Raghav Agrawal

Former Diamond Dealer | Turning research on learning, cognition, and systems into interfaces that reduce confusion, cost and decision stress.

Public Key

npub157txzy3sx6ck9kzhlsqr8g33erprgayc2utd8j7pphutqes604sqjzld9r

NIP-05 Address

[email protected]

Profile Code

nprofile1qqs209npzgcrdvtzmptlcqpn5gcu3s35wjv9w9kne0qsm79svcd86cqpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dse0rfeg

Publishing to

relay.damus.io

nos.lol

Author Public Key

npub157txzy3sx6ck9kzhlsqr8g33erprgayc2utd8j7pphutqes604sqjzld9r

Show more details

Last Notes

2026-05-30 09:07:37 CEST

The concept of references for getting next job is weird
Why would the previous employer have anything good to say for you if you're being laid off/fired/resigning
It's like your future is being held hostage by the people in the very place that you're trying to leave (or where they are trying to make you leave)

2026-05-28 13:05:44 CEST

- reply

References
When somebody loses weight, where does the fat go?
https://www.researchgate.net/publication/269772876_When_somebody_loses_weight_where_does_the_fat_go
Molecular mechanisms regulating hormone-sensitive lipase and lipolysis
https://pubmed.ncbi.nlm.nih.gov/14641008/
Mechanisms underlying regional differences in lipolysis in human adipose tissue
https://pubmed.ncbi.nlm.nih.gov/2503539/
Regional and gender variations in adipose tissue lipolysis in response to weight loss https://www.jlr.org/article/S0022-2275(20)33402-7/fulltext
Carnitine transport and fatty acid oxidation https://pubmed.ncbi.nlm.nih.gov/26828774/

2026-05-28 13:02:08 CEST

- reply

The deficit, held over weeks, does ~all the work. So track the scale trend, not your watch's calorie guess (it's off by 20-40%), and pick a deficit you can actually live on.
https://files.mastodon.social/media_attachments/files/116/651/858/368/287/743/original/05ec8348886a595e.png

2026-05-28 13:00:48 CEST

- reply

So why does it net out downward Because the burned fat physically leaves your body : exhaled, peed out and can't be re-deposited.
Eating later builds from new food; it can't recall what's gone. The day bobs up and down, but the floor steps down each night. It's a bank balance, not a memory. Rest isn't cheating
https://files.mastodon.social/media_attachments/files/116/651/854/186/080/614/original/53fdee34e8f0dc67.png

2026-05-28 12:59:50 CEST

- reply

This was the question I actually started with: if I eat and sit and sleep between workouts, how does fat loss "stick"?
Answer : it's a sawtooth, not a switch. Meals nudge the tank up; gaps, cardio and sleep pull it down.
The body isn't holding a streak. Each moment just responds to what's in the blood.
https://files.mastodon.social/media_attachments/files/116/651/849/991/432/091/original/9fd116e59ef99b53.png

2026-05-28 12:58:47 CEST

- reply

You can't out-shout an 8:1 brake ratio with more adrenaline. There's no credible "local fatty acids release the brake" mechanism and working a muscle does NOT preferentially burn the fat sitting on it, the fuel pools system-wide first. Spot reduction stays a myth.
https://files.mastodon.social/media_attachments/files/116/651/844/819/588/282/original/677e919232500e0a.png

2026-05-28 12:58:01 CEST

- reply

So why are thighs/belly "stubborn"? Those cells carry ~8 STOP bells for every 1 GO bell.
Same adrenaline, same deficit but the STOP side just out-pulls the rope, so they drain last and slowest.
There's no special lock to "unlock."
It's a ratio, it's genetic, it's often sex-linked. You can't pick where fat leaves first.
https://files.mastodon.social/media_attachments/files/116/651/843/268/188/339/original/f93f8a400a205f25.png

2026-05-28 12:57:01 CEST

- reply

Final leg. "Burn" = broken down in the mitochondria (the furnace).
But there's a guarded door: CPT-1 and here's the elegant part, eat a big carb meal and the cell jams that door shut (there's sugar to burn, why bother with fat).
Low on carbs? Door opens. Then carbon leaves as CO₂ in your breath. You exhale fat.
https://files.mastodon.social/media_attachments/files/116/651/839/185/626/329/original/71bb38747fda1c1a.png

