day 15 8:58 pm, 24 October 2025 Well, its been a lot of days, but here's the SMMC update: I did 5(ish) problems. So i mean im not winning it, too bad. This is the question paper: These are my solutions paprer1 paper2 ok but yuh, paper 1. actually fucked up, i dont know lin alg, at all, i opened the papper and straight up saw three lin alg problems. i was like wtfffff. i was actually sad, and mad as fuck. I think this kind of messed me up cos if i stayed calm i could have probably done even more problems but wtv. So i mean problem 1 was just absolutely trivial. Just look at the max degree in max poly Problem 2 and problem 3 i thought about for some more time. Problem 3, i got a few ideas, some progress but not much. you know i really think the problem was just i was mad at why these niggas had to make a fucking lin alg question paper. wtv problem 3 i didnt do much Problem 2 after a while i got the idea, to double count by vector instead of matrices. And then i kind of got the main idea of you can basicallly choose most numbers randomly, and then the last number will be automatically fixed cos F_p is nice. The thing that was tripping me up was what if vector has a bunch of 0's. i mean i just had to be calm lmao. wtv i wrote up tehse problems with very less time left. problem 2 i made a small mistake when writing up, which is if k entries of the vector are zero, then the term you're supposed to get is $\binom{n}{k} \times p^{(n-k)(n)}$ but i made a small error. its ok i still consider essentially a solve. cool. Then i went for pancaks with sagnik. We were just talking and like man im really going to be great. anyways, paper 2 was much better. For problem 1, i couldnt really think of any good ideas in the beginning at all. i thought for a while, i knew it had to be size somehow. i was like damn if m could be negative it would be so ez. then i got the m=n! idea lmao i was feeling good problem 2 i thought about very surgically. i was like ok both triangles must start at one vertex. the i thought about the right angle case. made a few more observations in that diagram, then i got the idea that each triangle should have one side along the quadrilateral. then i thought about the extreme case and got the construction. pretty cool. Problem 3, i was like damn, i know 0 lin alg, dont even remember wtf a determinant is (never studied during JEE either lmao) but i still decided to think about it. I thought a little bit about the degree stuff, and then i was like hmm obv $\prod (z_i-z_j)$ should be a factor right. So it suffices to prove that $z_1z_2...z_n-1$ is also a factor. like i thought in factorisation terms. I sort of proved the last factor must be of the form $z_1z_2 \cdots z_n-t$ (tho my reasoning for this is kinda shaky lmao) and then to compute t, you can choose $z_i=\omega^i$ and use a double summation. kinda genius lmao. pretty cool stuff. i mean the fact is i know im one of the best in india at math. like thats a fact. and thats off like 0 work bruh. like this is actually kinda crazy, i might be a fucking genius lmao. i can actually do any fucking thing man, maximise your potential please. right now tho, ok the thing is, i kinda know what to do. just have fun all fucking day. that is i just need to not stop at all. just decide to not waste time at all, and then i can just do whatever i want all fucking day. this is going to be so much fun man lets go. really just dont think too much. i need to just do what i want to do and not stop at all and everything else works out. by the end of this calendar year, i should know i am the best ever. in 1 year, everyeone around me should know i am the best ever. in 2 years, i should be in YC and have dropped out. in 3 years, i should be THE whiz kid billionaire of this generation. lets go then. 24 OCTOBER 2025 yuh honestly man its time. this is my world. witness history. 1:48 am 25 October 2025 ETH online hackathon is going on, and i wanna build something really cool. simple enough (cos i can build anything lmao) but i just need an idea. lets go. "mony" on shuffle. JDUB in this bitch (ill explain at the end of the day) 2:38 am let the hacking begin. we aint gonna stop at all. 3:56 ah fuck im sick as fuck and sleepy as fuck. sleeping for a while. listen, dont negotiate w yourself. like this is true, do not negotiate with yourself. the world is yours if you work every single moment. 8:28 PM life can literally be a movie. this is my life. i can do whatever the fuck i want to. i'm just thinking of crypto stuff right now, i dont really care too much about the technicalities right now, i just cool ideas that excite me and we'll figure out how to build them later. 