Molecular networks that can learn & GrapheneOS on Pixel phones - Hacker News (Feb 17, 2026)

Please support this podcast by checking out our sponsors:
- Invest Like the Pros with StockMVP - https://www.stock-mvp.com/?via=ron
- KrispCall: Agentic Cloud Telephony - https://try.krispcall.com/tad
- Effortless AI design for presentations, websites, and more with Gamma - https://try.gamma.app/tad


Support The Automated Daily directly:
Buy me a coffee: https://buymeacoffee.com/theautomateddaily

Today's topics: Molecular networks that can learn - Simulations of gene regulatory networks (GRNs) show Pavlov-style conditioning, rising causal emergence (phi), and hints of agency below organism level—keywords: Michael Levin, associative learning, causal emergence, integrated information. GrapheneOS on Pixel phones - A hands-on switch from Apple/Android to GrapheneOS: Pixel-only support, web installer steps, sandboxed Play Services, profiles for compartmentalization—keywords: privacy, security, Verified Boot, Titan M, Obtainium, Aurora Store. ASCII control keys explained clearly - A four-column 7-bit ASCII table makes CTRL-key mappings intuitive by zeroing high bits; it clarifies ESC as CTRL+[ and explains ^M with CRLF—keywords: control characters, caret notation, terminals, bitwise AND. Undo behavior across vi variants - A tour of undo/redo semantics from POSIX vi’s single-level toggle to Vim’s multi-level model and nvi’s dot extension—keywords: u, Ctrl-r, POSIX, compatibility, editor ergonomics. Dolphin adds Triforce arcade support - Dolphin Emulator’s mainline builds now run the GameCube-based Triforce arcade platform, covering hardware boards, security keys, Segaboot, and multiplayer/network quirks—keywords: Triforce, JVS, DIMM, Mario Kart Arcade GP, F-Zero AX. Rendering a realistic visible spectrum - An essay argues most “visible spectrum” images are wrong and proposes a physics-driven spectrum renderer with gamut mapping and Abney-effect corrections—keywords: CIE XYZ, sRGB, chromaticity, gamut, spectrumrenderer. Origami folds for disaster shelters - A 14-year-old used Miura-ori parametric design to create fold patterns that hold up to 200 pounds, pointing toward rapid-deploy emergency shelters—keywords: Miura-ori, strength-to-weight, deployable structures, testing rig. Fixing the incentives in science fairs - A critique of modern science fairs says they reward lab-affiliated ‘glorified internships’ over student inquiry, proposing categories like null results and replication—keywords: inquiry learning, reproducibility, incentives, Regeneron/ISEF. Why OKRs beat Deming in US - An argument that OKRs thrive because they reduce managerial complexity, while Deming’s statistical-process approach demands deeper system understanding—keywords: Drucker, Deming, management by objectives, variability, SPC.

-https://blog.tomaszdunia.pl/grapheneos-eng/
-https://garbagecollected.org/2017/01/31/four-column-ascii/
-https://www.smithsonianmag.com/innovation/this-14-year-old-is-using-origami-to-design-emergency-shelters-that-are-sturdy-cost-efficient-and-easy-to-deploy-180988179/
-https://www.newscientist.com/article/2513815-how-teaching-molecules-to-think-is-revealing-what-a-mind-really-is/
-https://asteriskmag.com/issues/13/rethinking-high-school-science-fairs
-https://utcc.utoronto.ca/~cks/space/blog/unix/ViUndoMyViews
-https://dolphin-emu.org/blog/2026/02/16/rise-of-the-triforce/
-https://brandonli.net/spectra/doc/
-https://surfingcomplexity.blog/2026/02/16/poor-deming-never-stood-a-chance/


Episode Transcript

Molecular networks that can learn
First up: the most mind-bending item of the day—biological simulations that treat “agency” as something that can show up far below the level of an animal.

The piece follows work associated with Michael Levin’s lab at Tufts, looking at gene regulatory networks—those dense webs of genes, proteins, and RNA that collectively decide what a cell does. In simulations built from real biological network data, researchers ran a kind of Pavlovian experiment: repeatedly pairing a “neutral” stimulus with a “functional” one by stimulating nodes together. After training, these networks responded to the neutral input as if the functional one were coming next—suggesting a primitive form of associative learning and memory.

Then it gets even more provocative: they quantify how “self-like” the network becomes using a framework called causal emergence, tied to the idea that sometimes the whole system has more explanatory power than its parts. In the simulations, learning increased this system-level integration measure—phi—and the more learning you saw, the bigger the bump. Critics rightly note this is in silico and simplified, not proof that real cells “think.” Still, it strengthens a broader, increasingly popular view: cognition and agency might be a continuum—from brains, to tissues, to immune systems, and potentially down into chemistry.