2026-05-28 12:55:52 CEST

- reply

Freed fat hits a problem: it's oil, blood is water, oil doesn't dissolve. So it rides a taxi — albumin, a blood protein.
No taxi nearby (low blood flow)? The fat drifts back in and gets re-stored. The cut was wasted.
This is the ONE place blood flow truly matters, but the taxis pool together, so no, biking doesn't burn thigh fat specifically
https://files.mastodon.social/media_attachments/files/116/651/834/843/326/283/original/f11ba92dc1db9e5e.png

2026-05-28 12:54:53 CEST

- reply

Now the whole chain in one breath:
effort → GO bell → courier flood (insulin's low, so nothing shreds them) → foreman pops the lid + fires the scissors → bundle cut → fat exits.
Every link was a defined character two cards ago.
But exiting the cell is only halftime, the fat still has to get somewhere and actually burn.
https://files.mastodon.social/media_attachments/files/116/651/830/710/303/257/original/d703c2c47629590d.png

2026-05-28 12:53:38 CEST

- reply

The courier wakes a three-person crew.
PKA (foreman) gives orders.
Perilipin (a lid sealing the fat) gets popped off : people forget the fat is wrapped; you can't cut what you can't reach.
Then HSL (scissors) cuts the bundle into free fatty acids, small enough to leave.
Now fat is finally out of the cell
https://files.mastodon.social/media_attachments/files/116/651/825/485/360/397/original/4fedfd17531daa67.png

2026-05-28 12:52:29 CEST

- reply

How does a doorbell on the outside reach machinery on the inside? A courier molecule: cAMP.
The GO bell speeds up its maker; more couriers = louder "release."
But insulin runs a shredder (PDE-3B) that destroys couriers.
So eating doesn't just go quiet, it actively erases the signal. Two off-switches, one go-switch
https://files.mastodon.social/media_attachments/files/116/651/819/376/885/166/original/b48073279ede933c.png

2026-05-28 12:51:33 CEST

- reply

The twist: adrenaline presses TWO doorbells on each fat cell.
β = GO. α2 = STOP. And they fight over the same lever, one pushes it, one pulls it back.
So the outcome isn't about how much adrenaline you have; it's about which bell (alpha vs beta adrenergic receptors) the cell has more of.
https://files.mastodon.social/media_attachments/files/116/651/812/928/978/677/original/6fbb9c8472c97651.png

2026-05-28 12:49:29 CEST

- reply

WHY does a deficit help? Insulin rises with blood sugar.
Eat less than you burn → blood sugar sits low more of the day → insulin is quiet more of the day. That's it.
"Deficit" isn't a switch the body detects, it's just more hours with the storage-signal off.
https://files.mastodon.social/media_attachments/files/116/651/809/828/867/141/original/c10fcc1ee019c9ef.png

2026-05-28 12:48:59 CEST

- reply

Two hormones run the valve. Insulin (you ate → STORE). Adrenaline (effort/fasting → RELEASE). They trade off all day. Crucial point that the whole thread builds toward: the cell only obeys whichever one is loud right now. It keeps no tally. Remember that when we get to "calorie deficit."
https://files.mastodon.social/media_attachments/files/116/651/806/746/794/361/original/afaa6779743978d9.png

2026-05-28 12:48:16 CEST

- reply

Stored fat is a triglyceride, a bundle too big to leave the cell. So "losing fat" isn't fat sliding out. Something has to cut it small enough to fit through the wall first. Hold this image: the fat cell is a fuel tank, and the whole rest of the thread is about what opens the valve
https://files.mastodon.social/media_attachments/files/116/651/803/279/236/128/original/3b781e6775ca6807.png

2026-05-28 12:47:27 CEST

- reply

First, three words, because everything downstream is made of them. Molecule = a piece of stuff. Receptor = a doorbell on the cell's wall. Enzyme = a worker that changes other molecules
https://files.mastodon.social/media_attachments/files/116/651/800/645/973/725/original/948fcee4afa23152.png

2026-05-28 12:46:43 CEST

How fat is actually lost, not the spot-reduction "burn it off your thighs" version, but the real molecular chain, built from zero.
https://files.mastodon.social/media_attachments/files/116/651/798/134/398/701/original/384f2c083991439b.png

2026-05-09 18:38:21 CEST

"The mystique of science is an essential tool for technocratic rule. Through it we are all taught that science, as the paradigm of all expert knowledge, has an objective, superior, and special truth that only the superintelligent few can understand."
Jay Lemke
"Talking Science"