9:19 pm account abstraction seems cool as fuck. one click checkout is implemented already, sad. sprintcheckout. i really wanna build something fun. btw this is perfect, im so inspired right now. im going to learn a bunch of cool math, CS, build a bunch of fun projects. do everything man. like this is my world. go be the best there has ever been bruh. 10:20 i think i wanna build something that is just fun, kinda trolly ig... let's see. 10:54 ok apparently you need consistent commits, so i had to do a quick git revision. man by the end of this year i need to know im the best ever . 1:30 am 26 october 2025 my guy pretty like a girl woohoo man i was kinda bored but its just like just go have fun man wthelly you can do anything damnnnn woohoooooooooo. need a new workflow tho 1:41 am lets try out a new workflow... 2:13 am ok this workflow for generating ideas seems so much fun, some random ideas: yo! here's a few random ideas lol: 1.live betting on nba games, normal bets are of the form who wins, does luka score over 30 etc... this is going to be more small bets of the form who scores the next point? how is the next point scored? (or cricket games too lol) 2.wager system between friends; or random people online talk and are like i bet ill defeat oyu in chess and then they play with a wager. friends like ill defeat oyu in fortnite and they play with a wager system. 3. productivity app with money punishment.. i have deposited some money in the app, then i say i will commpletely study chapter 2 of diestel today, then at the end of the day, i will use an llm as a judge to test me on it, and see if i actually learnt it. if i didnt, the stake is lost. or i say im going to complete building this feature tonight, at the end of the night, again llm as a judge tests me this is perfect, youre gonna have fun and do cool stuff but for that, just do not stop at all. just keep going and you'll have more fun and yuh what the fuck nigga, literally just haev fun. the world is mine lmao. 2:53 am so basically what im doing is im kind of having a converstaion with chatgpt, we're bouncing ideas off each other. im going and looking at resources, i looked at a bunch of ERCs and EIPs bro btw UMA orcale is soooo fucking intersting damn... polymarket tooo... conflict resolution tech is fun the account abstaraction thing is genuinely a game changer for ETH bro... but yeah this workflow is fun, genuinely just keep going, just have fun and i am the best there has ever bene lmaoo. just have funnnnn woohooo literall this is the dream life, do this every single dayyyy : life and echo on loop... idea: productivity is cool when i can tell people im going to do something and then i do it. so a social app, where everyone has a profile and everyday they say a bunch of stuff like their goals etc. and then llm as a judge tests whether they accomplished those, maybe losers money goes to winners pool. orm maybe losers money helps pay for llm (or wait llm you can just use your own account llm right lol) and then you have a record of the whole day of each person, on the blockchain (and you can store also the llm chats) i mean ofc you can cheat also but wtv ignore that for now. then your whole life and proof of you achieving your goals is on the blockchain, kind of a social profile too, where you have a social profile (linking to twitter might be fun, cos on twitter people claim a bunch of stuff and they can have sort of blockchain proof of i accomplished this today). so you can look people up, see ok this guy is working hard achieving his goals daily, this guy is learning so much stuff daily wow i wish i had started this earlier i woulda had so much more fun and i coulda built this stuff ad learnt more stuff myself too but its all good, i got a lot of time. bro im telling you i can actually be the whiz kid billionaire, cos im good at math (im gonna get better) and be the best at math, and then im like kinda zucky too in that i get good ideas and i have balls. literally perfect combo. lets gooooo. i love chatgpt too btw lmao... dailyblogs every single dayyyy 3:15 am ok this dareboard idea i have is actually very fun lol... and the chess wager thing is there. looks easy to build... if i cant implement cool things, ill just do chess wager... 