Oracle Cat

I decided to unpin the tweet I had about the ZK-EVM. I'm going to pin this one instead. What I wanted to say is this: Building it the way I did said a lot about me. I believe the way it was dismissed says a lot about them. I've moved on to other things. I feel increasingly better not thinking about it. It's just better off that we don't ever have anything to do with each other. I don't expect I'll ever try attending a crypto event ever again. It was only my mistake for expecting something different. I've learned a valuable lesson.

@0xschlboy Avalanche could have had my blockchain and another AI blockchain on their network but they chose not to support me. I was very attached to putting it on their network for some time. But they said no a few times in a row so I had to move on and just keep building independently.

@OracleThaCat Could have applied into build games with this imo. It's awesome tech, I'll tell you. Sheesh.

Honestly I have this crypto portfolio management project that is super good. Long story about why I was building this but it was basically to prove a point about something I was right about when I was working for a client almost 6 years ago. An async backend turns the app from a static portfolio management into a Bloomberg Terminal for your assets and view of the market. I could see something like this being able to get 100k+ users pretty handily. It is already better than the ones that get 500k users(with quite a few paying membership). Grok thinks a business like this would be able to generate $6m/year. It's nowhere near the top of my list of priorities but that's pretty incredible. Not sure what I'm going to do with it. I might consider cutting a deal to give the app to someone else.

I just checked my e-mail and I got a response from Apple just a week after applying giving me Endpoint security entitlement. This will allow me to really complete Beaver Warrior. They don't give these out to everyone it's pretty restricted, they manually review each application and it's pretty rare for someone solo to get it. I said to a guy on LinkedIn about a month ago who wasn't "getting it" that Apple would kill for exactly the kind of security application Beaver Warrior is. Local first, privacy is totally up their alley.

Brad Treliving was fired a few days as the GM of the Maple Leafs. He was originally signed for his "professional experience" a few years ago. When you have 9 years of "professional experience" in a role and you're in the bottom half of the league in performance I don't honestly think you should have been getting another job based on that experience. The Leafs went backwards from the "New School" stuff of basing their decision making a lot off of data & analytics to the "Old School" approach which leans more on personelle evaluation of players and what makes a good team. A lot of hockey guys feel that making decisions the "New School" way ignores the ability of your talent evaluators, what you get from watching a guy play. This carries over into other sports but for the most part in other sports "New School" has won. Why? Because it was better and the people who used it performed better. Hope their next hire is a fucking nerd who can rattle off advanced metrics of hockey players around the league. It's a fucking disaster what's been happening recently.

I pine for making new discoveries with AI assistance. I have quite a few that I've made but honest to goodness I want to make more. I only started making discoveries in the summer of last year. A few of the discoveries I've made were just at the tip of my tongue. Once you realize you can make discoveries with AI there's something of a snowball effect for finding more. Anyways, it feels better to me that a human being with AI assistance make these discoveries instead of AI doing it autonomously(which is what will eventually happen). I have to always think: what the hell am I missing? The best I was personally able to do before without AI was make the connection between an important cryptographic primitive that wasn't being used for attestation but should — the cryptographic accumulator. I think that's still genuinely impressive. I remember telling a former partner on my blockchain about this, it was really big, and actually that's about the time he said he would get back to me but never did. Screw him, that was fucking huge.

This is insane trust me. To get any kind of security today you have to give over the entire intel of what's happening on your device all the time to a big company. You pay a monthly subscription for them to protect you but what you also give them is the ability to spy on everything you do. Beaver Warrior is all the security(actually more) and none of the spying. Imo this is one of the most important/impressive projects that's been built+released in the last 5 years. I've obviously got to find a way to market it. I'm just going to see where things go with this other project I have in the next 2 weeks. But this is already an S tier app.

@OracleThaCat Okay, I get what you are saying now. Would give this a quick look.

Does anyone else just have a general gross dislike of Anthony Scarramuci? I saw him twice last year at Toronto's Consensus Conference and then again at the Avalanche Summit. He was becoming pretty buddy-buddy with Avalanche people. Just his presence set off red flags in the awareness my parents gave to me about certain other people. And then I listened to the things he says. You would hope somewhere in this life that as we get older we somehow become better and better people. I certainly feel that way about myself. This observation really escapes me about this guy.

Other stuff I have is a few tiers above. Like Beaver Warrior beaverwarrior.com This app is a structural shift in the security market. I also have another even bigger project that's coming late next week or early the week after. I'll probably end up putting it up anyways(I can still do multiple things at once). But I would totally prefer a way for me getting some $ and a % for building it and never have to think about running it as a business.

@OracleThaCat Don't give it to someone else gaddamnit! Make it work and take the credit.

