commerce > politics

i think my study of pleb politics is not quite as thorough as my study of pleb commerce. i'm not a foodie - stopped prioritising it while i was in college - but i ran a cafe for six years to make sure i had exposure to ordinary people. i almost never vote - but i'm not going to run for office, as i'm busy catching up on maths and other STEM shit i paused for two decades. but i am spending a bit of time to read and chat with ordinary people, and the propagandists who earn their livings in this realm. 

why do we use QKV and not fewer parameters?

Zoom out for a second. These are all just parameters - any attention block in any transformer layer can have an arbitrary number of parameters, wired up in information preserving ways.

Beause the overall transformer is trained iteratively, the system will store information somehow in whatever parameters are available without prejudice. If you give it one param, it will use it, if you give it ten, ditto. 

The net architecture is simply a decision of how much memory to allocate to internal nodes in a multinode NN. 

sex bots in Malaysia

conjecture : sex bots will be declared haram

more stuff for Malaysians to get political about in the future

---

Humour :

I don't need 72 virgins : the earth provides

I'd like to have 72 people who weren't apparently the sort of plebs I meet on earth every day though

Maybe this means, I go to hell 🙂

( this meme is inspiring me to target 72 as the team size for smart people needed at the centre of a sizeable organisation - lol )

tatacara kenegaraan : BM tidak cukup

Sebagai seorang yang ber PMR, SPM, saya rasa perlaksanaan ini tidak mencukupi. Tak boleh terus kata salah lah, nanti korang tak akan baca perkara seterusnya ... 

Dalam logik modal, adanya kerangka istilah, "perlu", "cukup", dan "mungkin". 

1. Adakah BM perlu, bagi perpaduan? Pendek kata, ya boleh diterima. 

2. Adakah BM cukup, bagi perpaduan? Semestinya tidak ... Jadi apakah unsur-unsur yang melengkapi penggunaan BM, supaya gabungan itu adalah cukup, bagi perpaduan? 

2.1. Pendek kata, latihan amali tatacara kenegaraan mesti lah ditambah kepada penggunaan BM.

Murid antara umur 10 dan 18 mesti menjadi fasih dengan tatacara :

- pembentukan dan pengubahan perlembagaan serta tiap undang di bawahnya

- pengelolaan jentera dewan undangan, kitar hidup wakil rakyat, dan hubungan seharian mereka dengan cabangan eksekutif, secara main-main dengan olok-olok kerajaan

- pengelolaan jentera perhakiman, dan hubungannya dengan cabangan kerajaan selainnya

- pengelolaan cabangan eksekutif dari PM, jemaah, dan badan-badan penguagkuasaan di bawahnya

- dengan peranan MRR sebagai cabangan keempat di Malaysia

3. Mungkin lah, matapelajaran yang tiada ini, LEBIH PENTING DARIPADA BAHASA MELAYU. Tapi perkara ini tidak penting. Yang penting ada lah, nak bagi tahu je ... BM ok, tapi tak cukup. Dan kita kehilangan satu kemahiran dalam HAMPIR SETIAP murid kita yang menjelang kedewasaan sebagai rakyat.

Negara itu lah jentera. Tak baik kalau bahagiannya tak fasih dengan fungsi jentera itu. Lari lah, peranan dan kendalian dirinya ...

reflection on a journey with poors

In the dreamy space before getting up today, I thought about my decisions to intentionally spend MORE time with people who were less economically privileged, from 2001 to 2023 ( ages 18 to 40 ).

I generally held the view that I did not understand them, and that it would be beneficial to live as poorly as possible ( intellectually, materially, emotionally, geopolitically, whatnot ), in order to learn about them, or at the very least to be able to say later that I had put significant time into making sure that I had tried to do so. 

Review questions :

1. Were my presumptions incorrect? What evidence is there?

2. Assuming correct presumptions, did I fail to achieve what I set out to do? What evidence is there?

3. Assuming correct presumptions, and success at what was attempted, was the project worthwhile? What evidence is there?

Discussion ( inconclusive ) :

- My presumption that I did not understand them was too risk averse - I already understood them quite well at 18. Nevertheless, it was worthwhile to check and make sure.

- Operationally, I made some costly mistakes. I would often write-off huge chunks of time, to assist specific individuals, on their whimsical directions. This would have been more valuable if I had apportioned smaller chunks of time to individuals, or the larger chunks to a larger body of stakeholders. In summary, I underdiverisified. 

- Was this chunk of two major phases, four years in academia, and 18 years in commerce, ultimately the best use of time? I really don't know. I don't ask much of life, so almost any results are welcome. 