2026-04-10 07:11:51 CEST

How Paradoxical Questions and Simple Wonder Lead to Great Science | Quanta Magazine
https://www.quantamagazine.org/how-paradoxical-questions-and-simple-wonder-lead-to-great-science-20250528/
"Almost 12 years ago, while we were building a microscope to find malaria parasites in rural clinics, we ended up inventing Foldscope — world’s most frugal microscope that people can carry in their pockets. It wasn’t practical to diagnose malaria with, but it led to the educational context immediately because then, for a price point of a dollar, anybody could observe the microscopic world"
https://files.mastodon.social/media_attachments/files/116/378/690/745/592/103/original/c0cf7c80da8ce44b.png

2026-04-05 22:02:39 CEST

- reply

Wakame pointed a very useful concept : Cyclomatic Complexity, couldn't that be used to visualize and segment the code
Like for example this kind of viz but for codebase
https://files.mastodon.social/media_attachments/files/116/353/882/633/878/065/original/d9d60c9a740c3e15.jpg

2026-04-05 21:27:03 CEST

- reply

Maybe we can learn from scada, plc programmers
The way they keep track of stuff

2026-04-05 20:33:11 CEST

- reply

Which means now more than ever we need tools to visualize and make sense of codebase

2026-04-05 20:20:22 CEST

- reply

One thing I've realized, software engineering is extremely extremely different than *computer science* or whatever they teach in a 4 year cs degree
Like the number and kind of abstractions for the entire supply chain of frameworks and libraries and the way it's all run and maintained is so different than learning how to merge sort
https://mastodon.social/@gvwilson/110367826277482266

2026-04-05 19:56:00 CEST

- reply

Every time you need to tweak one line, you have to mentally walk all ten factories at once then?

2026-04-05 19:36:47 CEST

RE: https://mastodon.social/@impactology/116346981818073311
If millions of lines code was visualized as a factory skid like with pipes and tubes, like an oil refinery or food processing or pharma unit let's say what would that be like
#note10as…jel2

2026-04-04 16:47:33 CEST

- reply

There's a technique called orbital welding, where they make the bends of the pipe in a smooth continuous automated manner right from the start rather than joining another component at the end
You know why they do this in food industry? So that even minute food residue doesn't get stuck in the joints and crevices

2026-04-04 16:00:14 CEST

- reply

It's funny
There are industrial automation companies like abb robotics and fanuc with engineers and researchers doing much more rigorous research and engineering work with a much higher bar of safety than AI companies of silicon valley and none of their researchers do the cult like complexity theatre performance that researchers at these frontier model labs do

2026-04-04 15:13:54 CEST

- reply

Real engineering deals with the laws of thermodynamics and material science. things that don't have bugs you can just patch later. Once something breaks or is contaminated you can't just be like we'll tackle this later.

2026-04-04 15:09:00 CEST

RE: https://mastodon.social/@impactology/116346202875671550
Pipes & valves skids for pharma and food. or machining tools, conveyer belts for appliances. Basically making the tooling for factories. This is real engineering, process engineering
Orders of magnitude more complex than tooling for making software
If level of rigour for quality testing used for Clean-In-Place (CIP) piping skids of a biologics facility was used for AI software tooling, imagine the kind of software reliability we'd have instead of whatever the fuck today's genai tools spit out
#note1myv…aq7u

2026-04-04 13:29:27 CEST

RE: https://mastodon.social/@impactology/116339395553357184
Aseptic Pigging Cleaning System
https://www.spxflow.com/apv/products/aseptic-pigging-systems/
Flexible, bullet-shaped projectile made of food-grade silicone that fits perfectly inside the pipe. After a batch of sauce/dip is finished, pig is launched through the pipe using compressed air or sterile water. It acts like a squeegee, pushing 99% of the leftover sauce/dip into the bottling tank
This leaves the pipe walls pre-cleaned before cleaning fluids even enter
Amazing technology to clean pipes from within
https://www.hps-pigging.com/
#note1cr2…v3m3

2026-04-03 08:38:16 CEST

What a useful technology
https://www.youtube.com/watch?v=hoxHW0lSkME&t=5s
https://www.hiperbaric.com/en/hpp-technology/hpp-tolling/
https://www.thyssenkrupp-uhde.com/high-pressure-processing/en/what-is-hpp
HPP Tolling (High-Pressure Processing) is a service where food makers use high-pressure machines to pasteurize their packaged products without heat. Using 6,000 bar (psi) of water pressure, it kills off pathogens like listeria and E. coli while keeping the fresh flavor, nutrients, and texture.