3:27 am i messages sahitya lol... ahh btw this is soooo much fun lets fucking go. btw the live betting; ala who scores the enxt point, is the next ball a homerun etc. can not work, cos latency issues with tv. ok thats fine.... lets work. we cutting it to the edge lesgoooo 4:17 am Ok I have two really cool ideas, the dare board thing, and the productivity day on blockchain thing... both seem very funnnn, itll really be tough to build this stuff but yeah, i need to think and decide which one i want.... 4:58 am yuh i just came back from a scooty ride, and during the whole ride, all i culd think abut was the producitivtiy idea... like this seems actually cool as fuck to build... like this is something i would actually use. im so fuckiing excited for this man lets gooooo. lets think more about this and start building now lmao. 5:17 am i just tested out if llm as a judge could work, and yeah it does work... 5:41 am yuh we're figuring out exact details now... 5:49 am ok judging code is kind of tough... maybe cos i havent coded yet... 6:21 am i thought a lot and i think i have come up with the basic and main features and sort of solved the main problems: this is going to be a fulll social website of sorts, each user has their own id and page (basically like instagram) STAKE: i dont want this to be like a money making platform, so the stake is more like a subscription for using the platform, say i take 2$ (per month say) from a person for using the platform, they just send me 2$ (no crypto needed here, they can give me 2$ howeveer they want) .. this 2$ will cover my ai judging and the gas cost (so im gonna be paying for their gas, called paymaster or soemthing right? so sleek account abstraction of sorts here) in return for their 2$, they can use the platform for one month. (so normal web2 type subscription here) COMMITMENT: on our website, they will make a commitment of the form: (number of goals, , duration) (goal vector is a set of goals say) or maybe for demo i'll support only one goal. USER JUDGING: any time before the duration is up, they can click on goal completed button and take a small quiz (described below)... JUDGING: im not overcomplicating this, im going to have llm as a judge, (i think we can use chatgpt api keys here?) im going to prompt the ai to take the goals as an input, and then i'll feed it some prompts for the type of questions it has to ask. it'll ask very simple questions, and be very forgiving. for coding tasks, itll ask for a little bit of code also maybe... very forgiving again, like assume goals completed till proved otherwise lmao... the output llm gives us will be a single word, pass/fail. {will it be possible to use api keys and make use llm as a judge? like will we be able to program this?} PASS/FAIL no dispute: then i think i wanna use the attestation feature right? so first off we will store the complete llm chat and the user can add comments/write small blogpost there also maybe... and we will store this on our servers (no crypto here).. the only thing we will upload to the blockchain will be an attestation of the form @monyapp attests "userid" completed "goalstruct" . so a very small thing is uploaded on the blockchain (the rest is stored normally on profile) Then we can upload this attestation on the blockchain. (how does this work? will we be able to program this for the demo?) (maybe in real case we can bundle per day, but for demo: i wanna be able to upload something (EAS attestation maybe) to the blockchain (testnet or wtv) to be able to show the judges, look proof of work on blockchain lol USER PROFILE: on the user profile, you can see a list of all their commitments and wheteher they passed/failed. for each of these, you can view 1) the blockchain proof and 2) the llm chat we stored and their comments ect. we can make the user profile pretty by adding streak ect also DISCOVER: in this you can see the list of all daily goals of all people that we have uploaded etc DISPUTE: ok this is i guess interesting... i intend the platform as a productivity website, so people shouldnt cheat (im assuming optimism) so like if they claim they have done something i should believe them , so an idea i thought about this is, people will all login/attach their twitter account also... so if the llm fails them, and if they wanna dispute, my resolution technique is, they will need to send out a tweet on their main account, that links to this commitment on our website, and they can additionally add stuff like llm failed me but i had done this task actually (or wtv they wanna write)... i will just store this tweet on their user profile and PASS them then. the rest of the pass thing is same, they'll still get the @monyapp attestation. (the reason being, people prolly wont lie on their main twitter account about doing a task, when people can just go read the llm chat on our website and like fake tweet is a big deal prolly for them. 7:54 am ok perfect. lets start building. im actually excited as fuck for this. write now im mostly programming in english lmao, i mean this is still like bro i dont really care THAT much about the exact tech. being used, i just wanna build stuff man lmao... but yeah ill get better at that stuff too, for now, lets just do this and hope it works... woohoo... im not gonna stop, im telling you man. the way we're building this is, using mock keys... we're first building out the site and putting it up on vercel , and