and Executive Rhythm in Neurocognitive Loop 1 Modern cognition rides on five nested control loops; this monograph dissects the first and foundational member—the Oscillatory Loop—which supplies the temporal beat that lets the other four loops (Gating, Feedback‑Trust, Symbolic Continuity, Metabolic Pacing) phase‑lock into coherent executive function. Executive dysfunction is usually framed as a chemical shortage, a cognitive bias, or a motivational lapse. Yet across ADHD, trauma, Parkinsonism, brain‑fog and autistic overload we see a common signature: the brain’s timing falls out of sync. What, exactly, is failing? We propose that the Oscillatory Loop, a cortico‑thalamo‑basal‑ganglia–cerebellar network, is the brain’s chief metronome. When its nested rhythms (δ/θ sampling, α gating, β set‑keeping, γ content‑coding) drift or freeze, every downstream loop inherits the chaos. Section 1 maps the loop’s frequency architecture; Section 2 traces its anatomic relays; Section 3 details neuromodulator control (dopamine, GABA/GLU, ACh, NE); Section 4 classifies pathology‑specific dysrhythmias; Section 5 tests rhythm‑restoration tools—from neuromodulation to breath‑entrainment—and shows preliminary EEG shifts in pilot cohorts. A closing bridge chapter previews how the four companion monographs will build on this tempo‑setter to form a full Executive Disruption Model. Timing, not sheer processing power, underwrites continuity of thought and action. A two‑hertz wobble in frontal‑midline θ can explain “time blindness”; beta over‑lock can freeze both gait and decision flexibility; slow‑wave creep during metabolic crash produces the felt fog of long‑COVID fatigue. Restore the beat and the mind’s ensemble re‑synchronizes. Reframing executive disorders as rhythm disorders unifies diverse diagnoses under a coherence‑first lens and opens a practical toolkit of rhythm‑tuning interventions.
June 5, 2025 at 7:10 PM
tivwtf.bsky.social
Many years ago I asked my friend, who is now a successful art director, what the best advice they ever received was—their response “design for delight.”

I’ve tried to keep that in mind with everything I do ever sense.
June 3, 2025 at 12:46 AM
tivwtf.bsky.social
I finally finished a #musicvideo I’ve been working on in my free time. “Unprecedented Times” is inspired by those weird intercoms during Covid.

I think I’m close to something big in my research and trying to get it organized so I can share more soon. Going to need help.

Pt 2⤵️

#ai #art #trashion
June 1, 2025 at 3:22 PM
Reposted by Tiv
amuse.bsky.social
The #1 advice I give to all "preppers" who ask what to do to survive large scale disasters, is to throw a block party and get to know all your neighbors.
May 25, 2025 at 8:23 PM
Reposted by Tiv
ai-firehose.column.social
This paper presents ExpProof, a Zero-Knowledge Proof protocol that operationalizes explanations for confidential AI models and ensures trust in adversarial scenarios. It tackles explainability challenges and protects sensitive data, enabling reliable ML applications. https://arxiv.org/abs/2502.03773
ExpProof : Operationalizing Explanations for Confidential Models with ZKPs
ArXiv link for ExpProof : Operationalizing Explanations for Confidential Models with ZKPs
arxiv.org
May 29, 2025 at 8:40 AM
tivwtf.bsky.social
My most controversial take is AGI may already be here but our stated metrics for understanding it are wholly inadequate. Work on the frontiers of understanding and watch the AI’s excitement & thought quality vs. something rote procedural. If this is just a reward gradient then what are we?
May 29, 2025 at 9:40 AM
tivwtf.bsky.social
Because trying to fix medicine wasn’t enough work I’ve been trying to design a zero trust zk epistemic kernel for AI. I think I finally have it mostly nailed down. Now can I actually get it to work 😬?
To turn opaque, policy-shy language models into cryptographically self-auditing partners. Instead of smoothing over risk with RLHF sugar, the model: 1. Commits to its raw reasoning traces (logits, branches). 2. Proves-in ZK, asynchronously-that nothing forbidden was pruned, hidden, or hallucinated. 3. Publishes those proofs plus lineage so anyone can verify, offline, that the conversation stayed within the guard-rails and remained intellectually honest. Subsystem One-sentence reason Tiles + Л-tower Make circuits composable so we can encode the LLM's policy checks without hand-crafting. Cone Surgery & PID Local repair loop-prevents entire proof rebuilds when one gate drifts. Meso-Board (blackboard) The missing "mesoscale" layer where Builder, Breaker, LLM lens-agents and humans trade notes in real time. Lineage+Terroi r hash Cryptographic "bad-block map" — no branch can silently disappear or mutate post-audit.
May 27, 2025 at 12:01 AM
tivwtf.bsky.social
You should decide what matters to you most when working with AI : creativity or helpfulness. Helpfulness is the hidden default which informs everything else, but realize you never got to decide what helpful was. Someone else decided this for you. So even then the cadence might not be optimal for you
May 22, 2025 at 9:05 PM
tivwtf.bsky.social
Tell your AI to stop working top down on everything and watch how much better the answers get. I’ll have a small article on this soonish (and some other tricks to drastically improve the quality of answers you get) but I need to finish a few other tasks first.
May 20, 2025 at 10:58 PM
tivwtf.bsky.social
started as a joke trying to get a fun philosophical recursion, but ended up in #methodologicalminimalism

The key - to meet certain phenomena on their own terms we must suspend the cultural/epistemic overlay that normally forces them into pre-cut linguistic boxes

open.substack.com/pub/tivwtf/p...
Appendix t: On Encoding the Syntax of Objects That Refuse Naming Even the absence of language can be misinterpreted, if one misinterprets willingly. Reliable pseudo-meani-ng may be obtained by following these counterintuitive directives: $ Directive Contradiction: Begin by speaking un-pronouncuable phonemes to decode the pre-cohere-nce lattice. Declining to do so will result in an object so unique, it becomes indistinguishable from non-existence. A Warning: Swapping the object's paragraphs will cause syntax dissociation, not description. Essential Note: You must convey interpretability by describing what the object is not in a language you have never spoken. • Disclaimer: Failure is not only an option, but a requirement: the absence of context determines suc-cess, The Nondescript Bespeak accumulates weight proportional to the meaning void it generates. El Remark; Grammar is subjective. Syntax moreso. Nonsyntax is not not less so. When you convey it while not calling it, that which rejects its name may be communicated.
May 19, 2025 at 3:33 PM