I absolutely love the minimalist UI.

I see code reviews is in the roadmap, I can't wait to try it.

Feels like one can just copy the UI and use it for forgejo. It would get something similar very quickly, and avoid handling all the difficult stuff I guess.

Drifting? I think it's there. basedpyright is awesome and super fast. Our latest services are all CI gated by type checking. Early in my career you'd hit so many dumb errors running your code - NoneType, attribute, value, and type errors. I'd say that's been cut over 95%.

How are you getting static typing without additional tooling?

I mean, of course there is tooling involved. But it's part of the language. Unlike TypeScript which is a superset of JS and requires transpilation.

What tooling are you thinking of that’s part of the language? I don’t think CPython ships with a type checker, does it? People typically use mypy or pyright and neither or these are part of the language.

I don´t get the design choice for Glimmer components:

  export default class CounterComponent extends Component {
    <template>
      <p>0</p>

      <button type="button">+1</button>
      <button type="button">-1</button>
    </template>
  }

This is obviously not valid JS. If they already had to create a DSL for components, why not embrace it fully and introduce a different keyword instead?

    export default component CounterComponent < Component {
        <template>
            <p>0</p>

            <button type="button">+1</button>
            <button type="button">-1</button>
        </template>
    }

JSX class components, even though not technically valid JS (render method returns HTML-like syntax), resemble a JS class much more as it requires methods to declare the template and handle component lifetime.

I understand this is project-scoped and not intended to replace a window manager, but it just occurred to me that this concept could be a cool experiment for a Hyprland plugin. Instead of multiple workspaces, a single, infinite one that could be navigated by dragging and zooming in and out. I would probably hate to use that, but I supposed it would be a lot of fun to develop it.

This is sad. I had a blast implementing my own interpreter from scratch. I saw, first hand, the effort Sarup put in trying to make CodeCrafters thrive. We had a few product feedback video chats, which were all much enjoyable. I tried to get my current employer to pay for CodeCrafters for the engineering team, but, as others in this thread already said, it's hard to justify this spending these days because of AI. Honestly, I think that it's a little short-sighted of the companies to not invest in hard skills for their employees, now more than ever. I definitely became a better engineer with the help of CodeCrafters.

This is awesome! OpenAPI integration is a great feature.

> Nobody writes essays in Slack

I 100% write long texts in Slack. I always try to provide as much context as possible when reaching out to someone with a question or request.

Do you start every response off with "that is a great question"? I don't know any human who does. "that is a great question" is reserved either for really hard questions, or sarcasm. The majority of questions are not great, they are just things the asker needs a simple answer.

I remember listening to this Freakonomics episode [1] on the phrase.

It covers a variety of times people might use it. Sometimes it's genuine, other times it's flattery. Some people use it all the time (and in the episode they talk about calling someone on it). One guest says it's a bridge, to go from the question asked to the question you want to answer instead.

I don't see it in this transcript, but I also thought I remembered hearing/reading that it can be a sign the speaker doesn't know the answer offhand, and needs to consider the question to formulate their answer. I think I'd classify that as a genuine usage: the question is something the speaker hasn't considered before, and thinks is worthy of consideration.

[1] https://freakonomics.com/podcast/thats-a-great-question-rebr...

That's a great question! Ahem.

I remember being advised to do this ~20 years ago when I was going to be answering questions from a group of people. I was told that it's good practice to say something like "that's a great question" every time someone asks anything, as a form of social lubrication, to encourage others to ask questions. I can't say whether it works, and it was advice for a spoken context rather than written, but I don't know how to finish this sentence.

Math professor here.

When I go to research lectures, I sometimes hear that in response to audience questions, although not especially consistently. Some speakers do this more than others, I don't think anyone does it all the time.

It was so long ago that the specifics have faded, but I remember I was coached to use a variety of positive responses. "That's a great question," yes, but also things like "I'm glad you brought that up," and "I was hoping someone would ask about that!" It wasn't my cup of tea, too artificial, but the advice was contemplated.

The next question (which is a great one, from what I understand) is: Why do LLMs use these phrases so much if humans rarely use them in written form? Maybe a fair portion of training data comes from lecture transcripts, where such responses are common when responding to direct questions? And/or system prompts are just instructed to be like that?

> Why do LLMs use these phrases so much if humans rarely use them in written form?

As far as I understand, it's due to RLHF. The reviewers the AI companies use don't necessarily know what kind of question is a good one, so when the LLM answers "That's a good question!", they tend to rate the answer higher because they like being flattered. Proxy models that are themselves trained on RLHF inherit this pattern. Similar effects contribute to sycophancy.[1]

[1] https://arxiv.org/abs/2310.13548

I've always seen and used "That's a great question!" and similar vacuous phrases when speaking as a polite way to buy some time while formulating an actual response to the questioner.

That's a good question... => I didn't think about this, I don't know the answer yet.

