ME|META| (@interMemetica) header
ME|META| (@interMemetica) profile picture

Weaving a tighter fabric of universal self-discovery. || membre.lla STUDENTS || -kapnet || Encyclopedia InterMemetica by @LaggardLabs

Everywhere at once
Joined July 2015
Photo by ME|META| (@interMemetica)Photo by ME|META| (@interMemetica)Photo by ME|META| (@interMemetica)Photo by ME|META| (@interMemetica)Photo by ME|META| (@interMemetica)Photo by ME|META| (@interMemetica)Photo by ME|META| (@interMemetica)Photo by ME|META| (@interMemetica)Photo by ME|META| (@interMemetica)
The human psyche, concentrated in the individual, interacts clumsily with itself in human relations not specifically pyche-centric.
1
10
16
fruit flies and humans (!) share the same set of genes to form a body plan. whenever i see them i see the power of generativity granted by computation: one shared set of rules giving rise to endlessly varied forms.
10
24
3
234
7,264
ME|META| retweeted
Covering your device's camera with tape blocks the lens, but it does nothing to the sensor right beside it. In January 2024, MIT CSAIL researchers published a paper in Science Advances showing how the ambient light sensor on a tablet can be repurposed into a coarse imaging system. On Android, ambient light sensors require zero runtime permissions. Any app can read the raw lux values in the background, with no permission prompt and no green camera dot on your screen. The sensor is just a single photodiode designed to auto-dim the screen. It has no lens, no shutter, and no pixel array. It only outputs a single brightness number over the internal bus. The researchers turned that single number into an imaging tool using dual photography, treating the display screen as an active array of light sources and the photodiode as a single-pixel detector. As the screen displays changing light patterns, that light reflects off your hand hovering above the glass back into the sensor. An algorithm called Plug-and-Play Quantized Compressed Sensing combines an L_p-norm dequantizer with a deep denoiser to solve the inverse problem. By matching the known display output against the tiny fluctuations in reflected light, it reconstructs an image of the hand from the perspective of the display. The research was demonstrated on an unmodified 17.3-inch Samsung Galaxy View2 tablet, recovering 32 by 32 pixel images of hand positions, pointing, pinching, and scrolling. The physical constraints are severe. The attack needs objects in contact with or centimeters from a large display, and small smartphone screens do not provide enough surface area. Reconstructing a single frame with subsampled patterns took 3.3 minutes, while natural video took tens of minutes. The experiment demonstrates that zero-permission passive sensors leak spatial interaction data when paired with an active screen. Operating systems could neutralize this by placing the sensor behind a runtime permission prompt, capping its sampling rate to 1 Hz, and quantizing lux readings into coarse steps. Until mobile platforms restrict the bus, physical tape on the lens leaves the adjacent light diode completely accessible.
6
41
5
387
34,850
No tokens. No yield. Just bits of patience. -kapnet hbu
1
6
5
57
ME|META| retweeted
I've always believed that intelligence should come from simple, general building blocks that learn their own abstractions directly from raw data; in other words, the bitter lesson. I think we're finally seeing a real step towards AI that can learn from all data. Language modeling on raw bytes is a microcosm of this ideal, and to me it's a central problem in AI: if we can't build LLMs that learn end-to-end from bytes, how will we handle the rest of the world's data that we understand even less? Last year we introduced hierarchical networks (H-Nets) as a first step. Our new version finds up to 3-5x efficiency gains in controlled scaling law experiments, and at larger scale matches an open-source model trained on 11T tokens with ~10x less data and compute. The posts below walk through what we found and why we think it works.
261
826
287
7,385
1,028,272
New paper from my group. We gave coding agents a formal executable semantics of the language plus check.intentcomputing.org, a kit of proof procedures, then asked them to prove the code they generate correct. ➡️ HumanEval: 164/164 proved, and all passed audit. ➡️ arxiv.org/abs/2609.39022
8
14
4
111
17,956
Software development is now capped by math ideation and that limit is directly set by your knowledge of the furthest extent of developed math.
