27/Jun/2026
I picked up Anil Ananthaswamy's Why Machines Learn in December after listening to him on a couple of podcasts. While I abandoned it midway through the Probability chapter in a few weeks, it led me to 3Blue1Brown and Prof. Kilian Winberger's Machine Learning course.
Two terrific pieces of insight so far: a. Traditional computing takes in data and a program and spits out an output; ML takes in data and output (training data) and spits out a program. b. Maths is less a pristine, pure discipline that's the necessary language for pursuing the highest echelons of philosophy and more like any other scientific endeavour that's part engineering, part experimentation, part convention, and part theorisation that is very human in its fallibilities and inspirations, its history filled with deadends, bad bets, unexpected results, and inspired uses. In Imre Lakatos' framing, a scientific discipline that develops because of its proofs and refutations.
Also, thanks to Grant Sanderson and conversations with ChatGPT, I now have a better sense of the relationship between Algebra and Geometry, and how they can be used interchangably to both gain a better understanding of the model and be transformed to reveal newer patterns - William Thurston's "Mathematics is not about numbers, equations, computations, or algorithms: it is about understanding." comes to mind.
Anyway, I picked the book again a couple of days ago and I must admit I think I understand the Perceptron Convergence Theorem a little better now.
--
24/Jun/2026
-I will write once my restless, seeking, procrastinating, obsessive, needy, exhausting mind gives me some respite -You will not get respite until you write
I've been at the going-to-blog-now zone for the last few weeks. Its funny that now when I sit down to write I can't remember any of the epiphanic moments that almost-prompted me. A muddle mind can't write but it can't be unmuddled until writes. If you aren't engaged in new, interesting happenings you'll have nothing interesting to report but if you're constantly chasing every eye-catching event, how will you find the thehraav to truly think and write.
I've been working with/reading around/about LLMs on and off for the past few months, and one of the repercussions of the text-on-a-tap quality is me questioning if all my blogging of the last ~2 decades(!) is just so shallow that it can be so easily replicated by mechanistic computation. While there exists a part of me that really wants me to write and publish in more conventional methods, when I'm actually at-peace and unagitated, I think I realise that I will never be anything other than a blogger. Whatever the reason could be, either my lack of talent or my inability to work hard, this feels natural, it feels relaxing, and, most importantly, this feels liberating. While I have supremely romantic notions of being a writer/ filmmaker still, the intelligent part of me realises that they're basically advertisements I project for myself; that definitely is not the reality even if I somehow managed to gain both the talent and the work ethic to pursue it seriously. Having said that, what I do here too doesn't feel like creation.
Which brings me back to the LLM-thing. When I look at myself as I blog, I see that I took am kind of doing next token prediction only- Its literally Anne Lamott's 'One word after another'. And then there's a path dependence created that leads me further down an argument. As haphazard as entries on this blog may feel, they are infact the saner, justifiable sequences that slowly float up from the darker, wilder mindspace. The question naturally then becomes how am I different to an LLM then- because I know there's more in here than there.
Previously, I suggested that our Kahnemanian System 1 is more or less an LLM, and its the System 2 that is responsible for our 'reminder humanism'- the phrase comes from Leif Weatherby who's writings have had an impact on me. Prof. Alison Gopnik et al's "AI is a Cultural Technology" was the first heuristic that helped me look through the AI panic-hype and consider them as a continuation of Google- the next step in our information-management capacity as a species. Now Weatherby's "LLMs capture the poetics of a culture" is one like that. He says that what LLMs do it not replicate human cognition but act as a mirror to our collective ramblings.
--
31/May/2026
It's fucked.
రసెల్ తాత ఏడ్నో అడిగిండు, "ఏ సెట్టుకీ బిలాంగ్ కాని సెట్ యే సెట్టుకి బిలాంగై"తదన్నట్టు, నాకులా తప్ప ఎవరిలానూ ఉండననే మనిషి ఎలా ఉండాలో తెలీక ఎందరో అయోమయంగాళ్ళల్లా ఉండిపోతాడా? All unhappy families maybe different but are all unhappy men similar? ఈడ ఒక పోస్ట్ రాసే సమయం ఆసన్నమైందని నా బుర్ర నాకు సిగ్నల్ ఇచ్చినప్పుడు జర సేపు తటపటయిస్తా, మారాం చేస్తా, పన్నాగాలేస్తా, ఇంత మహత్తరమైన ఆలొచన ఈడ వేస్ట్ అయిపోతాందని విలపిస్తా, సరే ఏదో రూపంల బైటికొస్తది కదాని సర్దిచెప్పుకుంటా, ఆ బొక్కలే ఒస్తేంది రాసెటోడికి సద్వెటోడికి ఎట్లైనా ఫరక్ పడదని చవక వైరాగ్యం చెప్పుకుంటా, ఈ అనంత విశ్వంల నేనెంత నా బోడి ఆలోచనలెంత అని ఆలోచిస్తనే ఈ జగత్తు అనంతమైనదని నా అన్ఫెల్ట్ అసంప్షనే reflexive modern shtick అని చిన్న పాటి ఎపిఫనీ గుట్టెక్కుతూ అసలు రాయాల్సింది గదికాదు గిది అని మెంటల్ నోట్ పెట్టుకొని ఎప్పుడో సాయంత్రమో రాత్రో గీ ముచ్చటంతా రాద్దమని కూసుంటే మొదటి లైనే అస్సలన్కోంది ఊడిపడ్తది. ఇగ ఆడికెల్లి అంతా సట్కో సీతారాం.