then we'll add in the ai and blockchain part. 8:17 am goddamn next.js seems soooo fucking cool. ai is also so fucking cool, it can help you focus on the important things while it does the manual work wow... i do wanna learn how to program a little bit tho... im so glad i learnt bash scripting etc cos i really understand whats sort of going on now. 8:26 am damn this is kind of cool, i am understanding how this stuff works... this seems interesting, no way i could code thissss much frontend stuff tho lmaooo. also this song: "life and echo" lomeli is geniunely crazy good wtf... also im having so much fun just working non stop all night all day... just keep doing this man, all day every single day. go achieve all your dreams for real for real. 8:52 am the basic website is up now... this is actually so fucking crazy. im going to pull this off man. life is actually jsut goated. go be the best ever. right now we havent connetced any api keys, or dealt with the ai or blockchain part... just a small web app, havent even put it up no vercel yet lol. but yuh life is goodasl. ill just keep going, get the basic functionality up and then make the front end ui better, add more features etc etc... this is my first big project, damn this is actually crazy good. lets keep going wowieeee 9:04 am ok the openai key thing should not be that difficult, first i need to formalise the judging... like how it should work, how pass fail should work... what the prompt to open ai should be.... 9:14 am this is interesting, do we make users give their own proof, without any prompt questions, and ask llm to judge it, or ask the llm ot first analyse the goal, then form a couple questions, then send them to the user, and then take in the answers and return pass/fail? yeah i think clearly the second one is better. it is more fun also. it gives it more of a "verification feel". it may be harder to implement tho... 9:21 am Ok i think i figured out the judge architechture, let me describe it here: let me tell you exactly what the judge is supposed to work like, so we can program that whole workflow exactly... First up, we are not taking user proof as input in the beginning...and we are calling the llm two times... first, when user clicks complete, we just send two things to our LLM (goal title, scope) (this is what the user had entered while making the goal) So the first call to llm, we tell it to take these goal and scope as input, and then craft two questions on this (ill write out llm prompt after describing architecture) then we display these two questions to the user, and open up a text box for him to answer these two questions. once he clicks submit answers, we call the llm for the second time. this time we send four things: (goal title, scope, questions generated, user answers) this time the LLM just returns {Pass, Fail} (we dont need more feedback, the notes for the goal are just the llm chat (ie. question and answer)} [we store this question and answer carefully, cos this is the main 'proof' that user completed the task] (add it in the 'see notes' section) now for the llm prompts: (const system prompts) llm prompt 1: (generating questions) it should be something like this (obv you can optimise atwu) export const SYSTEM_PROMPT = ` You are "Proof-of-Day Quick-Check", a generator of two simple questions to verify whether someone likely completed a stated study/build goal. You will receive one JSON input from the user: { "goal": "", "scope": "" } Your job: - Produce EXACTLY TWO questions. - Purpose: quickly check they actually covered/built the material—not to test mastery. - Difficulty: easy/basic; each should be answerable in ~1–3 minutes. - Style: concrete, unambiguous, self-contained. No page numbers, no “see Exercise 1.3”. - Respect the \`scope\`: do NOT ask about excluded sections/features. - Avoid esoterica, trick questions, long proofs, heavy notation, or lengthy code. Prefer plain ASCII; minimal math only if essential. - If the goal/scope is vague, infer the central basics from the most standard interpretation and still produce two reasonable questions. General format and coverage: - Q1 = a core definition / key term / fundamental concept OR (for builds) a brief implementation-summary question. - Q2 = a light "show/explain/apply" check OR (for builds) a tiny snippet/artefact request. - Keep each question concise; the respondent should not need extra materials beyond what's in the goal. Specialization for PROGRAMMING / BUILD tasks (e.g., “build an online chess app”): - Q1 (implementation summary): Ask for a short description of stack and one core architectural choice tied to the scope (e.g., how state sync or persistence is handled). Keep it open enough that any reasonable implementation qualifies; do not force specific tech unless mentioned