Two of my favorite movies, but Interstellar was better. Truly two great movies!!

Now the book Contact, that was amazing.

Chicago- Saturday in the Park "Live" (1972) YouTube video by ChicagoKid1969

Saturday In The Park - Chicago

youtu.be/UTFD1C4tVIg

I am shocked, shocked I say!

Who would have thought, in any possible future, that the proliferation of betting in sports would lead to corruption and the fixing of games.

Why, I never, ever thought that would happen…

Floor Jansen of Nightwish.

Great post - 100% agree! Between great tests and git, it is much more manageable.

Agreed! We were talking about this in the early 1990s. If you’ve heard of the SOLID class design principles, there were five more principles that focused more on dependencies. The Stable Dependency Principle was one of those!

Working Man - Rush

Saturday In The Park, Chicago
Lakeside Park, Rush

Pat Benetar - hit me with your best shot.

Cinderella Man - Rush

You’ve got pockets of blue in the bigger cities and a whole lot of red everywhere else. We’ve tried…

You Took the Words Right Out Of My Mouth- Meatloaf

Goodnight Kiss - Dream Theater

REO Speedwagon - Roll With the Changes (Black and White Version) YouTube video by RSpwagonVEVO

m.youtube.com/watch?v=4PdU.... Gotta keep on rolling!!

The Intelligent Terminal Warp is the intelligent terminal with AI and your dev team's knowledge built-in. Available on MacOS, Windows and Linux.

By the way, my new strategy is warp.dev. Completely and totally love writing code through it.

Rush
Dream Theater
Steely Dan

The chances of him knowing anything about the story are minimal at best, either now or after.

So what was the problem??

Potentially the worst song ever… We Built This City.

I was a huge fan of Claude 3.7 for coding, and was paying for the Pro version. Then Anthropic launched their Max tier and cut the resources available to Pro by it feels like 75%.

Coding sessions that used to go for a few hours now get stopped after an hour or so.

Looking for another strategy.

At least for me, LLM-based coding consists of a conversation to explore the solution and problem and then rounds of code and improve.

Maybe the better way to go is to use one LLM for the discussions and another for the coding, with explicit instructions…

Otherwise it just doesn’t work.

My favorite law of all time is the Law of Unintended Consequences!

I took the reply fed it to ChatGPT, asking for any semantic content:

Great question — and the short answer is: very little.

Trump’s reply, as quoted, is a rambling, vague, and often contradictory statement. It lacks clear structure, concrete ideas, or a direct answer to the question.

And so on..

Bat Out of Hell. I just realized that 20 years after hearing the song for the first time.

As much as I loved using Claude Code, it is *expensive*! I’ve had really good luck with Claude Desktop, with two MCP servers, desktop-control and sequential-thinking.

I give it tests and a problem and it iterates until they all pass. Then it refactors for me, uses great names, small funcs.

I’m sure that’s part of it, but threats to themselves and family must play a big part.

If they’re that afraid, they should resign instead of capitulating. It’s the power they’re in love with.

Keeping a population uneducated is a lot harder in the day of the internet. You can see news from around the world, learn about any topic you want, and with AI, you can have a lot of the more complicated things explained to you reasonably well most of the time.

Times are just different.

I agree about Claude Code. I’m hoping the token changes they made are going to come to Code soon.

I am finding that Claude with a couple MCPs is a good replacement.

- AI is terrible at following directions consistently. I have several operation directives to work in a TDD way, and think about small pieces. Constant reminders.

- this is letting me fall in love with development all over again.

- AWS permissions are the bane of any developer’s existence. Code has been finished for a week. Spent the rest of the time battling policies and files and services, oh my!… not productive time. Horrible idea, but over-providing permissions will get you don’t a lot faster.

Couple of things I’ve noticed while building a canary in AWS using AI.

- AI can be tenacious in following a path. They get an idea and drive down that road. Spent a ton of time debugging some issues that we avoided at a suggestion from me. Moral is don’t rely on them for creativity in design.