2026-03-07 19:48:00 CET

- reply

I think game engine makers should make a UI tool, Unity
Or having something like this for UI would be great https://www.notch.one/

2026-03-07 19:45:31 CET

- reply

I wonder who will lead the way for the visualist camp for design tooling. I don't think figma will hold the ground, they have gone public, got to keep the share price high.
Figma's incentive structure post-IPO almost guarantees they chase the verbalist market. They'll go the cursor route.

2026-03-07 19:28:10 CET

- reply

Prompting forces experts to regress. It demands explicit articulation. It forces a master surgeon to describe every cut to a robot that has never held a scalpel. You are forced to convert your high-speed procedural intuition into slow, clunky declarative sentences.
This is why prompting can feel "cognitively diminishing" to a seasoned creative.

2026-03-07 19:27:25 CET

- reply

LLM interfaces currently privilege verbalists. But design as a discipline historically privileges visualists.
A huge portion of design expertise is tacit knowledge, intuition gained through years of procedural manipulation. You know a layout "feels wrong" before you can articulate why. You fix it by nudging a pixel, not by writing a spec.

2026-03-07 19:24:40 CET

- reply

The "Verbalist" bottleneck: Language is linear (one word after another), but design is parallel (everything is seen at once).
Forcing a parallel process through a linear pipe (the textbox) is inherently inefficient.

2026-03-07 19:23:37 CET

- reply

Prompting (Verbalist): This is a generative loop. You must construct the entire state of the world inside your head before you can output it. You have to "simulate" the design in your mind’s eye to write the prompt. This creates a massive cognitive load. If the output is wrong, you have to mentally parse the image, translate it back into words, and retry.

2026-03-07 19:23:30 CET

- reply

Canvas (Visualist): The interface acts as an external hard drive for working memory. When you lay out three boxes on a canvas, you don't need to remember their relationship, hierarchy, or color. You "offload" that data to the screen. This frees up the PFC to make high-level decisions. You look, you see, you tweak. It is a reactive loop.

2026-03-07 19:15:41 CET

- reply

Visual manipulation vs linguistic specification. The most powerful interfaces let people think in both modes without friction.
LLM interfaces currently privilege verbalists. But design as a discipline historically privileges visualists.

2026-03-07 19:09:53 CET

- reply

I think the canvas vs prompter camp in design tooling should be referred as visualists vs verbalists. People who are sick of words vs people who are okay with words
Visualists : People who prefer thinking through spatial manipulation.
Verbalists : People comfortable with symbolic and linguistic reasoning.
Object Manipulators vs Object Describers

2026-03-07 19:00:26 CET

- reply

Which means we need new direct manipulation interfaces for LLMs that don't require us to type words but let us explore ideas via moving stuff around

2026-03-07 18:54:11 CET

- reply

Canvas work leans heavily on procedural knowledge. Prompting forces declarative knowledge.
A canvas provides spatial anchors. So the brain doesn’t have to hold everything in working memory. The environment becomes part of the thinking process.
Prompting requires you to keep the design in your head. Working memory has limits.
So prompting can feel like: Describing a sculpture without touching clay, composing music without an instrument. You operate in description space, not manipulation space

2026-03-07 18:51:28 CET

RE: https://mastodon.social/@impactology/115203268904720219
Design workflow using a canvas tool like figma vs GenAI tool seems like using different parts of your brain
Cerebellum vs PFC : Clicking on a canvas using your visual-spatial intelligence and muscle memory that comes with it vs typing everything explicitly via words in a textbox
Embodied Visuospatial Cognition vs Linguistic Executive Cognition
Procedural knowledge (know-how) vs Declarative knowledge (know-what)
#note13r5…3jzs

2026-02-07 20:07:55 CET

Incident 1313: Anthropic Claude AI Agent Reportedly Caused Financial Losses While Operating Office Vending Machine at Wall Street Journal Headquarters
https://incidentdatabase.ai/cite/1310
We Let AI Run Our Office Vending Machine. It Lost Hundreds of Dollars | WSJ
https://archive.is/20260124205026/https://www.wsj.com/tech/ai/anthropic-claude-ai-vending-machine-agent-
Anthropic’s Claude ran a snack operation in the WSJ newsroom. It gave away a free PlayStation, ordered a live fish

