peacebeard

Thanks! I must have had a typo when I searched the page.

~That line isn't in the file I linked, care to share the context? Seems pretty innocuous on its own.~

[edit] Never mind, find in page fail on my end.

The name "Undercover mode" and the line `The phrase "Claude Code" or any mention that you are an AI` sound spooky, but after reading the source my first knee-jerk reaction wouldn't be "this is for pretending to be human" given that the file is largely about hiding Anthropic internal information such as code names. I encourage looking at the source itself in order to draw your conclusions, it's very short: https://github.com/alex000kim/claude-code/blob/main/src/util...

In my view an unappreciated benefit of the vanilla setup is you can get really accustomed to the model’s strengths and weaknesses. I don’t need a prompt to try to steer around these potholes when I can navigate on my own just fine. I love skills too because they can be out of the way until I decide to use them.

My experience is that Sonnet can be a bit verbose and prompting it to be more succinct is tricky. On the other hand, Opus out of the box will give me a one word answer when appropriate, in Claude Code anyway.

This is definitely part of it.

I think another part of it is that AI tools demo really well, easily hiding how imperfect and limited they are when people see a contrived or cherry-picked example. Not a lot of people have a good intuition for this yet. Many people understand "a functional prototype is not a production app" but far fewer people understand "an AI that can be demonstrated to write functional code is not a software engineer" because this reality is rapidly evolving. In that rapidly evolving reality, people are seeing a lot of conflicting information, especially if you consider that a lot of that information is motivated (eg, "ai is bad because it's bad to fire engineers" which, frankly, will not be compelling to some executives out there). Whatever the new reality is going to be, we're not going to find out one step at a time. A lot of lessons are going to be learned the hard way.

Huh, there must be some reason it shows up for some people but not others. Weird.

I don't see it there, but I do see it in screenshots online. Maybe it was removed or moved.

Yes, all the time. I understand that if you have a setup where you do everything in your IDE you could reasonably leave it full screen all the time and I get why that works for some people. I'm not one of those folks and I use separate IDE, terminal, browsers, and other windows and use window management to allow myself to see multiple of them at the same time and switch between them by clicking on what I want.

Also just want to be 100% clear: Tahoe is bad and I hate the changes and I don't think the OS should prefer one way of working over the other. I just hope it's helpful to explain my perspective.

Pretty insubstantial high level tour of broad AI pushback. Goes from "It was just not 'elegant.'" [sic] to "I don't want to give up my brain" [sic].

Nobody likes the feeling of being hunted.

This doesn’t address the author’s point about novelty at all. You don’t need 100% accuracy to have the capability to solve novel problems.

Their vulnerable introduction leading into the 4 task bar screenshots made me laugh out loud.

Everyone is different, and even though I don't share your experience, I don't view yours as either good or bad, it just is what it is. My experience is different but I'm not planning on ever telling anyone "Oh don't worry about it just have kids it'll be the best experience of your life" in blind faith.

Honestly I think this is the primary explanation for why there is so much disagreement on if LLMs are useful or not. If you leave out the more motivated arguments in particular.

In my experience trying to push the onus of filtering out slop onto reviewers is both ineffective and unfair to the reviewer. When you submit code for review you are saying "I believe to the best of my ability that this code is high quality and adequate but it's best to have another person verify that." If the AI has done things without you noticing, you haven't reviewed its output well enough yet and shouldn't be submitting it to another person yet.

Agreed. When you submit code you must take responsibility for its quality. Blaming AI for low quality code is like blaming hammers for giant holes in the drywall. If you don't know how to use AI tools without confidence that your code is high quality, you need to re-assess how you use those tools. I'm not saying AI tools are bad. They're great. But the prevalence of people pushing the tools beyond their limits is not a failure of the tools. Vibe coding may be fun but tight-leash high-oversight AI usage is underrated in my opinion.

Sure! I went through my Claude history and pulled out some of the most recent ones (anonymized a bit). I had it write code when it was rote / boilerplate work to save time, and other than that I used it more to discuss the code rather than write it for me. In between these, I was writing code myself.

Writing code:

- can you remove this <zustand store> and move the state here into <a new slice in a different zustand store>

- Can you update the unit tests for the working copy changes that need test updates / created (it wrote a bunch of tests which were satisfactory, I just deleted some redundant ones)

- I removed this <TypeName> type can the usages in this file be safety replaced with <OtherType>? (it analyzed the type differences, confirmed it was safe, and made the replacement, though an IDE could have done the replacement too)

- Can you fix this type error (it built a type guard function to address it, which is boilerplate code)

- is there a way to add the output here not just the input (it found a way to plumb some context through the codebase that I needed, pretty rote)

Other stuff I did in this time:

- can you review the code in commit <hash> (been doing this a lot)

- Seems my changes in this branch has broken <feature>. can you add logging to help me diagnose what is now wrong, and also analyze the git history to maybe get some theories on what maybe broke it (this can save a lot of time, digging through git history manually can be time consuming)

- This file doesn't work as expected anymore given that we've implemented <feature>. Can you investigate (another good time save, it's good at digging through a bunch of changes quickly)

Hope this was of interest to you!

When doing more technical / novel work you can’y vibe code it but you can still use ai tools to make things faster. Having Claude implement small chunks with strict direction and oversight is underrated in my opinion. Or just using it to search the codebase (where is the code that does x), implement tests, and do code review are all helpful. There is a lot of e-peen measuring around vibe coding but I think it’s really not the must useful workflow, it’s just chasing a dream.

That's a really good point. Maybe in part because Beyond had a highly visible IPO they became the poster child for the success or failure of meat alternatives but in reality their story is pretty much just their own story.

My personal experience is that Opus 4.6 degrades after a while but the degradation is more subtle and less catastrophic than in the past. I still aggressively clear sessions to keep it sharp though.

I never got on board with wireless headphones.

* Having to charge them is a PITA

* Having to pair them is a PITA

* Having more points of failure is a PITA

* Paying more is a PITA

On the other hand:

* Wires are fine

A little cynical of the arts isn't it.

We're going to have to collect data on the ratio of bathroom graffiti between tags and limericks. Anecdotally, where I live a it's mostly silly goosery.

Just to be totally clear, here is an example. Please cover up my user name for an authentic experience.

A dozen, a gross, and a score, Plus three times the square root of four, Divided by seven, Plus five times eleven, Is nine squared, and not a bit more.

Your “in practice” is doing too much heavy lifting here. This comes across as more of a prejudice on people than a fair assessment of the tools and techniques.

There is a big difference between "asking an editor for suggestions" and "vibe posting".

You don't lose your voice if you ask for advice and manually incorporate the suggestions you agree with.

You might lose your voice if you say "Improve my comment to make it better" and copy-paste the result without another thought.

The big question is: if copyrighted material was used in the training material, is the LLM's output copyright infringement when it resembles the training material? In your example, you are taking the copyrighted material and giving it to the LLM as input and instructing the LLM to process it. Regardless of where the legal cards fall, this is a much less ambiguous scenario.

You should consider “Make better care more affordable” as an alternative to “Take worse care of your kids.”

The new bottleneck is code ownership. You have to understand what it does and how it works to maintain it long term. You can LLM into a maintainability disaster but you can’t LLM out of it. Biting off more than you can chew is more dangerous than ever.

Works fine for me in sessions that use a lot of context. My workflow is to keep an eye on the % that shows how soon it will auto compact. And either /clear and start over, or manually compact at a convenient place where I know it'll be effective.