I proceed with these thoughts in mind. 

embarassing reflections

I have some embarassing reflections to share, in my fatigued state, this week. 

1. Probably the one Malaysian value I picked up before even graduating from highschool, was laziness. I have been thinking about how to make the absolute minimum amount of money for most of my life, probably starting around the age of ten. Back then I thought maybe 500 RM/month would be enough. 

2. The people I do meet who want to make money, have disappointingly low notions of how much money to aim for. But I guess I have chosen my environment ... I began studying commerce in 2006, in Malaysia, a country whose notion of a unicorn company is one that remains 17 B USD in loss after 14 years in business.

2.1. Speaking of which, I have understandably mixed feelings about that particular unicorn. They are very nice people, and make lots of money as employees ... many yuppies I met in the 2000s, now work there. But I could never bring myself to take it seriously as a tech company - perhaps because I don't personally use such products at that price. (Male privilege and the 6 RM taxi ride to work, in 2005.) Plus it's not very high-tech, more of a service co, so off flavour for tech. Furthermore just thinking about Uber and extrapolating, I could never convince myself that the privatisation of public infrastructure would have any viable exit except back to the public sector in the very long run. Long, painful, cloudy, investment horizon. Doesn't matter much to wage-workers, I guess. Mattered a lot to me. Hm. Weird.

2.1.a. I'm not averse to lossy ventures at all. But I prefer to fail as small as possible. I think the only time I took outside money, I capped losses under 100 grand, after six years.

3. I find that I work best when I allocate large periods of time to give unwieldy projects a lot of attention. Smaller projects are not risky enough, and shorter timeframes seem too risky. Stupid personality quirk, I suppose. Pre-college was a decade. College was four years. Post-college was 20 years. Sabbatical is ten.

3.1. Corollary - I am not so practiced at dividing my attention. I can make money as a trader, but if I start studying something else at the same time, I lose money. If I'm doing well in business, I'm not learning anything, because predictability is orthogonal to learning.

4. I have been testing a restricted diet for science, for a few years. It feels efficient, but not for performance growth or hypertrophy. So I am pensive, in a bad way.

Ah well, on with the program for now. Today is day ... 1171, 32.08% in.

revision : toe and finger tip tracking

Latest run : proprioception of fingertips and toetips is always a good dataset for optimisation purposes. At the extremities, any instability is amplified. It turns out further that it sufficient to track digits 1-2-5 on each paw, reducing the number of tracked points to 12. This insight is interesting because it can be applied in moving systems of any kind, such as robots.

Proprioception of the head is a complementary concern, but not the main study today. Footstrike depends heavily on digits 1-2, while 5 is needed to check toe splay. For hands I favour digits 1-3-5, as it feels more balanced. 

apostasy in Malaysia

This seems to be raised in multiple posts by a Muslim-identified spokesperson, now as we ramp up towards elections, the parties are looking for stub issues to campaign on. As elections are announced, parties will consolidate messaging around key issues.

My disagreement on their interpretation is shown. 

"1. Tiada halangan dalam perlembagaan

2. MRR > MKI + ketua islam negeri > mufti : berhak mendirikan pandangan dan mengundangkannya

3. Kalau (2) bercanggah dengan hak asasi lain dalam perlembagaan, kes sivil berperlembagaan boleh dilangsungkan bagi menyekat (2) ikut perlembagaan ( ada duluan perhakiman dalam percanggahan jenis ini )

4. Sampai hari ini, saman jenis ini kurang berdokumen secara awam. Segala isu perkara ini, sudah lama disunyikan - tak nyata la, ia sekatan tersirat macam adab, atau tersurat macam OSA. 

Rakyat satu hari, wajib dapat tahu. Bagi kemuliaan negara Malaysia. 👍🏼" 

Elections : Interfaith Muhibbah

It is normal to be divided in politics, but it is (always) sad, to see Malaysians being fearful, angry, rude, and violent, about politics ... because some evil (maybe human) spirit has told them that this is a correct (bersopan) way to approach political (civil) life.

As a point of reflection, for non-Muslims (such as myself) and Muslims, perhaps we can all reflect on this frequently read Muslim text. Muslims are of course familiar with it, but it is less common knowledge to others.

Maybe this is a good reminder about how to engage with each other. Our responsibility to ourselves is to believe what we believe - and our responsibility to each other, is to encourage each other, to believe whatever they believe. 🙂

• Surah al-Kafirun (The Disbelievers)
• https://ms.wikipedia.org/wiki/Surah_Al-Kafirun

I think, the overall scope of discussion, is quite suitable for children also, on a daily basis. Hopefully if kids are raised to be more chill, they won't grow up into adults who are always yelling at each other.