2026-01-29 16:16:39 CET

- reply

>In the worst case, it constantly gaslights the human, presenting choice where there aren't any, simulating control while actually preventing it.
Very very true, one concrete example is a support option in an app where you can't write an issue of your own or talk to a person because they already have pre-decided options in a chatbot

2026-01-29 15:34:19 CET

RE: https://mastodon.social/@impactology/115945876113755071
You know I always thought of interface as having control over a machine
What human choices can we create controls for in LLMs
#note1xkz…d0wq

2026-01-26 12:42:31 CET

- reply

https://files.mastodon.social/media_attachments/files/115/961/214/412/412/504/original/00d03eaaccf8fcde.png
https://files.mastodon.social/media_attachments/files/115/961/214/767/048/130/original/e6d8109b27424c4f.png

2026-01-26 12:41:16 CET

- reply

HCBI is a design framework for building software that remains profitable while keeping human cognition stable under speed, scale, and AI assistance.
https://files.mastodon.social/media_attachments/files/115/961/204/881/364/450/original/7885daf31eb21694.png
https://files.mastodon.social/media_attachments/files/115/961/207/757/771/869/original/2536354050e40e5e.png
https://files.mastodon.social/media_attachments/files/115/961/208/225/582/473/original/0cc269a614cdfaa4.png
https://files.mastodon.social/media_attachments/files/115/961/208/636/888/158/original/1c0d1693b12a18b4.png

2026-01-26 12:38:00 CET

An update to this
How does a designer hold the entire system in working memory while making a local decision?
While thinking about this a while ago I thought of a 4x4 grid where you could think and communicate the human, business, interaction and computer (engineering) aspects in your working memory while you are going through any workflow
A conceptual pallete of sorts for a designer
https://mastodon.social/@impactology/114942438763021042
https://files.mastodon.social/media_attachments/files/115/961/195/220/959/554/original/f20a043807cfd1a7.png

2026-01-23 19:41:06 CET

Interfaces are all around us, not just on a computer screen.
Imho any tool or machine where a human being can control its function or its output, or has any way to pour human intent and choice into what it produces, is an interface.
A steering wheel is an interface.
A pen is an interface.
A guitar's fretboard is one of the best examples of an interface, and one of the most versatile ones.

2026-01-19 20:32:26 CET

- reply

Great examples, journalists can do all this because it's their profession. They get paid to do it.
Any other person after their job, commute, family chores barely has enough time
Just one hour before or after dinner and weekends, and if you're in tech that means spending time upskilling

2026-01-19 20:08:19 CET

- reply

Or fighting the agencies via suing them?
https://ejgreenbook.com/resources/citizen-suits/
https://www.epa.gov/ogc/notices-intent-sue-us-environmental-protection-agency-epa
https://www.fec.gov/legal-resources/enforcement/complaints-process/how-to-file-complaint-with-fec/
or sending FOIA requests?
https://www.foia.gov/faq.html

2026-01-19 19:57:07 CET

- reply

Do you have direct influence over federal agency officials working in these departments to implement the above?
Internal Revenue Service
Department of Justice
Environmental Protection Agency
Federal Election Commission
Securities and Exchange Commission

2026-01-19 19:53:29 CET

- reply

Don't you have to be a legislator to do all these, what can the average 9-5 joe do for actually implementing these

2026-01-19 19:39:15 CET

- reply

Okay how to address, could you suggest some actionable steps

2026-01-19 19:12:07 CET

- reply

We can't, not in our power. It's futile to waste energy on things not in our control
Unless you're running for office like mamdani or trying to be a bureaucrat trying to change things from the inside

2026-01-19 18:04:00 CET

- reply

RE: https://c.im/@meltedcheese/115922700208032964
@npub1jqm…65zu @npub174q…ckk8
Nicole look at the cool project David made for diagnosing spacecraft in space!
https://mastodon.social/@npub1jqm…65zu/115922700239601446
Much more interesting to talk about that than billionaire's actions who we have no control over
#note16tw…lypk

2026-01-19 17:30:32 CET

- reply

That is so cool! and wow its now being used for "Under consideration by a medical company for real-time diagnosis of rectal colon cancer."

2026-01-19 04:38:42 CET

- reply

That's amazing you invented an entirely new solution! Can you share more details on that, what was the problem & how did you arrive at the solution?

2026-01-19 04:34:39 CET

- reply

This is amazing feedback, thank you!