3
3
4
45
1,776
BREAKING DISCOVERY: Humans alternate between conscious and unconscious states 15 times per second.
60
211
40
3,533
259,031
ME|META| retweeted
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works already-clean.goodfire.com/
890
2,645
1,459
19,911
4,700,289
Announcing LayerTree: a new tree-based scaling solution for Bitcoin that permits O(1), independent of the number of users in the system: teapawt.org/lightning/status/21108… A single UTXO can hold the balance of millions of users as an off-chain Merkle-sum tree. Transaction introspection techniques permit batched exit from the system in a single transaction w/o any pre-signing: layertree.tech
Introducing LayerTree, a non-custodial scaling solution on bitcoin that connects to Lightning. Using covenants, one UXTO can be shared by a million users. Any user can leave at any time in a single transaction. Users don't need to be online to receive. Funds never time out. LayerTree is the next step in bringing bitcoin to billions. Learn more: layertree.tech
58
125
24
493
47,255
65
165
26
3,233
49,724
Do you suffer from AI psychosis? The robot doctor is in, and ready to check your vibes.
1
2
7
590
wow the base model understands when it is appropriate to create a 2-simplex of 1-cells as a witness for composition of relations between relations of concepts in theta space ontologies.
4
6
1
43
1,205
(1/7) Why do different languages/modalities have similar "platonic" representations in neural networks? Why does similarity peak in the middle layers of LLMs? When does it enable cross-lingual transfer? Our new paper develops a theory based on shared hierarchical structure. 🧵
2
49
4
302
11,924
100%
Compositionality in ontologies help semantic crawlers traverse composed elements in less steps
17
We've been working for decades to reduce the amount of cognitive and keyboard overhead in simply communicating with the machine while developing software. LLMs are a realization of the first natural language programming interface that actually works, and for some reason people are very confused about it. Some folks think it's their new God and others think it's an unholy terror when it's just a natural language interface between man and machine that actually works.
This is backward. Software used to be about programming, which was about thinking and typing. Now it's about truly building things, pretty much at the speed you can think. The big difference is taste. 10 years ago a great programmer didn't need taste. There was too much work to do, someone else could handle that part. Now you need taste, and you face a constant stream of decisions. You can't hide in the bits anymore! When we built TMOG, it wasn't a prompt - it was HUNDREDS of prompts placed by multiple devs for months. AI didn't create the interface - we did, with hundreds if not thousands of small decisions. That's the irony - to me, it was ALL craft. And I didn't miss the typing at all. tmog.org
4
6
27
1,228
Who is AI coming for next? My hypothesis: the rest of the g-loaded/verifiable quadrant. How the others fare is an open question.
125
65
42
759
265,459
ME|META| retweeted
Our new preprint is now available! Thalamocortical credit routing from the basal ganglia and cerebellum: anatomical constraints and circuit principles biorxiv.org/content/10.64898…
1
29
1
125
5,045
ME|META| retweeted
𝗛𝗲𝘁𝗲𝗿𝗮𝗿𝗰𝗵𝘆 𝗶𝗻 𝘁𝗵𝗲 𝗯𝗿𝗮𝗶𝗻 New preprint with Andrea Gambarotto. Neuroscience often treats control as a hierarchy: the structures behind a behavior hold a stable rank. We argue this mistakes a configuration for an architecture. 🧵 arxiv.org/abs/2610.04643
12
87
8
441
21,786
Amazon AGI presents Adaptively Looped Diffusion Language Models Outperforms all evaluated DLMs and the corresponding AR baselines in average benchmark score proj: alo-dlm.github.io/ abs: arxiv.org/abs/2610.04198 repo: github.com/amazon-science/AL… model: huggingface.co/amazon/ALoDLM…
8
74
4
534
26,582
ME|META| retweeted
Ever seen a child vibe code before? Kids today are growing up knowing that they can learn and create anything they want. That confidence in your own abilities is what I believe is the biggest civilizational unlock of AI. It’s the beginning of a new human era.
8
13
152
4,508