Ok so I understand there's been a change with Claude Code recently that you will get throttled for less usage during peak hours: weekdays 5:00 AM – 11:00 AM PT / 1:00 PM – 7:00 PM GMT. It's not going to mean that your weekly quota will drain faster though which is good. Now waiting for my reset tomorrow at 6 am.

Hit my Codex(OpenAI) weekly quota in less than a day(even though usage limits are doubled right now). I got 70-80k LOC which will need to be audited when I get Claude Code weekly quota reset tomorrow. Expensive, but I think it was worth it to keep me on track; on the timeline I had in my head.

It's been a really long time since I've been able to plan out a week in advance. Every day for a really long time has been a combination of building as much as I could, at the same time as figuring out how I could make any bit of money. The building went really well, the moneymaking not so much. I'm so glad to be feeling like I could be near the end of this. It's been such an incredibly difficult journey.

The cost to participate is what most people can manage. For $200 a month you can play. That's $2400 a year. Right now this is a lot to me but I'm at the tail end of all this. I don't underestimate that $200 can be a lot to a lot of people. If you've received a few million in funding you have incentive to let your engineers keep going with near as many credits as they want. What your engineers manage to implement is probably worth it overall. I can say the vast majority of time I've been doing this has been while using 1 terminal or what I used to do before realizing Derr there were tabs, was open up 2 windows. Most of the work here was sequential anyways it would have been challenging to parallelize. Anyways, it helped me compartmentalize so not a huge deal. But just saying you don't actually need to have unlimited money running 6 tabs to get somewhere. But if you are doing that with a budget I am a bit scared of what you're capable of doing.

I just realized that if you're married and living with a partner you can work on the same project together, and probably Anthropic would let you have a Claude Code Max account each — that's not against the rules. Parents: "Why did you get married and start living together?" Married Couple: "Well we realized this was the loophole to getting 2 Claude Code Max accounts and saving money." Parents: "We did it for the tax benefits." You: 😮‍💨

Codex also had a 2x going on at the same time as Claude Code. Last month must have just been an insane time to be building; especially for solo-independents that don't have funding resources. I always wonder who else is out there in the universe.

For all the really really smart people in the world who are doing research & engineering with AI, there's a really fat-easy one that's been discovered by only me or hasn't been made public. When you figure out that one you'll probably figure out a bunch more. For example Beaver Warrior: beaverwarrior.com This isn't normal. This is an alien mothership. Smart people would question this much security being possible to run locally. It is possible. I would have never endeavoured to build a security app otherwise.

If there's positive news about Ponce potentially being out for the season it's gotta be for Jose Berrios. He's mentioned having regrets about leaving the team during the post-season last year(even though he was injured). I'm sure he'll use that as motivation to have a really good year when he gets back. Prior to the injury affecting his pitching he was having a great year with a 3.75 ERA. The prior 2 years he pitched to 3.65 and 3.60. People are mega-down on this guy expecting regression in his Age 32 season. If he was 35 I'd be a little worried. If this guy is your durable #5 starter you are lucky. Still holding out hope Ponce will be able to return near the end of the season. He looked so good and he just made it back to the big leagues. To me this looks like the kind of injury that's not actually a big deal unless you're competing at 100%; rest and recovery are a slow process. If I ever have a platform(which I might) I would want to talk about the idea of bringing healing peptides to professional sports. You get to use them if you're injured but then lose that privilege after you're healed so you can't use them for performance enhancement(this would be fairly minor but it still is).

I'm also doing expansion on a crypto portfolio manager app I worked very minimally on last year but is actually a very decent app. Not sure what I'm doing with this but there's a story of why I wanted to build it.

Currently working towards adding 200-300k lines of code to an agentic cli using GPT. There's currently someone working on a heavily starred repo(50k) translating Claude Code's leaked code into Python for some reason, after he deleted the original code from his repo. Soon Rust as well 😂. Doesn't make sense unless you actually have an LLM. Anyways, I shall never be accused of stealing any ideas from Claude Code's leak. I have no ideas on doing this I'm leaving it entirely to GPT to just make it the best possible. I'll still audit it with Claude Code though.

I'm a person that believes that unless there's some kind of architectural inefficiency more code is usually a good thing. This is not always true I have seen 2000 lines of code be better than 100k. A lot of people will say then it becomes difficult to maintain but what does that really mean? You are going to still have AI to deal with everything that's built. Code is usually making something better in some way. So right now I'm just getting more and more code for a coding cli. Honestly this isn't my department for knowing what's good. I'm letting AI make those decisions for me.

Just the early thing I can say about Codex, but I think this is generally as it is, is that this is very hands off I'm not really thinking. I can run 10 or more agents at a time. But the quality of each session isn't as great and I'm not sticking around to make sure the code has no issues(which I'm sure it does). Every time it stops I just paste "don't stop keep building". This has value for creating code which I will then use Claude Code to audit. I'm at a stage anyways where this isn't so much about thinking. So at the moment I'm not regretting the purchase.