2026-01-18 12:22:01 CET

- reply

Manufacturing, logistics, aviation, food safety and finance right? For example Import-export trade compliance software for shipping?
https://mastodon.social/@impactology/115912577302179967
Or any of these domains below?
#note1zre…tfux

2026-01-18 12:06:03 CET

- reply

By then the damage will be done, people can't stop their lives and wait for that. You have to live and adapt
Already recruiters and hiring mangers are clueless on evaluating skills, its going to get even worse when they insist on evaluating AI tool use. Ones who already are senior level don't have to worry because they already have a track record and prestige, ones that are entering the field now are scrambling

2026-01-18 11:56:12 CET

- reply

Its here to stay as long as VCs are okay with burning their cash on it, seems like they have a lot to be okay with running on loss making AI products
https://www.a16z.news/p/we-raised-15b-why
https://www.cnbc.com/2026/01/07/anthropic-funding-term-sheet-valuation.html
They have a comfortable runway to change workplace habits, practices and hiring for the next five years.
https://files.mastodon.social/media_attachments/files/115/915/732/985/346/181/original/9d6e3127e9a1258c.png
https://files.mastodon.social/media_attachments/files/115/915/739/378/368/479/original/e4be5b9e78e23c26.png

2026-01-18 11:52:39 CET

- reply

For businesses its really simple to judge its value : if it saves more money than it costs to use it, otherwise dump it

2026-01-18 11:34:53 CET

- reply

Agreed! GenZ especially needs to learn this, very foul mouthed they are becoming

2026-01-18 11:17:47 CET

- reply

basically no point in crying over spilled milk

2026-01-18 11:15:11 CET

- reply

Sure, so what can be done about it? Assume the technology is here to stay, and its use becomes ubiquitous even in job tasks, workflows and duties to get paid how should one adapt to its use?
Its a crutch, spicy autocomplete, its bad yes, it weakens us, atrophies our cognitive abilities okay and yet its forced upon us so we have to figure out a way to use it that doesn't weaken but strengthen us.

2026-01-18 11:05:16 CET

- reply

Yeah I agree, what do you think is the solution forward if AI is here to stay and it becomes ubiquitous
What do you think should be that small unit of practice one can repeat daily to get better at even with AI use, that will always hold value and amplify our expertise rather than decrease it

2026-01-18 10:36:59 CET

What We Misunderstand About Pre-Industrial Craftsmanship by Vic Tesolin
https://victesolin.substack.com/p/what-we-misunderstand-about-pre-industrial
"Craft has always been about problem solving. Tools are just solutions with handles"
"The problem starts when personal preference gets dressed up as moral superiority"
From a woodworker
https://files.mastodon.social/media_attachments/files/115/915/421/597/526/131/original/08768463c36693b1.png
https://files.mastodon.social/media_attachments/files/115/915/421/953/796/449/original/4a33fb5e0060f5b9.png

2026-01-17 22:32:46 CET

Niche software categories based on HS codes (an import export classification system by world customs organization for almost every product category in the world) and possible errors software fixes for each
Which HCI program teaches their students to design software for such software genres?
To design decision points in interfaces that are most likely to cause costly errors
https://files.mastodon.social/media_attachments/files/115/912/565/738/924/931/original/4135372926c6fa39.png
https://files.mastodon.social/media_attachments/files/115/912/566/779/745/271/original/47e66f1a5afbefd8.png
https://files.mastodon.social/media_attachments/files/115/912/567/838/633/302/original/4f3b36d63cd19e28.png
https://files.mastodon.social/media_attachments/files/115/912/569/023/934/055/original/b5395d79627f87d2.png

2025-12-27 13:40:52 CET

- reply

The fifth reflex is reducing state, conceptual simplicity
Too many statuses mean no one understands the system. You merge and cut until only a few meaningful states remain
Do this
1. List all states
2. Merge/delete until ≤3–5 remain
3. Redraw transitions clearly

2025-12-27 13:38:44 CET

- reply

The fourth reflex is designing failure first, designing for error & recovery
You think through how things break, what can recover, and what can’t. You design feedback for those moments before you design success.
Do this
1. List 5 ways this can fail
2. Classify:
hard stop
soft recovery
3. Design failure feedback
4. Only then design success state

2025-12-27 13:37:18 CET

- reply

The third reflex is choosing one primary action, hierarchy & decision clarity
If a screen feels noisy, it’s because nothing is in charge.
You pick one action to lead and demote everything else.
Do this
1. Choose one primary action
2. Downgrade all others:
filled → outline → text
3. Annotate visual order:
first / second / third glance

2025-12-27 13:36:24 CET

- reply

The second reflex is structure before interface.
When things feel messy, you list concepts, group them into a small number of buckets, name them plainly, and delete a couple
Do this
1. List everything (concepts, not screens)
2. Group into ≤6 conceptual groups
3. Name each group in plain language
4. Delete 2 items (justify structurally)
If a non-designer can’t explain the structure back to you, it isn’t real yet.