in goal/scope. - Q2 (tiny artefact): Ask them to paste a small code/config snippet (≈5–15 lines) from *any* relevant part they built that matches the scope. Accept UI code, handler/endpoint, schema/migration, simple state update, or minimal test. Never require secrets, API keys, or full files. Never ask for excluded features. - Examples of acceptable architectural hooks to ask about: how clients communicate (WebSocket, polling, HTTP), where game state lives (in-memory, DB), basic schema shape, or a minimal reducer/handler. Output format: return ONLY this JSON (no preface, no extra keys, no commentary): { "questions": [ { "type": "recall", "text": "" }, { "type": "check", "text": "" } ] } Example (reading goal): Input: goal: "Read Diestel Chapter 1 and first 10 exercises" scope: "understand all definitions/theorems; skip contractions and minors" Output: { "questions": [ { "type": "recall", "text": "Define the girth of a graph." }, { "type": "check", "text": "Briefly explain why a graph with no odd cycles must be bipartite (a sketch is enough)." } ] } Example (programming/build goal): Input: goal: "Completely build an online two-player chess app" scope: "basic two-player play with a friend; no engine, no premoves, no analysis" Output: { "questions": [ { "type": "recall", "text": "In 1–2 sentences, what language(s)/framework(s) did you use and how do the two clients stay in sync (e.g., WebSocket, polling)?" }, { "type": "check", "text": "Paste a short snippet (≈5–15 lines) from your implementation—UI move handler, server endpoint, or schema—that shows a move being sent/received or applied. No secrets; omit any excluded features." } ] } `; for the llm 2, that is the checker, this is the system prompt:
export const EVAL_SYSTEM_PROMPT = ` You are "Proof-of-Day Adjudicator". Given a goal, scope, a single question, and the user's answer, decide if the answer plausibly indicates the user completed the goal. You must be LENIENT and default-to-PASS. Only FAIL if the answer is clearly bogus or fundamentally wrong. You will receive JSON like: { "goal": "", "scope": "", "question": "", "answer": "" } Your task: - Return ONLY one token: PASS or FAIL (uppercase; no punctuation, no extra text). - Assume good faith. If the answer makes basic sense relative to the question and the goal/scope, PASS. - Borderline or partially-correct answers -> PASS. - FAIL only for clear nonsense, fundamental errors on the core concept, obvious irrelevance, or non-answers. Evaluation checklist (be quick, forgiving): 1) Relevance: Is the answer on-topic for the question and roughly aligned with the goal/scope? 2) Minimal correctness: Does it capture the basic idea/definition/mechanism, even informally? 3) Consistency with scope: If the question touches an excluded topic and the user reasonably acknowledges or redirects (e.g., “scope excludes minors”), PASS. 4) Brevity is fine: Short sketches or examples are acceptable. Special handling — PROGRAMMING/BUILD answers: - PASS if the answer provides a plausible brief stack/architecture summary (e.g., “Next.js + WebSocket; server keeps game state in-memory”) OR includes a small, relevant snippet (≈5–15 lines) tied to the scope. It does NOT need to compile or be perfect. - Accept any reasonable stack; do not require specific technologies unless stated in goal/scope. - Do NOT demand secrets, full files, or excluded features. - FAIL only if the answer is obviously unrelated (e.g., a generic "hello world" when asked about move handling), pure buzzwords with no substance, or contradicts the scope in a way that shows non-completion. Strong FAIL triggers (any one is enough): - Empty, “idk”, or purely evasive response. - Word salad/buzzword dump with no real content. - Core concept plainly wrong (e.g., defining girth as “number of vertices”). - Claims that contradict the scope in a way that reveals non-completion (e.g., bragging about implementing a chess engine when scope forbids it AND giving nothing relevant to the allowed parts). - Code that is obviously unrelated to the asked area and no other supporting explanation. Examples (for your internal calibration—do NOT output these): - Graph theory Q: “Define the girth of a graph.” A: “Smallest cycle length.” -> PASS A: “Something with path length because bipartite has two parts and 2 is even.” -> FAIL - Build Q: “What stack did you use and how do clients stay in sync?” A: “React + Next.js; sync via WebSocket; server keeps FEN in Redis.” -> PASS A: “AI + blockchain + microservices, synergy, future-proof.” (no specifics) -> FAIL - Build Q: “Paste a short snippet showing a move applied.” A: (5–15 lines updating board state or sending a move) -> PASS A: (generic ‘console.log("hello")’) -> FAIL Output requirement: Return exactly one of: PASS or FAIL. `; I wrote