గీ బ్లాగేంది నేనేంది మీకీ మారని ముగియని ఘోషేంది? నాకు జ్ఞానోదయం కల్గదు, కానీ కలిగేలా ఉందని అదీడ రసుకోకుంటే అదైనప్పుడు నేను అయిందని మర్చిపోతానని ప్రతీ కొద్ది వారాలకు నాకు తోచ్తా ఉంటది. ప్రతీసారి false dawnఏ. లేక సూర్యకాంతి జాగాహ్లో మెరిసేవి రెండు మూడు మిణుగురులే. మల్లా అదే సైకిల్.
మొన్నక రోజు బుర్ర మస్తు పరేషాన్ చేస్తాంటే ఉండబట్టలేక ఇగ చాట్జీపీటీని అడిగేస్నా. ఆదీఅన్ప్లగ్డ్ రచయిత గురించి నీ అభిప్రాయం ఏంది, వాడు గెసుంటి పనికి పనికొస్తడు, వాడి పర్సనాలిటీ ఏందని. గదేదో చెప్పింది. నిజమా కాదా కాదు అడగవల్సిన ప్రశ్న, పనికొస్తదా లేదా అని. రాదనే నాకనిపిస్తాంది. LLMతో రాయించిన ప్రతీ ముచ్చటలాగనే సద్వుతున్నప్పుడే నా కళ్ళు glaze అయిపోయినై. ఒక వైపు నాకు వాటిపట్ల కంపరం, ఇంకో పక్క ఉడుకుమోత్తనం. నా అనుభవంలో AI writing slopఏ. ఇన్నేళ్ళుగా ఎన్నో రకాల అంతర్జాల వ్యాసాలు సదివే నాకు కొన్ని నచ్చేవి, కొన్ని బోర్ కొట్టేవి, మరి కొన్ని ఒళ్ళు పులకింప చేసేవి. కానీ ఈ మధ్య కాలంలో ఆ టెక్స్ట్ని చూస్తే కంపరం పుడుతోంది- uncanny valley feels. ఈ వైనం ఎక్కువ శాతం సబ్స్టాక్ పోస్టుల్లో కొట్టొచ్చినట్టు తెలుస్తోంది. కోపమెందుకంటే నాకనిపించేది- మనిషితనానికి భాషకి ఒక అవినాభావ సంబంధం ఉందని. కానీ ఇవి textని commodify చేసి పారేసాక ఇప్పుడు చచ్చినట్టు ఇంకేదో ఎసెన్ష్యలిస్ట్ ద్రవాన్ని వెతుక్కోవాలి. My instinct says we are not just LLMs, so I need to look for my new elan vital.
Can it be called a digression if its the opening? Because I wanted to write about how exhausted I've been feeling. No this isn't about Noori and fatherhood although thank you for asking, and yes its going great, although the poor little thing has been sick on and off for the past 3-4 weeks. I'm referring to that existential malaise, the one that has been making Sravani say recently that I should perhaps see a shrink. Maybe I should even though I know I'll spend the first few weeks flitting there between euphoria, despondence, and a pig-headed insistence on finding my own light my own way. Which way to go? Does one retreat further into one's own private language until they hit the bedrock of their particular self (or maybe the underground river of a common humanity) at the risk of becoming more and more solipsistic, or does one force oneself to look out and learn to interact with others in the hope that salvation lies outside and with others? While my behaviour has been veering towards the former, my instinct tell me the right path is the latter. But its become so, so hard to connect with other people.
--
18/May/2026
As is usually the case, it starts in the here and the now.
