I’ve registered as a Cardano DRep. I spend pretty much every waking moment working to improve the Cardano ecosystem. Throughout this time, I was able to identify and resolve a large number of critical exploits in production open-source DApp protocols. Last time a large-scale network attack was conducted on Cardano, I stopped the attack and looted their funds in the process. I hope to use my technical expertise in smart contract development and protocol security to help vet proposals. DRep ID: drep1ydvwp8x9j8u4flwjqwxf8jdjwp9p4zf2x4rk7vv7fgzk6j2k7c6

Pretty cool!

We just launched ZK (the privacy kind) contracts on @MidnightNtwrk mainnet today 🙂

"Bro, engineers and math people are bullshitting you; anyone can build anything with vibe coding; no one needs to read the code anymore, we're past that, it's all imagination land now." AI psychosis, believing that you have solved the three-body problem because new shiny model told you it did, and then believing that the solution was formalized because the new shiny model told you it was. Meanwhile it formalized some random properties like when you press escape the animation stops. If you cannot form your own opinions, and think logically, it doesn't matter how powerful AI gets, you will only ever be able to produce slop.

Incredible work. Cardano is one of the best places for complex asset issuance

The team clearly put a lot of effort into the port, was receptive to feedback, and quickly resolved audit findings / architectural concerns whenever we reported them. I am happy with how the final design turned out, especially with respect to its ability to keep in sync as the Cardano ledger progresses and new features are introduced. Huge shoutout to Jann Mueller, Amir, and koslambrou who all made massive contributions to the original design and implementation. Fun fact, the original effort now has an academic citation in: From lambda to ledger: An architectural comparison of Plinth and Plutarch It takes a village for any feature of this size, so it is very satisfying to see it come to fruition. Looking forward to the deployment!

Throughout the development lifecycle of Midgard, Aiken has been put through the gauntlet. Hundreds of contracts, a massive test-suite, and hundreds of thousands of lines of Aiken. Our contributions to this release include: 1. Major compiler and optimizer speedups — PR #1399 The Aiken test-suite in Midgard was taking 485 minutes to complete; this was a real performance issue and heavily slowed down development loops. This PR introduced optimizations to the Aiken compiler and optimizer ultimately resulting in: 767-test suite was 485 minutes → 5 minutes 27 seconds, approximately 89× faster 2. Reduced CEK evaluator memory blow-ups — PR #1410 Aiken was regularly causing seg-faults and OOMs when building Midgard, causing our developers computers and our remote development machines to regularly crash on certain large scripts. This PR optimized addressed these memory issues resulting in: Deeply recursive list accumulation target consumes 48 GB RAM causing OOM → the same list accumulation target consumes only 1.1 GB RAM 3. Parallelized UPLC optimization during test runs — PR #1410 Another large performance improvement improving wall-clock times by roughly 3.5x for `aiken check`. Midgard also uses Plutarch extensively, and we have made similar contributions to its maintenance during this timeframe. A raising tide lifts all boats.

These people are truly some of the least intelligent individuals in all of human society. Even cavemen desperate for hunter gatherer comrades would ostracize people this level of psychosis. The only explanation that doesn’t make me worry for the future of society is if this movement of relentlessly whining over codex reset handouts is if this is a paid effort by Anthropic to socially pressure OpenAI into deciding to not issue resets. At-least then there is a rational explanation for this behavior.

Some light bedtime reading. Results of the defi audit from @alphagrowth1. However, this is just documentation, a paper-pushing exercise, NOT the actual real work towards outcomes. That work is happening in the trenches. Really smart people crave really hard problems. You won’t ever see them complain about things being unfair or too hard. 💪

Audit in! Every blockchain ecosystem has to understand what is missing before it can grow. Incentives can bring attention, but users and capital stay only when the infrastructure is ready. The audit gives PRIME a working map. Now the work is execution. 💪

This is a really impressive tool. Cardano DevEx is rapidly ramping up.

We tried to do this for years with Tweag. Amazing what the vibes bring

This looks really clean.

You can run full Haskell Cardano node in a web browser. Also balance and submit transaction fully on a browser as well.

There are lots of talented teams in the ecosystem. This is a huge opportunity for everyone to step up and build out the gaps to dramatically improve DeFi on Cardano. Review the categories and send in a proposal!

Chris is heading to Boston Blockchain Week today! If you're serious about infra, go find him. He was building Linux from scratch 30 yrs ago, he understands everything from the kernel-Kubernetes. Exactly the kind of person you want building resilient blockchain infrastructure.

This is so unfortunate for those impacted :( This explains a persistent "bug" that frustrated me when building float.sundae.fi though. I happened to be building the real-time updates features at the time, and several charts showed sudden sharp red candles.

This is actually braindead and worse than just not doing anything at all. By doing this, you reveal to the world that there is a vulnerability, and leave the money at stake. Essentially you are broadcasting to every blackhat everywhere “hey guys look there is an exploit here, we just used it to steal $100 (also illegal btw)! Definitely don’t try to figure it out yourself!”.

Fun fact, this is a result of the incredibly stupid watermarking added to Claude. The technique in discussion is canonically referred to as “red/green test driven development” in software engineering textbooks. Normally the model would be inclined to produce something like: “I’m running the red tests” but Claude’s watermarking system uses word replacement when doing so doesn’t (appear to) break correctness. This watermarking system allows the replacement of “red tests” with “red checks” because it considers both to be “correct”. This allows a probabilistic flagging of this response as Claude generated, because a human would not say “I’m running red checks”.