That's a great question! => I can tell you understood what I explained and used that understanding to reach the next step of reasoning just like I did.

In a spoken Q&A setting opening every single response with "that's a great question" or "thanks for asking that" or whatever is pretty common as a way to fill a few seconds while you think about your response. This is obviously unnecessary on slack.

Well before the LLM explosion I would often preface my answers with some form of praise for the question. It depends a lot on audience of course, but it’s amazing how many people tend to perceive direct answers to their questions as negative… and just as amazing how far a little strategic sycophancy goes to temper that. Even though everyone knows it’s half-sincere dead weight.

And in parallel: a non-trivial number of people in class or at work are asking questions for no reason other to establish their own social credibility as smart and or knowledgeable.

Validating their ego and effort and social position can fulfill their social desires regardless of an answer that dismisses, say, buzzword soup as both inappropriate to the current context and incoherent to people who know what those words mean.

It might be a difference in culture, but I definitely remember people saying you should say stuffs like that to encourage others to ask more questions. The rationale is if you are in a more powerful or senior position, like a college professor vs a student, other may be not feel comfortable asking questions because they fear appearing dumb or inappropriate and it hampers communication.

I would probably reply with "that is a great question" only as a euphemism for "I don't know"

From Schlock Mercenary: "'That's a great question' is a terrible answer."

I have only heard this phrase in american tv shows and movies

People use that too when speaking in real life when they are stalling a bit to conclude their internal thoughts before providing an answer.

And worse, you use it so much that when the question is actually great, you can’t convey that convincingly anymore.

I didn't, but then I watched SICP and the guy kept saying it to utterly dumb questions.

And then, somehow, he kept turning the dumb question into a deep and insightful one.

So I've adopted that technique but if I say "that's a great question" think for a while and come up with nothing, it's probably a moronic question. I'm not perfect.

work at a crown corp and you'd change your stance on above...

<context> <tutorial> <anecdata> <answer> <sumary> <funny hook>

Introducing AI made markdown tags for conversations so others can only see what the wanty

Could add a <vitriol> tag to that - but yes, if that was auto assigned by LLM - i could see that.

Could even add a "Autistism" filter, preventing conversation digressing, filtering out only points that stay on topic and only the <summary>, that way.

> Could add a <vitriol> tag to that

Is that equivalent to the already popular <rant>?

Hah, can we do that for recipes next?

Conversation add blocker unlocked

Exception that proves the rule. You know what context that specific recipient needs from you. GenAI usually doesn't.

There's a difference between a long text and an essay. You wouldn't spend hours typing a message and formatting it with headlines for example. You wouldn't insert loads of unessorasy creative writing techniques in to a message asking for help.

I do too, but it does seem like fighting the tool. I wish more people would use email.

Your co-workers must love staring at the "SwiftyBug is typing..." prompt for 15 minutes

Please learn to love bullet points for the sake of the reader.

When a person expands bullet points they add extra information from their own knowledge and research in the process of writing. When AI does it, it adds filler and repetition.

Long form writing itself isn't a problem, it's the empty fake long form we have now.

Came here to say this as well. I've written and read full human-written essays on slack before AI.

With that said, I don't disagree with the article. Don't use more word when few work.

Sorry for the long response, I didn't have time to write a short one.

But this should not be an excuse applicable to an LLM.

Speaking as a random internet stranger, it depends entirely on context.

Sending me a message saying "Hi, I'm getting a Frobnizzle not found error" is a waste of both our time. Explain what you're doing so that I can reproduce it, even if it takes a few paragraphs. Maybe send me your user ID so I can check our logs. I don't care if you're declaring "these are the facts" because the facts are what I need to help you.

If it's a massive wall of text with a defensive tone during a discussion, yeah, sure, that's bad. Do you work somewhere where that's common?

I made a basic reference site around similar ideas a while ago: https://quick-answers.kronis.dev/

I think some people just prefer a more conversational format.

Good, realistic examples! Potentially related word/concept: alogia.

The classic "Hi" message never gets old... no wait, sorry, it was always old.

Some people, like me, have developed this communication style because it turned out that when they didn't they were very often misunderstood. When properly applied (i.e., not excessively, no actual walls of text), giving appropriate context helps focus the thinking of the receiver in the right direction.

...but yeah, every now and then I do get comments that I'm very wordy.

> It's much harder to have a proper back and forth with a firehouse than it is a few sentences at a time.

Sometimes, a back-and-forth is not needed, and the entire response is necessary for someone to understand to interact with.

This is when I open up a text editor, draft it, and paste that into Slack.

> It declares authority "these are the facts" rather than "let's discuss ideas" and if you haven't fully earned that authority it honestly just kind of smells of insecurity.

Not at all.

1. Someone is coming to me with the question. They're doing so because either the question is about my area of ownership and I have that authority or because I'm a subject-matter expert and I have that authority.

2. I don't know what the other person knows around the edges of the domain that the question is in, I don't know if they understand the constraints of the domain nor do I know what constraints their specific problem has.