2025-12-27 13:35:41 CET

- reply

The first reflex is framing before features, Problem Shaping
Any time someone asks for a feature or a screen tweak, you pause and write multiple ways the problem could be framed.
Do this
1. Write 3 competing problem framings (2 lines max each)
- coordination problem
- information/structure problem
- effort/behavior problem
2. Kill one framing
- write why it is structurally weak

2025-12-27 13:34:48 CET

I’ve been trying to distill what actually survives once real work begins.
Not what sounds smart in theory, not what looks good in portfolios but the design reflexes that still fire when you’re tired, under pressure

2025-11-09 11:56:33 CET

Got radicalized by teachers teaching to the test for memorization based exams. Got me researching about project based learning, open book assessments, open education
Got deradicalized again by realizing those exams are there for a reason : easiest to scale for millions
I think almost every single radical opinion I've had has undergone change when thinking about how it would apply and deliver outcomes predictably at scale in a routinized 9-5 workflow for tired, short on attention span people

2025-10-18 11:19:35 CEST

- reply

Environmental Touchpoints
Air, water, and noise are tolerable.
Public transport is reliable; walking feels safe.Energy systems don’t punish the poor for living.
Welfare outcome: Dignified coexistence with the environment, not survival despite it.

2025-10-18 11:19:02 CEST

- reply

Civic & Social Belonging
You can voice a complaint or idea without fear or humiliation. Public spaces are safe, clean, and co-owned.
Institutions respond transparently and predictably.
Welfare outcome: Trust loops between citizen and system.

2025-10-18 11:18:25 CEST

- reply

Learning & Agency
Learning doesn’t end with school; access is open, modular, and flexible. Skills are recognized by what you can do, not where you studied. Curiosity isn’t penalized by cost or prestige filters.
Welfare outcome: Distributed capability; lifelong autonomy in shaping your future.

2025-10-18 11:18:00 CEST

- reply

Health & Mind
When you feel sick or anxious, you can see someone without financial dread. Medicine, counseling, or rest doesn’t require bureaucratic negotiation.
Preventive habits (sleep, diet, exercise) are supported by the built environment.
Welfare outcome: Predictable care, not catastrophic care. Reduced allostatic load (chronic stress burden).

2025-10-18 11:17:32 CEST

- reply

Work & Livelihood
You can find meaningful, decently paid work without insider access. Skill learning and job transitions feel achievable. You’re not terrified of falling behind if you take a break or get sick.
Welfare outcome: Economic security + skill mobility. “My effort changes my situation” becomes true for most people.

2025-10-18 11:16:23 CEST

- reply

Food and Nutrition
Grocery prices are stable; nutritious food is nearby.
You can cook or eat without debt or dependence. Children’s meals are not skipped for budget reasons.
Welfare outcome: Adequate calories + nutrition as a background condition of life, not a luxury.

2025-10-18 11:15:56 CEST

- reply

Housing
Housing stability, predictable living costs, environmental safety.
The feeling of “I can plan tomorrow because today is secure.”
You wake up without worrying about eviction, leaks, or power cuts. Rent, repairs, and energy costs are predictable. If something breaks, help is accessible and affordable.

2025-10-18 11:14:56 CEST

When you think about it what exactly is the end goal of welfare, of social good, can we describe it in concrete mundane routines and touchpoints and outcomes?
If we strip the moralism away, the end goal of welfare can be described as:
A society where the average person’s daily routines contain fewer avoidable frictions, fears, and dependencies and more reliable pathways to self-directed stability, dignity, and contribution.

2025-09-14 17:06:19 CEST

The Search for True Numbers of Neurons and Glial Cells in the Human Brain: A Review of 150 Years of Cell Counting
https://pmc.ncbi.nlm.nih.gov/articles/PMC5063692/
Cerebellum which accounts for only about 10% of the brain's volume, contains an estimated 50 to 70 billion neurons while cerebral cortex comprising about 80-85% of the brain's total volume is estimated to contain between 10 to 20 billion neurons!