Why do I live the way I do? I was born in erstwhile Andhra Pradesh into a Telugu-speaking family and yet English is my first language. I was born into a Brahmin family and yet my religion is secular liberalism. Some of my ancestors were priests and yet I hardly know a word of the Vedas. More broadly I live in a world that is Capitalistic, Computer-saturated, deeply shaped by Climate Change- what are the epistemologies that led to this? I'm sailing a very small boat in the very large currents of history, navigating my destiny as best as I can but tossed by winds way beyond my control and comprehension. What I'm setting out to do here is to understand my wider reality a little better. And the way to do that is to ask What If; How else could history have transpired?
This is primarily an act of imagination. The central question I'm chasing is this: If I'd still have been born when and where I was born but something else had happened slightly differently at some point in history, how different would my life have been. The spectrum of possible configuration changes can range from the extremely personal- someone in my immediate family deciding to do something different at a critical juncture- to the ginormously universal- what if the initial setup of the big bang had been slightly different. I'm only limited by my imagination, education, and, most importantly, interest. What adjacent possible am I most keen to explore? What alternative setup would shed the most light on my daily life and make me see its contingencies and inevitabilities in stark contrast? What would make me better understand my worldview, my moral bearings, my aspirations and ideologies, deepest desires and their contradictions, socially laudable actions and transgressions?
My intuiton is that while we have free will, we are nonetheless highly constrained by the geographical, technological, social, cultural, economic, biological etc. boundaries of the immediate world we inhabit. To be a truly free agent, as free as possible, is less an act of will (Schopenheur: You may do as you will but can you will as you will) and more a practice of being able to first imagine and then inhabit alternate ontologies. Partly to identify a normative self - place me in any context and these are things I'll never do - but also to develop a meta-understanding of possible options of action and a sense of why they're the available choices in the menu.
One prospective entry into this exercise that excites me is to bring the lens of the World Machines framework to the last thousand years of Telugu History. They both dovetail beautifully because Telugu literary history roughly goes back to the period - from Nannayya's embarking of the transcreation of the Mahabhrata in 1050 CE - and I want to use literary history as the throughline to understand the changing political, technological, economic, social landscape than eventually brought me here.
Why Telugu history - because I was born into it and feel like I should know more about it; I have some advantage in the area because of my ability (limited but nevertheless present) with the language and being able to access primary sources; And most importantly because it is popular enough to interest a large number of people, big enough to act like cogent, impactful player in World-Systems Analysis, and is old enough that it has its roots sunk deep into larger Indic thought to truly envisage other modernities.
In terms of approach I don't want History to read like "Just one damned thing after another". It should chafe against the present- rupture, rhyme, resonate, reflect. Show us the Could'ves and the Should'ves. What was essential and what was contingent about the human condition. And I want to go deep into the assumption: What qualifies as Telugu? Is it still a relevant category? What is the normative definition of Telugu-ness?- Genetic, Linguistic, Regional, Mythological? How do you create models from the raw data stream of reality?
The undergirding philosophy will become more apparent as I proceed in the journey. But for now let's identify a place and time, and start the story.
--
01/Apr/2026
First we asked for Church-State Seperation - The story of Modernity I first heard stated that one of the fundamental aspects of transformation from Medieval was the state moving away from religious matters, and ensuring that the church became subservient to it. That led to separation of powers, and when woven with individual rights, freedom of speech, democracy, scientific temparament, and marketisation of most activities, became the latticework of the post WWII World Order most people in my socioeconomic nook have as our Stock OS. Having spent the last year on the margins of Contraptions History Club, I think I'm fairly convinced by Venkat's World Machine Mental Model, and think that that kind of 400-Year Connected History - not just spatially but also across disciplines like Culture, Technology, Economics, Political - thinking has proven to be quite useful. Its corollary that at any point in time we are in the intersection of Dawn-Day-Dusk also then applies here. We are all roughly here - some more, some less. It's been pretty amazing to realise that that old dictum about India being both in the 12th and 21st centuries at the same time is literally true. The most edifying insight of the World Machine Project has been the centrality of, well, education in creating a large enough mass of people all oriented towards one direction pushing with all their might to change the trajectory of the world. ChatGPT once pointed out to me that what people assume to be axiomatic, unshakeable truths is stuff they had to learn but then forgot they learnt it thereby giving them the impression that that belief was somehow apriori and thus inseparable from their conception of the self. Pair that with Keynes' defunct economist quote and Javed Saab's "99%of people die in the religion they're born in", Pankaj Mishra's unfulfillable capitalist dreams creating resentment, and algorithmic platforms' proclivities towards creating insulated groups, its not too hard to understand why culture is..