gulugawa

Coding without LLM AI.

Another piece of evidence that Claude Code is a scam project. It is being subsidized by venture capital funding.

I have a proof of concept here: https://dmvboardgames.com/

Here are more details about the planned architecture: https://createthirdplaces.org/events/techOverview

I'm still going to develop open source projects. They are rooted in real-world considerations that AI models have trouble understanding.

Sounds like a great way to make sure students understand fundamental concepts.

MyselfGPT

Sound more like modern day colonialism than an attempt to promote useful tech.

This deal seems to reflect the desires of a company that wants to profit from AI , and not what regular people care about.

It's also worth noting that sometimes, groups are on Meetup because it is one of the least bad options.

I still host board game events on Meetup, although I'm hoping to switch to my open source alternative soon. I think the other commercial options such as Discord and Facebook are worse, and people have gotten used to Meetup.

Agreed. I'm working on a non-commercial open source Meetup alternative.

Misanthropic and Open Artificial Ignorance are misleading the public.

They are leeching off the work of mathematicians and pretending that new math was invented.

Misanthropic's entire existence is built around AI misuse.

Some friends I've known for over a decade clearly have ADHD, and I'm not going to tell them unprompted.

I feel like it is very disrespectful, especially when the conversation is about something completely different. Also, I'm not a medical professional, and it is possible that their ADHD-like symptoms are caused by something else.

That being said, if they ask if I think they have ADHD, or any other form of neurodivergence, I will say yes.

Is there a way to make this work with a wider variety of tools?

It appears to only be useful for React, and there are many other tools used for frontend development.

Can it use MyselfGPT?

How about putting the brakes on LLMs and prioritizing other types of AI.

When I think of AI, my mind goes towards automated Metro trains that do a great job of driving on fixed routes. On the other hand, it's obvious that the train can't do something like generate dinner recipies.

I'm not a fan of any large projects from Amazon, including data centers, warehouses, and the HQ2 office.

Cool, this project looks very interesting!

I tried uploading a .tex file created earlier this year and was able to see a PDF preview without any additional configuration.

As someone who has a local TeX setup, I'm a fan of your approach to loading dependencies on demand with a service worker. The installation process on my computer took a while because it was downloading a bunch of dependencies I didn't need.

I do most of my coding with a model called MyselfGPT.

If I ran a business, I would intentionally make it harder for AI search to view my website.

I agree with the headline, but I don't think LLMs are a reliable way of optimizing software.

First of all, coding speed is not the main bottleneck. I've been working on my own JavaScript framework. Most of my work is talking to people about challenges they have with web development, doing research on existing tools, and thinking about features to add.

Coding more slowly without LLM usage is a benefit because it helps me understand every line of code. The slower pace also helps me think about making things intuitive and reflect on my experiences with React.

While my framework is outperforming React, that isn't particularly impressive. On the other hand, it demonstrates how ignoring LLMs improves quality. Also, my monthly token spend is $0.

This site has excellent articles on learning to use native web technologies with minimal dependencies. I've found them to be useful resources when learning how to create web components.

https://gomakethings.com/guides/?filter=html

The article makes a great point.

Frankly, I think we should move to have harnesses do 100% of the work. In my view, the key is the fact that harnesses are deterministic tools created with a clear sense of intention.

I'm using a framework built from scratch with vanilla JS. It includes logic for writing templates.

I plan to publish an updated version soon.

I use web components with a render function to generate HTML and templates to provide fine-grained reactivity. Targeted re-rendering of items with relevant state changes makes performance significantly faster and also makes forms easier to write.

Here is an example of a template used to render a list of events.

const EventItem = () => {

  EventItem.eventTime = (eventData) =>{
    if(eventData.isRecurring){
     return `${eventData.dayOfWeek}s at ${eventData.nextEventTime}`;
    }
    return `${eventData.nextEventDate} at ${eventData.nextEventTime}`
  }

  EventItem.url = (eventData) => {
    return `/html/groups/event.html?id=${eventData.eventId}&groupId=${eventData.groupId}`
  }

  EventItem.location = (eventData)=>{
    return  `${convertLocationDataForDisplay(eventData.eventLocation)}`
  }
    
  return `
    <li>
      <a 
        class="btn secondary"
        {{href=url}}
        {{eventName}}
       >
       </a>
       <div 
         class="event-time" 
         {{eventTime}}
       >
       </div>
       <div
         class="event-location"
         {{location}} 
       >
       </div> 
      </li>
   `;
 }

 BaseDynamicComponent.defineTemplate(EventItem,"EventItem");

This is how the template is used in a component.

  <ul
    data-array=data
    data-template-name=EventItem  
  ></ul>

I actually agree with you. I was making the point that not using LLM assistance to code produces decent results.

Also, I'm planning to release my changes as a framework that can be used for a variety of interactive websites.

The field can be enhanced with JavaScript to support filtering or typo mitigation.

The grouped details element sounds very useful. I'm going to add it to my website.

I'm a 1 person dev team who wrote frontend code that outperforms React.

Is was successful because I didn't use any sort of LLM assistance.

The author's library is only 2.47kB.

This reminds me of watching JIRA go downhill over 10 years as it incorporated new technnology.

I used to be a scrum master who helped manage the team JIRA ticket board. When running into the inevitable problem during a meeting, I would sarcastically make a comment about "JIRA's monthly downgrade". By the time I stopped using JIRA in early 2025, it had become noticeably slower and had a bunch of UI clutter that made it difficult to do basic tasks.

I checked the homepage for Atlassian, which owns JIRA, and it has a headline saying: "Unleash your teams and their agents"