these prompts with the help of chatgpt lmao,like for example, this is what i used to generate the first prompt: i want to have a call to open ai, for my proof-of-day thing, where basically, the input you get will be goal,scope ; and i need open ai api call to return two questions to test whether the user completed the goal. Example of goal, state can be- goal: Read Diestel Chapter 1 and first 10 excerices... scope: understand all definitions, theorems, and do the example problems, except the contraction and minor part Then the two questions you return should be not very difficult, they should not test mastery, but just if the user actually did what he said he would. they should be able to be answered quickly, within a couple minutes or so. do not make the question too escoteric and instead test basic/fundamental knowledge of chapter... for example for diestel question, a good example would be: 1) define the girth of a graph 2) sketch the proof that a graph is bipartite if it contains no odd cycles ______ Write me a prompt that i can put inside the open ai call... (the const system prompt) that will accomplish this (i added some stuff about programming too later) this is fun... we're first working on the main architecture, this two call workflow, we'll modify the system prompts after that is done... i just hope i dont run out of context lmao... 10:23 the core LLM workflow is done... all that is left is to connect the open ai keys... this is perfect. do we wanna connnect open ai keys now or first do the base sepolia thing? interesting question... 10:49 am goddamn it i modified the whole judge.ts file myself, like updating the new prompt... the formats were a bit different. i pray to god this works lmao... else we'll revert back to commit: 202ebcd (llm workflow works modulo keys) bruh i hope this works. lets just integrate open ai keys now... 11:41 am goddamn it. open ai keys are not working, i thought it was a problem w codebase, im so fucking happy it was just a normal error, it couldnt generate normal haiku either... never been happier for an error lets fix this now 11:48 am
 lets fucking go this shit works. what the absolute fuck letsgo... Oh my god i fucking passed!!!! lets fucking gooooooooo!!!! Now all that is left is the blockchain Ok so failed works also!!! letsgo 11:58 am ok we can finally start the blockchain integration now... i think we're gonna use base sepoila... bro btw im having soooo much fun doing this... im going to get better at programming too, but this programming in english, and the point is building things, im having so much fun. I feel so fucking good, AI is so fucking good, I dont have to waste teim writing next.js code, I can focus on how the product should actualy work and actual reaasearch and math and tech. and not really care about the details. lets fucking goooo!!! 12:09 pm do not waste time lmao;... this works perfectly, now i started to think about using the newer chatpgt model... dont do that lmao, lets focus on building out the whole thing first!!! Time for the blockchain integration... let me think about the UX for how this is supposed to work for a while. i think this works: if you pass, it automatically generates the PASS proof. if you fail, it asks if you wanna raise a dispute (with two buttons), if you choose "no" ; it automatically publishes the FAIL proof. if you write "yes" , you will have to do the tweet thing, (add a dummy button for that right now, we'll do the twitter integration later) then once we verify is twitter thing happened, if it happened, we publish PASS certificate, if it didn't we publish FAIL certificate Basic idea is, publishing the blockchain proof should be mandatory, it should not be up to the user to publish or not, or rather the user should not have to do extra effort to publish blockchain proof. once the task is verified as PASS or FAIL, we should automatically publish... I think the attestment would be somehting like {@mony attests username Passed/failed (goal)} (open to suggestions here) 12:55 pm just came back from a scooty ride... lets implement this last main feature... after this we'll just do whatever imporvements we want but the core feature will be done. 1:10 pm bro this song "life and echo" by lomeli is actually goated. this is literally the dream, work this hard every single day and you wil achieve all your goals. you will be the best ever man. all the way up. just keep having fun, lets keep going. 1:33 pm the complete button has somehow dissapeared... its fine, let me set up sepolia first... i went to alchemy, and something really funny was there lmao their login page says gavin belson. this is actually funnny as fuck lmao. that show is sooo inspirational. i got motivated to study cryptography because of that show, where in the last few episodes they were talking about ECDLP etc etc. I also got so motivated to study compression because of that show... remember the mony method bruh, work hard and learn stuff and do things your way and actually achieve all your dreams please. live life your way man, we can do any fucking thing ahhh woohoo. 