Jacob Enders

One of the most nostalgic games ever

@unclebobmartin All the kids these days are writing their code by hand, it's the new fad

Authoring code by hand HAS GONE AWAY. Engineering module structure and architecture has not.

@unclebobmartin x.com/FaytuksNetwork… Opinions on this?

So get this. The Democrats in Virginia are so upset about the Supreme Court of Virginia‘s decision to throw out the redistricting map that they are seriously considering lowering the retirement age of all the Supreme Court justices to below the age of the youngest member so that they can throw out the entire court and then appoint new judges who will accept the redistricting.

@theo @dagecky The YouTube replies

@dagecky I didn’t see much hate on this one?

Please stop submitting AI slop code pull requests to RPCS3. We will start banning those who do without disclosing. There are plenty of resources online to learn how to debug and code instead of generating slop that you don't understand and that doesn't work.

@unclebobmartin @BillySlang @YouTube Grrrr, I hate that the Democrats control the house, senate and presidency! We need someone to oppose these crazy liberals!

@BillySlang @YouTube In this case there was no blame. VDH was covering the facts. Frankly, neither he, nor most republicans, want one-party-rule. Power corrupts. Absolute power corrupts absolutely.

Winning the Map: Why the Left’s Gerrymandering War is Backfiring | Victo... youtu.be/exnLBvSb0sU?si… via @YouTube

This is the greatest piece of Star Wars media ever produced. I will not be taking counter-arguments at this time.

@unclebobmartin Soo... Not trump?

Here’s a tip to help you vote. Choose the candidates who engage in civil discourse rather than hyperbole.

@theo About what?

How are you guys feeling about 5.5 so far?

Microsoft accidentally told the truth about AI

this tweet low key ruined the world lol

Star Wars: Knights of the Old Republic (2003)

@transkatgirl @eriskiiii People still play chess or knit shirts. Even though AI and factories do both things better.

@eriskiiii because it's depressing and demotivating to see most people accomplishing things way faster because they have access to a tool you can't afford.

freaking the fuck out after trying AI coding tools a bunch i just realized that this is the end of software dev as something you can start doing for essentially free if you have a computer

@MarcoNL88 @krishnapro_ @housecor Do you trust the human to write the correct prompt to make the AI generate the correct tests?

@krishnapro_ @housecor Do you trust the tests written by coworkers lol? Humans are lazy and just want to hit coverage thresholds and press merge. AI doesn't get lazy and covers all edge cases.

The acceleration from AI is unbelievable. After *only 5 weeks* of development on a new project with 4 devs, we have over 4,000 tests. That's ~200 tests per dev/week! Absolutely no way we could've done this by hand. Vitest runs them all in under 15 seconds on my M1 Mac.

@unclebobmartin Everything good?

@JacobEnders I'll be here.

My thoughts on the Iran issue. Thank God that there are leaders who understand the threat and are taking the necessary action to address it. I've been waiting for this for 47 years. I remember, all too well, the day the American hostages were taken, and the 444 days of their captivity. I remember, all too well, the Iranian diaspora, and the horrors that the regime imposed upon it's own people and upon the world. And when it is gone, and buried, with a stake through its heart, I will smile and say: Good Effing Riddance.

@icanvardar A vibecoder is just a tool that the AI uses to get its job done

are vibecoders even programmers?

@LGrapenthin @unclebobmartin I find it much easier to express my ideas in a language designed for software instead of english

Yet "deterministically generate desired behaviors and structures" has already been solved via programming languages, designed because natural languages lack the required semantic unambiguity. I expect your attempts of filtering out natural language ambiguity and misinterpretation via methodology, to converge towards another programming language again interpreted by a neural net instead of a compiler. Because what is it, that you want from programming in a natural language, if not the questionable freedom of being ambiguous and vague?

AIs are just another step up the semantic expression ladder. We initially expressed our semantics in binary, then assembler, then Fortran, then C, then Java, then Python, etc. AI is just the next step up that same old ladder. And when you take that step, nothing else changes. You are still expressing behavioral semantics. You still need to express structural semantics. All the old principles still apply. You still have to be concerned about design and architecture. And even though the syntax allows informal statement, you cannot abandon formalism. When you express behavior you need a formal way to enforce the behavior you want. I use Gherkin for this. It seems to work pretty well. Consider that Gherkin is written in triplets of Given/When/Then. Each of those GWT triplets is a transition of a state machine. A full suite of Gherkin triplets is a formal description of the finite state machine that represents the behavior of the application. Other formalisms that matter are things like module dependency graphs, testing constraints, complexity constraints, and many others. This step up the semantic expression ladder provides you with an enormous amount of options. But you'd better choose those options wisely!

@yoemsri @unclebobmartin Some people care about the things they build.

@unclebobmartin AI optimises for the metric not the goal, same as humans, the fix is also the same, make the metric harder to game than the actual work.

With AIs “distance” is important. One AI that writes tests, and another that makes them pass, sets the two at odds with each other and prevents either from cheating. This is very much like the separation of powers in a democratic constitution. Another form of distance is semantic. I have the AI create a parser that parses gherkin into an intermediate representation file. Then I have it write a generator that reads the IR file and produces executable test code. That semantic distance interposed so many intermediate goals that the AI cannot reach through them in order to cheat. Or, to say that differently, the intermediate goals become higher priority than the overall goal of making the test pass, making reinforcing the test higher priority than cheating to get green.

@unclebobmartin How much does this cost you to do? Do I need to pay claude 500 dollars a month to be a software engineer now?

i have a theory that all the people complaining about Claude getting worse actually just vibe-coded themselves a massive ball of slop without paying any attention to architecture and they gradually start noticing poor Claude struggling with it more and more