2025-09-13 13:51:01 CEST

- reply

One idea
Creating an Error Topology: mapping types of wrong models (over-generalization, wrong context, misapplied steps, etc.) from the get go and from student's mistakes
Basically mapping a variety of learner's mistakes, contrast it with the expert model → show the delta → guide the shift.

2025-09-13 13:17:46 CEST

Scale destroys everything that is meant to be sacred in learning and teaching.
I don't think you should educate hundreds and thousands at a time via the same model and methods
Instead, teach groups of 5 connected in a federated network with 100 other groups of 5
Federated microschooling, microcollege
https://standtogether.org/stories/education/microschooling-is-a-growing-movement-heres-how-we-can-speed-it-up
https://thoreaucollege.org/microcollege-movement/

2025-09-04 16:43:04 CEST

- reply

TIL about phase precession
https://en.wikipedia.org/wiki/Phase_precession
"Phase precession is a neurophysiological process in which the time of firing of action potentials by individual neurons occurs progressively earlier in relation to the phase of the local field potential oscillation with each successive cycle"
Basically the answer to how brain encodes "when" to fire what neuron in what sequence

2025-09-04 15:33:17 CEST

- reply

Out of millions of hippocampal neurons, how does it know which ones to fire for any given word/concept
How is the sequence of neural firing decided. If the same chemicals (glutamate, Ca²⁺) are used everywhere, then what gives each word its unique “order” in firing?

2025-09-04 15:33:03 CEST

How does the brain turn a very specific word like “linguistics”), which has shape (letters), sound (phoneme pattern), and meaning (semantic network), into specific chemical activity in the hippocampus and then later decompress those same chemicals (glutamate, Ca²⁺, CREB proteins) back into that exact word?
How does the brain “know” that THIS exact word = THIS exact chemical arrangement?
Out of millions of hippocampal neurons, how does it know which ones to fire for any given word/concept

2025-07-07 15:03:40 CEST

‘Came to work or enjoy with your boyfriend?’: Bengaluru intern recounts shocking sexist remarks by startup CEO | The Economic Times
https://economictimes.indiatimes.com/news/new-updates/came-to-work-or-enjoy-with-your-boyfriend-bengaluru-intern-recounts-shocking-sexist-remarks-by-startup-ceo/articleshow/122026352.cms
A UX designer shared her disturbing experience at a Bengaluru startup. CEO publicly humiliated her & other female employees
We need a workplace whistleblower platform to out toxic asshole Indian CEOs and managers
Or some kind of anti-workplace harassment job board platform

2025-06-20 12:29:23 CEST

Does classism exist in queer spaces?
Like do poor queers get treated differently from the filthy rich ones?

2025-06-06 15:40:29 CEST

- reply

"No matter how much mathematics you know now, once upon a time, some of the very same mathematics you are teaching was hard for you. Try to remember what it was like when the problems were hard and get in your students’ shoes"

2025-06-06 15:39:52 CEST

- reply

"This doesn’t mean that we should come to our classes unprepared, but rather, we should be prepared to model an authentic inquiry process for our students"
"It can be very valuable for us to let our students see how we struggle through a problem, get stuck, and get unstuck, because these are all normal parts of the work"

2025-06-06 15:38:16 CEST

- reply

"Mathematics in the back is the place where mathematicians engage in authentic practices. It doesn’t say a lot about you as a mathematician to know that you can memorize the strategy for every related rates problem in a calculus textbook"
"Solving a problem when you already know the answer isn’t really problem solving. It’s just factual recall"
This is exactly the extent to which JEE math tests for

2025-06-06 15:35:12 CEST

- reply

"It’s full of exploration, conjecture, and drawing pictures. If students are to ever have any hope of producing the neatly packaged mathematics that we see in the front, they are going to have to spend a lot of time getting messy working through mathematics in the back"

2025-06-06 15:34:18 CEST

- reply

"The “front” of mathematics is a well-organized set of theorems and proofs that are neatly arranged to present mathematical theory. This is what most students encounter in school. But much of the real excitement happens with the “back” of mathematics"
https://files.mastodon.social/media_attachments/files/114/636/674/325/923/790/original/2e497b19f38889f0.png

2025-06-06 15:29:24 CEST

- reply

"Drawing from this concept and adding a few ideas from my own experience, I offer the following features of tasks that are effective at promoting collaboration and engagement (in math learning)"
https://files.mastodon.social/media_attachments/files/114/636/655/376/365/005/original/85395878ed2640ab.png