3. Often the answers to any actually decent question at work are fairly nuanced, and to understand the nuance you need at least a level set of context.

It's a lot more dismissive and rude to answer with excessive brevity if you treat the question in good faith. For simple questions, sure I don't need to write an essay. Some questions I answer with "Got 10 minutes so we can chat about this?" because it just needs a conversation. Or I answer with questions of my own. But if the question is well-formed, answering it starts with providing the necessary context to understand my answer so we're on an even playing field and we can effectively communicate ideas.

How long have you been at it? Because some of us grew up writing letters with pen and paper, sending them to people in the mail, and getting something back a week or two later. You just have to actually sit down and READ closely what people are saying, sometimes multiple times, to make sure you are clearly understanding what they’re saying rather than skimming everything you encounter for information to extract.

It is actually quite easy to communicate a problem in the middle of a wall of text. You simply refer to the phrase and then explain why it doesn’t hold. It is also fine to simply present your perspective to people without invitations to “discuss ideas.” You can open a discussion if you want, but if I’m telling you something then you can rest assured that those are the things I believe to be true, and if I am uncertain about any conclusions I will include caveats to indicate uncertainty. You have free will and are perfectly capable of taking or leaving anything being said to you.

> It's much harder to have a proper back and forth with a firehouse than it is a few sentences at a time.

Sometimes just providing the context about an issue is itself enough to warrant a few paragraphs, let alone a few sentences.

Obviously, it's best to express a problem, its options, and its recommended solution(s) in as few words as possible, but it's unwarranted to hold an absolute position that every discussion must be an inefficient exchange of mere sentences at a time.

I'd rather have an outline-style essay given to me in one go for me to digest and research async rather than be subjected to a barrage of back-and-forth pings. That's the real disrespect of other people's time.

The next step is to not talk with each other at all.

Just have a LLM that "knows you well" in all your position argue by points and values assigned to the points with the LLM of the opposition.

If value alignment exists, a actual conversation may be engaged.

One approach I've been really starting to enjoy is to use use Tailwind alongside scoped styles (in Svelte and Vue). This keeps template pollution minimal while still allowing for the conveniences Tailwind brings:

  <div class="counter-component">
      <button @click="count++">+</button>
      <span class="count" :data-is-even="count % 2 === 0">{{ count }}</span>
  </div>
  
  <style scoped>
  @reference "tailwindcss"
  
  .counter-component {
      @apply flex items-center gap-2;
  
      button {
          @apply bg-gray-800 text-white;
      }
  
      .count {
          @apply italic text-teal-500;
  
          &[data-is-even="true"] {
              @apply text-rose-500;
          }
      }
  }
  </style>

yeah, likewise, I locked in on this very early on even though it goes against what the tailwind creators recommend. But I've never regretted this approach and it works well.

Adam always did say best practices don’t work.

Wow, you should really go to a butcher shop nearby and buy some filet mignon. Try making some steak au poivre, a classic of French cuisine. It's not too hard and it's delicious.

I wonder how well Mandarin works for LLM-based programming. On one hand, it's very token efficient as Mandarin script is very dense in meaning. On the other, I suppose this can increase ambiguity.

Character-density and token-efficiency are different things. Latter is data and, therefore, tokenizer specific e.g. take GPT-5's tokenizer o200k_base and run mandarin text and its translation through. Some amount of the time en will beat zh. I just tested with news articles and wikipedia.

After all `def func():` is only 3 tokens on o200k_base.

I can speak, read, and write Taiwanese Mandarin (which is likely relatively underrepresented in the training sets and, which is, in my practical experience, materially different in its usage.)

The authoritative answer for this question would best come from the millions (or tens of millions) of Chinese-speakers who are currently using LLMs to write software.

However, it is my suspicion that you would see no advantages using any language other than English. While there is a certain token-level density to written texts, it seems the benefits of this (and the more recent discussion around “caveman talk”) are quite limited.

Furthermore, consider that the vast majority of textbooks, technical documentation, blog posts, StackOverflow answers, &c. are originally in English. Historically, where these have been translated to Chinese, the translations have often been of very poor quality (and the terminology and phraseology is often incomprehensible unless you also understand some English.) I would suspect that this makes up the overwhelming majority of the training sets for these models.

That said, my experience using the most recent models, is that they are surprisingly language-agnostic in a way that surpasses readily-available human capability. For example, I can prompt the LLM to translate English into something that uses German grammar, Chinese vocabulary, and Japanese characters, and I'll get an output that is worse than what a human expert could do… but where am I going to find a multilingual expert?

(Of course, I have so far only ever been impressed that a model could generate an output but never impressed with the output it did generate. Everything—translations, prose, code—seems universally sloppy and bland and muddy.)

So what I would anticipate the biggest benefit for a Chinese-speaker today… is that if they are disinterested in working internationally, they have significantly less dependency on learning English.