HNHacker News
TopNewBestAskShowJobs

bfung

1,856 karma · joined June 10, 2009

https://bfung.github.io/

https://benson-guides.blogspot.com/

fung [dot] benson [at] gmail [dot] com

submissionscomments
bfung··on GitHub Is Degraded
Not purely an advertisement, it’s a real app! https://www.datadoghq.com/blog/updog-ai/

But it does help with marketing and showing off our tech.

bfung··on Where things stand with the Department of War
I don’t feel that old, but I guess being 45 is ancient in tech.

The Silicon Valley tech jobs we have now has a history rooted in World War 2 and funding of it by the US gov.

https://youtu.be/ZTC_RxWN_xo?si=gGza5eIv485xEKLS

I’m not saying war is good or anything, but also don't ride a high horse cause none of it would be here w/o WW2.

bfung··on Where things stand with the Department of War
I’d argue it’s come full circle and it hasn’t changed a bit.

There wouldn’t be a Silicon Valley without World War 2 and US gov. funding of Stanford to develop radar basically.

The initial investment from then gave critical capital mass for Stanford, the VCs, and the tech companies of today.

https://youtu.be/ZTC_RxWN_xo?si=gGza5eIv485xEKLS

bfung··on When AI writes the software, who verifies it?
When humans write the software, who verifies it?

half sarcasm, half real-talk.

TDD is nice, but human coders barely do it. At least AI can do it more!

bfung··on Sizing chaos
Hot take (?):

The random sizing today is great:

* If you want a better fit, go physically to a store instead of shopping online and try them on.

* the vanity part is also fine, no need to cause outrage at raising the number and making people depressed cause they think they're even more "fat". It doesn't need to be "optimized"

* Only serves online retail to "standardize", but guess what, 15th standard also sucks... <cue xkcd comic about standards>.

Enjoyed the presentation of the site. :)

bfung··on Gemini 3 Deep Think drew me a good SVG of a pelican riding a bicycle
In the spirit of Winter Olympics, I vote “Lion on a bobsled” next bench . :)
bfung··on Ex-GitHub CEO launches a new developer platform for AI agents
`/init` is good enough.

Model improvements will take care of the rest.

bfung··on Experts Have World Models. LLMs Have Word Models
Or the opposite, that humans are somehow super special and not as simple as a prediction feedback loop with randomizations.
bfung··on The silent death of good code
If it’s easy to read and understand but doesn’t work, or is slow to execute, or costs a lot to run, is it good code?

If the function is a black box, but you’re sure the inputs produces a certain output without side effects and is fast, do you NEED “good code” inside?

After about 10yrs of coding, the next 10 of coding is pretty brainless. Better to try and solve people/tech interaction problems than plumbing up yet-another-social/mobile/gaming/crypto thing.

bfung··on The Great Unwind
Dunno about the website or corp, but the occupy wall st movement was/is true. Happened right after 2008 stock market crash and people camped out on Wall Street in protest of bailing out the banks.
bfung··on AI code and software craft
> If you're fine with software which gets a little bit harder to work on every time you make a change and which might blow up in unexpected ways,

human written code is similarly fine. Save for very few human individuals.

bfung··on AI code and software craft
Is clothes today really worse?

We have clothes and materials like gortex now that blocks rain and snow no handmade jacket could ever hope to perform at the same level to be lightweight AND dry.

bfung··on I was a top 0.01% Cursor user, then switched to Claude Code 2.0
use both ~ Claude Code as main driver, hook it up to Cursor with /ide in Claude code to review or make other manual adjustments.

Have OpenAI Codex do code reviews, it’s the best one so far at code reviews. Yes, it’s ironic (or not) that the code writer is not the best reviewer.

bfung··on Software engineers can no longer neglect their soft skills
Totally agree.

The uncomfortable truth is that the skill of _writing_ code is becoming commoditized.

Reading, understanding, designing efficient systems, and planning changes based on that, at least for now, will still be for human experts (and AI already a great assist here).

But churn out yet another webpage/website? People doing this will need to move on from this as their primary job.

bfung··on 2026: The Year of Java in the Terminal?
Article meta comment: show me, don’t tell me.

No show —> not easy enough —> too lazy to look it up, already got python/go/ruby quick cli methods. LLM generated even easier.

bfung··on Things that aren't doing the thing
Doing the thing that nobody wants is doing the thing.
bfung··on Google to merge Android and ChromeOS in 2026
Sooooo ChromeOS 26 with Liquid -Glass- Material Design? :trollface:
bfung··on LLM Observability in the Wild – Why OpenTelemetry Should Be the Standard
TL;DR - follow https://opentelemetry.io/docs/specs/semconv/gen-ai/
bfung··on Stop writing CLI validation. Parse it right the first time
Looked human-ish to me, what signs did you see?
bfung··on AGI is an engineering problem, not a model training problem
Yep, the Attention mechanism in the Transformer arch is pretty good.

Probably need another cycle of similar breakthrough in model engineering before this more complex neural network gets a step function better.

Moar data ain’t gonna help. The human brain is the proof: it doesnt need the internet’s worth of data to become good (nor all that much energy).

bfung··on If we can find information by asking GenAI, who needs the Web?
What new data will GenAI train on if there’s no Web?

a nothing burger article with no real ideas.

bfung··on Ask HN: Do you struggle with flow state when using AI assisted coding tools?
no need for flow, just vibe with it. /s
bfung··on Show HN: A GitHub Action that quizzes you on a pull request
That was my first reaction: now I gotta build a gpt wrapper, oops, I mean agent, to answer questions to this quiz
bfung··on What went wrong inside recalled Anker PowerCore 10000 power banks?
^^^ is closer to the truth for the swap out; eek out those margins.

Culturally, chabuduo is more of an excuse/greed of the lazy to convince you it’s close enough [that it won’t make a difference to you] and they’re too lazy or out of patience, for many reasons they also don’t want to explain to you.

bfung··on What explains the liberal-conservative happiness gap?
Ignorance is bliss, right? :)
bfung··on The Grug Brained Developer (2022)
Coding too hard, this grug use AI.
bfung··on Seven replies to the viral Apple reasoning paper and why they fall short
Any links or examples available? Curious to try it out
bfung··on Stack Overflow is almost dead
To add a bit more nuance, SO has a question-answer type format, which leads very well into prompt-rely format to train these chat applications. Most of the other sources do not, except for Github issues maybe. Without this question-answer format, there'll be a need for a bigger data labeling effort to train LLMs on new stuff, no?
bfung··on Stack Overflow is almost dead
Where will training data come from for new tech & programming languages if SO dies?
bfung··on Human
Enjoyed the read, has a similar vibe to Asimov's The Last Question (https://users.ece.cmu.edu/~gamvrosi/thelastq.html)
← PreviousPage 2 of 27Next →