Discussion :

2026-06-10 :

Part 1

Strategically, Malay centrists who seek political gain, need to have a structural narrative response to the Malay right's appeals to nativism and theism.

The Malay left is too tiny to make an active impact, given the laws and taboos against communism - whereas their position is favoured by the current phase of the global economic cycle where economic inequality is peaking.

I have detailed thought on each of these issues, but it is really just me hanging on the dividing wall, sipping a soda watching it play out in someone else's family. 🤣

Part 2

Earlier I posted an FMT op-ed by a Malay centrist, who had a view that there are many types of Malays. 

Q : Someone in the comments asked why the Malay right seeks to unify against non-Malay citizens.

A : One of the driving factors of course, are our civil society NGOs with roots in the 1970s Muslim Brotherhood movement, like ABIM and ISMA, which our dear PMX helped to form a strong identity base back when he was a young man. See EUBI, and all those historical events back then.

These are of course, grassroots clubs with a common identity. Sometimes we make fun of them by saying it is not clear if race is riding on religion, or if religion is riding on race - it actually doesn't matter, these are our citizens and it is altogether part of our national society. 

Whether we agree or not about how to run the country, we are still living together and must cooperate, to increase cultural exposure for our kids, so that everyone has a better understanding of everyone else ... 

... instead of turning into adults who have gotten stuck in the idea, "my culture is my culture, their culture is their culture, we do not talk or share these with each other" - these are the people of all races who reduce GDP 15% annually, compounded for decades, because people are worried about their neighbours instead of worrying about all the other countries coming to eat out lunch.

Tradeoffs

I always wonder.

Currently I study small details of nutrient contributions to daily productivity. Did I make an efficient gamble, by spending the first 20 years post-college, simply buying my way out of this concern? Hard to say. 

Similarly, did I make a good gamble by postponing maths study during this period?

How about the gamble of avoiding intimate relationships during college and the first few years of jobbing?

And how about the positive allocation of all that time to cultural studies of the Malaysian commercial environment, with hardly any cultural study of the Malays? Was it worthwhile?

Generally my approach, is that if I think X is not interesting, I will make extra time to double check it. All these gambles. 

Why (current) LLMs are fat, and dumb as shit

Consolidated notes over a few days, nothing really new, but the talking points were related, so here they are. Mainly response to he following propositions made by others :

❌ "We don't yet have the language to understand LLMs"
❌ "LLMs are bad at maths"
❌ "World models are timeless"

• 2026-06-06 0133
• "We don't yet have the language to understand LLMs"
• WTF is this nonsense - there are infinite opportunities to make language more complex. We have a perfectly precise language for describing what LLMs do ... it's the literal data, and algorithms operating on the data.
  
  If you say "it's not clear in English," well read the maths. If you say, "I don't understand the maths," then it's either a you problem, or a request for poetic expansion of English, which the LLMs are more than happy to to provide if you append "ELI5" to the name of the step in the demised computation 😂
• 2026-06-06 0809
• "LLMs are bad at maths"
• Just have ALUs do math. Use LLMs for guesswork. The dumbest shit architecture in the world making frontier models learn maths as if they are made of meat.
• 2026-06-07 1743
• "World models are timeless"
• You need to give an example of thinking that is time independent. All thinking is intrinsically time embedded
  
  Thinking= computation
  Thought objects = data
  
  Comp on data
  Operations on operands
  
  Concept of causality / implication depends on concept of time ( Good place to start )
  
  You're probably thinking about transformer archi as a way to parallelise the temporal dimension. It is not "independent".
• Let's go with a simple example.
  
  Say the input is a ball balanced on a pencil, standing upright on a flat surface, on a gravitational planet.
  
  Let's say the input is just a photo.
  
  Just to get from photo to physical model requires inference (comp) by the bot. Then more comp is needed to forecast the "next incremental state".
  
  There are two ways to do the comp.
  
  A : transpose the time dimension to space, which is what embedding layers + attention weights do : these contain a memory of past events, and associate the input data with similar past events.
  
  B : after that association is done, logic is performed, in the temporal dimension, at the FFN layer ( after the attention layer in the transformer ). Typically if you have a transformer doing logic, maths, physics, it is happening here.
  
  Because transformers are not Just A, but A+B, there is always a temporal i.e. logical computation step during inference. Only A is time agnostic. B is not.
  
  Currently LLMs are fat because they try to move more B to A.
  
  But as we all know, sometimes just a bit of B time will save a lot of A space, plus all the time needed to create A space in the training step.
  
  So it comes down to target workload. Fat A is not for everything.