GrapheneOS on Pixel phones
From deep biology to everyday privacy: a detailed field report on switching away from Apple’s ecosystem and mainstream Android, and living on GrapheneOS.

The author’s motivation is part curiosity, part politics: reports that France wanted to force a backdoor into this security-hardened Android variant made it feel worth investigating. GrapheneOS is open-source and AOSP-based, and its big architectural move is removing Google services from the operating system itself—while still allowing Google Play Services to run in a sandbox if you choose.

There’s an important constraint, though: official support is Pixel-only, precisely because Pixels offer strong hardware security primitives like Verified Boot and the Titan M security chip. It’s a little ironic if your goal is to “de-Google” your life, but the trade is clear: you’re separating Google-the-hardware from Google-the-services.

The walkthrough is practical and careful. You use the official web installer, a data-capable cable, and a Chromium-based browser. You enable OEM unlocking, unlock the bootloader with Fastboot, flash the factory image, and—crucially—only re-lock the bootloader after verifying the system boots and behaves correctly. Lock too early and you can turn a minor hiccup into a brick.

What makes the post especially useful is the author’s operating model after install: keep Google Play Store and Play Services only in the Owner profile for a small set of apps that truly require it—like a bank app needed for NFC contactless BLIK, or a carrier app for perks—then use a second profile as the daily driver. That second profile also becomes a “clean handoff” option: if you’re forced to unlock your device at a border or inspection, you can delete that profile without wiping the entire phone.

For apps, they lean on Obtainium to install and update open-source packages directly from upstream sources, and Aurora Store as a Play Store front-end for proprietary apps without full Google Mobile Services. They also flag the downsides: anonymous account reliability, terms-of-service risk, and an added trust layer that could, in some threat models, matter.

And in day-to-day usage, GrapheneOS’s tighter permission controls stand out—things like denying network access or sensor access to apps by default when it’s not needed—plus a “Private space” feature akin to Secure Folder. Not everything is perfect; in their setup, NFC payments still behave best from the Owner profile. But the verdict is positive: it’s a meaningful way to reduce dependence on both Google and Apple—plus a reminder that projects like this survive on donations.

ASCII control keys explained clearly
Now, two quick but satisfying entries for people who live in terminals and text editors.

First: a “four column” layout of the 7-bit ASCII table that makes old terminal lore suddenly obvious. ASCII’s 128 codes can be understood as two high bits that choose one of four columns, plus five low bits that choose one of 32 rows. The neat trick is that many control-key combinations keep the low five bits and zero out the high bits—effectively a bitwise AND with 0b0011111.

That’s why CTRL+[ becomes ESC. The character '[' lives in the same row as ESC, but a different high-bit group. Pressing CTRL removes those high bits, and you land on Escape. The same pattern explains why you see caret notations like ^J for newline, ^H for backspace, and the classic “why is there a ^M?” moment when tools reveal Windows CRLF line endings.

Second: a discussion of undo in vi and its descendants—one of those deceptively small design choices that shapes your whole editing workflow. Classic Bill Joy vi—and POSIX—define 'u' as a single-level toggle: press it once to undo the last change, press it again to redo by undoing the undo. The author argues that this standardization is a mistake in practice.

Vim-style editors took a different path: multi-level undo with 'u', and redo on Ctrl-r, both taking counts. That model is predictable even if you move around, search, or yank text between undos. Nvi sticks to the POSIX toggle behavior but adds a dot extension—like 'u..'—to go back multiple steps, though it can feel like you’re entering a fragile “undo mode” where cursor movement risks breaking the sequence. If you ever wondered why different machines feel subtly different when you type 'u', this is the lineage behind it.

Undo behavior across vi variants
In gaming and emulation news: Dolphin Emulator has published a deep dive into the Nintendo–Sega–Namco Triforce arcade platform, and—more importantly—proper Triforce support has landed in official Dolphin builds.

Triforce is basically a GameCube motherboard repackaged for the arcade world, with extra boards for arcade I/O and video output, and a media board for storage, networking, and management. The ecosystem is full of arcade-specific constraints: optical discs were considered unreliable, so games often loaded into DIMM RAM with battery-backed retention; Namco used NAND cartridges; and every title needed matching security keys.

Dolphin’s story here is also about software stewardship. There used to be a hacky Triforce branch that was dropped from mainline years ago. Now, as of a recent development build, it’s integrated cleanly—credited largely to developer “crediar,” plus a long review cycle focused on memory safety, crash prevention, and usability.

