Frontend development rescue - I can help
Glasgow
Joined January 2008
- Tweets29.1K
- Following38
- Followers527
- Likes846
A warning to vibe coders.
I'm using Claude Code to build a football match predictor. I wrote a proper plan for it. Trained model, learned parameters, evaluation pipeline. The plan lives in the repo, and I told Claude to read it every session.
When I asked it to build the model, it ignored all of that. Gave me a Poisson simulation with hardcoded multipliers instead. No training step. No learned weights. Just `home_lambda *= 1.05`.
The scary part is that it worked. Real data, plausible predictions, clean API. I would have shipped it if I couldn't read the code myself.
If you can't verify the output matches what you actually asked for, the agent is driving, and you're just watching.
A warning to vibe coders.
I'm using Claude Code to build a football match predictor. I wrote a proper plan for it. Trained model, learned parameters, evaluation pipeline. The plan lives in the repo and Claude reads it every session.
When I asked it to build the model it ignored all of that. Gave me a Poisson simulation with hardcoded multipliers instead. No training step. No learned weights. Just `home_lambda *= 1.05`.
The scary part is it worked. Real data, plausible predictions, clean API. I would have shipped it if I couldn't read the code myself.
If you can't verify the output matches what you actually asked for, the agent is driving and you're just watching.
#vibecoding #claudecode #claude
I don't think a lot of people get that LLMs are all about plausible and not about accurate #LLM #Claude #claudecode
I wrote about AI coding — the good, the bad, and the downright dangerous. Hold on to your hat!
frontendrescue.com/posts/202…
#Claude #ai
The duplication in LLM code is something I find staggering. I asked CLAUDE why LLMs default to duplicating code rather than defaulting to existing code, the reply is very interesting.
LLMs generate token by token based on pattern matching.
Writing new code is easier than understanding existing
code and figuring out where to integrate.
We don't truly reason about architecture. We see the
immediate task and generate a solution. Checking
existing files, understanding their structure, and
extending them requires more cognitive steps than just
writing something new.
Training data is mostly standalone examples - "here's
how to do X" - not "here's how to integrate X into an
existing system".
So the default is always to generate, not to understand
and extend.
#ClaudeCode #LLMs
AI builds fast. Then you pay for it. AI knows how to build. Knowing what not to build is still a human skill
Here is what I am finding
frontendrescue.com/posts/202…
#claudecode #vibecoding
When the AI hype finally settles, and the generated code actually lives in production, things will burn. Give it 6 months
Not because AI “can’t code”, but because it produces volume without precision and ownership.
Huge diffs, duplicated logic, silenced types, unreadable PRs — all fine in demos, expensive in reality.
AI is a tool. Codebases still need restraint.
Math in motion—polar equations turning angles into symmetry and pattern.
cutting.scot/viz/polar
#typescript #react. #viz #dataviz #javascript
dagda1 retweeted
Reminds me of some similarly cool visualizations that @dagda1 put together, e.g., cutting.scot/viz/sinusoidal
dagda1 retweeted
Some tips on how to learn math independently:
1) Active Learning
You don't become proficient in math just by reading books or watching videos like those from 3Blue1Brown or Khan Academy.
This is known as "passive learning."
To truly grasp mathematics, you need to engage in "active learning," which means solving lots and lots of math problems.
For more advanced learners, this also includes writing lots of proofs.
While many recommend 3Blue1Brown as a good starting point for advanced topics like linear algebra, I believe this is misguided.
Their videos are beautifully presented and enjoyable (I love them myself), but they are most beneficial _after_ you have a solid understanding of the underlying material, not before.
dagda1 retweeted
if you need to learn Linear Algebra or Calc for applied reasons (i.e ML)
don’t use a textbook, it’s time inefficient
here’s what you do:
- buy a 5 pack of these pentel pen they’re so smooth (optional but recommended)
pentel.com/products/energel-…
- get a ream of printer paper.
the cheapest ones possible
a.co/d/hIrhTyN
- go to mathacademy.com
- focus and grind for 4-5h / day
(diminishing returns past this point, i assume you have other things you need to allocate time to)
- do it daily
it’s not bait edutainment like unlike brilliant et al. you must actually put in work.
theres a topic dependency graph that serves topics based on your knowledge level
spaced repetition is also built in
it is expensive and the marketing does not have good vibes
but outside of getting a broke math postgrad to 1:1 you, nothing beats it
This Post is from an account that no longer exists. Learn more
I wrote this so that my daughter can remember what a,b,c and d are in
y = Asin(Bx-c) + d
cutting.scot/viz/sinusoidal
#react #typescript #d3 #dataviz
you just get immune to the crazyness of frontend development. All this messing about with adding and removing .js extensions for imports with moduleResolution Bundler and nodenext in .ts files is bonkers #typescript #javascript
there are lots of development courses I would love to do but once I see it is a bunch of videos, I don't buy.
Watching videos does not encode any worthwhile memories rendering the whole thing useless.
Videos are the worst courses