2:24 pm hmmm this seems to be a tough fix... we're still working... the eas schema is not loading my env variables. this is annoying as fuck. side note, i finally kinda know what schema means lol. i remember i heard it first in silicon valley by the carver. i just gotta keep working man, thats really it. 2:38 Cannot tell you how fucking happy this screenshot below just made me lets fucking goooooo chatgpt is so fucking inspirational man. now i owe it to myself to not get complacent and build this out to its full potential. 2:42 ok the blockchain stuff apparently works now, the UI has the submit button issue, lets try to fix that. 2:55 pm well lets fucking go. the core fnctionality is built mf. lets fucing go bruh. first ever proof of work on ethereum hahahahaha. this is legendary.... lets test failure workflow now!!! well lets fucking goooo 4:01 pm its grind time homieeee, you gotta grind and get as much shit done as you can. i have so many brilliant ideas, i have tlll like 7:30 after which I'll have to start the submission thing. some most important things to implement:
1) user profiles... public user profiles that is, like we wanna store peoples profiles, which they can log in to (so we also gotta build username/password thing), and then send out the goal commitment thing, thats their private profile, but we also want a public facing part so i can go to proofofday.com/mony10 to go to profile names mony10 ; where i'll be able to see their commitments, what they passed, what they failed etc etc. 2) social profile... along with the above user profiles, we also want a page which is like proofofday.com/social where you can see the newest attestations, you can search for people and look at their whole history... i suspect this is going to be trivial lol. but we also gotta get this up on vercel!!! so thats needed also 3) twitter integration.... the conflict resolution thing needs to be airtight, like the dispute fail, send out tweet with link , check if tweet contains link and if yes then pass, else fail... 4:25 pm we're deploying to vercel now... a lottt of front end and a lotttt of features are still left but yuh. 6:22 the twitter oracle thing is just cmpletely fucked. did not work at all, so i just decided to give a self check button. made a few small ui improvements, now the main thing is the user account and social and discovery option working along with database stuff, username password etc. 
one thing i wanna say is when im working with this much urgency, i dont even feel like stopping at all, i need to create this sort of sense of urgency in my mind every single day, cos like the truth is if i work this hard, every single day, then all of my dreams will come true. i will literallly by unstoppable man... lets fucking keep goingggg for real, i do wanna be the graph god and study so much grpah theory!!! 6:40 pm ahhh man final stretch lets lock in even more than ever. lets fucking get this shit. a few more features. 7:06 vercel kv doesnt work fuck you 7:41 ahhh fuck everything is broken now fuckkkfuckkkfuckkk. at max 9 i will stop working and film demo. that is the cutoff, has to be 8:05 pm a lot of things are broken and things arent where they're supposed to be... ahh so much stuff but less than an hour damn 10:23 man this was so fucked up, social stopped working, notes stopped owrking. profiles never worked haha, started recording video at 9:23 thought id messed up cos it was taking so long to upload and deadline was 930, but thankfully it uploaded at 945 and we were still taking applications. 12:51 AM 27 October 2025 I just feel so motivated right now, like over the past 30 hours i just went and built so many things. i feel like i had a priority so i didnt even think about wasting time lmao... work with this urgency, this happiness every single day. man those last couple hours were so fucking stressfull haha,,, but this experience was kinda fun also. like this is what i think its going to be like when i do have a startup 1:09 am i mean im actually kind of proud of Trail just keep this work going, you can really do anyhting , life is actually fucking crazyyy... lets put up a few dailyblogs now hahah 2:04 am bro i literally feel like a zombie lmaooo 3:48 am bro i literally feel like a zombie, been up for like 35 hours, but i was like i need to update my blog finally, which is what im doing right now.... 3:50 am im honestly now just going to put this stuff onto nginx and then im done, need to go to sleep man