For users, Dolphin will auto-detect Triforce titles and switch modes, but you still need to configure the virtual baseboard and map Coin, Service, and Test inputs—because arcades aren’t just controllers. Segaboot—the operator menu environment—matters for full service access, and Dolphin’s guide explains where to extract it from game files.

The support is already surprisingly complete: in-article screenshots were captured from Dolphin itself, single-machine multiplayer works, Android is supported, and there are options for third-party servers for certain networked cabinets. Some big challenges remain—like touchscreen emulation for The Key of Avalon, and unresolved linking for F-Zero AX without real packet captures—but the direction is clear: Triforce emulation is no longer a side quest; it’s becoming a first-class target.

Dolphin adds Triforce arcade support
Two stories today orbit around “how we represent reality,” one in pixels and one in classrooms.

On the pixels side: an essay argues that many internet images of the visible spectrum are simply wrong—banded, mis-proportioned, and far brighter than what you actually see through a prism or diffraction grating. The author takes a physics-first approach to rendering a rainbow on a computer, walking through Grassmann’s laws, metamerism, and why human color perception effectively collapses a full spectrum into three dimensions—hence color matching functions and spaces like CIE XYZ.

The core problem is practical: many pure spectral colors don’t fit inside common display gamuts like sRGB, so you need a mapping that keeps gradients smooth without destroying hue and brightness. The essay compares several gamut-mapping strategies and then hits an interesting snag: the blue region. Simple “mix with gray” methods can drift toward violet on screen, unlike the deep blue you may see in real spectral light. The author attributes this to color appearance effects—especially the Abney effect—and chooses a pragmatic correction using equal-hue curves rather than implementing a full appearance model like CIECAM02. The result is an open-source spectrum renderer, plus fun outputs like a periodic table of atomic spectra derived from NIST line data.

On the classroom side: a strong critique of American high-school science fairs, arguing they’ve shifted from student inquiry to status competition—often rewarding what amounts to a prestigious lab attachment rather than a student-driven question. The author contrasts projects that meet “high-level inquiry” criteria—student defines the problem, designs the method, interprets the data—with projects where the core intellectual work is imported from a professional lab.

The proposed fixes are refreshingly concrete: add a “Null Results” category that rewards good experimental design even when nothing pops; a proposals-and-pilots division that emphasizes planning, budgets, and IRB realities; meticulous replication tracks that teach reproducibility; and even a fraud-exposure category that trains students to spot manipulation and errors, reviewed privately.

As a counterpoint, we also have a story about a ninth grader, Miles Wu, turning an origami hobby into serious engineering exploration. He focused on Miura-ori—famous for compact deployable structures like space solar panels—and ran a parametric study: dozens of variants, over a hundred folding trials, and load testing until failure. Some samples held up to 200 pounds—far beyond his expectations—suggesting a path toward emergency shelters that are strong, cheap, and fast to deploy. Experts also note the hard part ahead: scaling to real materials, real joints, real buckling, and multi-direction loads. But the blend of curiosity and rigor is exactly what science education claims to want.

Rendering a realistic visible spectrum
Finally, a management lens for anyone working in large organizations: why OKRs and management-by-objectives culture became dominant in the U.S., while Deming-style thinking often feels like the niche alternative.

The argument is that it’s less about who was “right,” and more about what’s easier to run. OKRs compress messy organizational reality into a handful of objectives and measurable key results—numbers that fit neatly into spreadsheets and slides. That’s a relief valve for managers with limited time and attention.

Deming’s approach, by contrast, is about understanding and changing the system—especially its variability—using statistical process control rather than simple target-setting. A process can be stable but off-target, or unstable but sometimes near the target, and the correct intervention depends on which situation you’re in. That demands sustained investigation, and as Deming’s camp liked to emphasize, many of the most important figures are “unknown and unknowable.”

So the thesis is blunt: Drucker-style objectives won because they reduce the mess for managers, even when they risk oversimplifying what’s actually happening. Deming often offers the better compass—but it asks for deeper, continuous learning, and that’s a harder sell.



Subscribe to edition specific feeds:
- Space news
* Apple Podcast English
* Spotify English
* RSS English Spanish French
- Top news
* Apple Podcast English Spanish French
* Spotify English Spanish French
* RSS English Spanish French
- Tech news
* Apple Podcast English Spanish French
* Spotify English Spanish Spanish
* RSS English Spanish French
- Hacker news
* Apple Podcast English Spanish French
* Spotify English Spanish French
* RSS English Spanish French
- AI news
* Apple Podcast English Spanish French
* Spotify English Spanish French
* RSS English Spanish French

Visit our website at https://theautomateddaily.com/
Send feedback to [email protected]
Youtube
LinkedIn
X (Twitter)