tezza

Wait… tailscale connection to your own network, and unsupervised sysadmin from an oracle that hallucinates and bases its decisions on blog post aggregates?

p0wnland. this will have script kiddies rubbing their hands

You’re not alone. I do a small blog reviewing LLMs and have detailed comparisons that go beyond personal anecdotes. Gemini struggles in many usecases.

Everyone has to find what works for them and the switching cost and evaluation cost are very low.

I see a lot of comments generally with the same pattern “i cancelled my LEADER subscription and switched to COMPETITOR”… reminiscent of astroturf. However I scanned all the posters in this particular thread and the cancellers do seem like legit HN profiles.

this is truly bizarre.

It’s as if they’ve never heard of Maslow‘s Hierarchy of Needs before and further did they don’t know Self Actualizing is right at the very top.

Without that key stone on the top the human being is still a wanting animal. And if you somehow “mission complete” one Self Actualizing, then you immediately start to want something fresh “purpose” etc.

And obviously Self Actualizing doesn’t have to come in the form of work, although often it does.

and with types, it makes it easier for rounds of agents to pick up mistakes at compile time, statically. linting and sanity checking untyped languages only goes so far. I've not seen LLM's one shot perl style regexes. and javascript can still have ugly runtime WTFs

Yeah, it needs a steady hand on the tiller. However throw together improvements of 70%, -15%, 95%, 99%, -7% across all the steps and overall you're way ahead.

SimonW's approach of having a suite of dynamic tools (agents) grind out the hallucinations is a big improvement.

In this case expressing the feeback validation and investing in the setup may help smooth these sharp edges.

Had a nightmare getting a holiday only e-sim in Australia

Couldn’t set up easily because no wifi

then I just simply could not cancel the damn thing… It required being in Australia, and like the article needed a SMS code and the support was only contactable Australian working hours… who wanted the SMS code again.

So once back in Blighty there was no way… had to cancel the credit card to stop payments

So you are at the mercy of the competence of the provider

Yeah… even in the Web interface if you crack open Developer Tools and look at the json, the timestamps are all there, available in the data model. Those values are simply not displayed to the end user.

I was looking to write a browser extension and this was a preliminary survey for me.

There are remote team video games like (mine) https://epicwin.team .

Slightly different approach, each plays in their browser/phone and there is no “main display”. And my games are cooperative. DM if interested.

I’m releasing a new game BossBattle soon so keep your eyes peeled.

While this topic is here

Also see specialisms WITHIN Autism that are different to the mainstream Autism

The one I know most about is

PDA: Pathological Demand Avoidance [1]

PDA presents differently and needs very different strategies to mainstream Autism.

Main signs… kids under 12 attend school. However they explode at home or in private. At school the PDAers are masking (pretending to fit in) which is draining. When they get home the pent up frustration is released (explosively). So the family at home see a very different kid to the one that school/extended family witness. If this is an A-Ha! lightbulb moment for you or your child, see the questionnaire at the PDA Society[1]

[1] https://www.pdasociety.org.uk/what-is-pda/

Qwen won the latest models round last month…

https://generative-ai.review/2025/09/september-2025-image-ge... (non-pro Nano Banana)

If you’re interested in side by side analysis of carious image gen tools, i review them:

https://generative-ai.review/2025/09/september-2025-image-ge...

Main problem for me is that the quality tails off on chats and you need to start afresh

I worry that the garbage at the end will become part of the memory.

How many of your chats do you end… “that was rubbish/incorrect, i’m starting a new chat!”

Plugging away with reviews of Genrative AI tech with detailed comparisons. I announced the launch on HN a while ago, thought I’d use this month’s for a status update.

I just took Qwen-Image and Google’s image AIs for a spin and I keep a side by side comparison of many of them.

https://generative-ai.review/2025/09/september-2025-image-ge...

and I evaluated all the major 3D Asset creators:

https://generative-ai.review/2025/08/3d-assets-made-by-genai...

this happened to my mother-in-law, where she was the crasher.

in North London there is a large Turkish centre that hosts Turkish weddings. She was invited to a wedding there.

Traditionally, the bride and groom stand in the centre of the room and then family members lineup next to them all in a procession.

As you enter the room to reach the bride and groom, you must shake the hands in turn of all of the people in the procession.

When my mother-in-law eventually got to the bride and groom, they realised that the bride and groom were strangers. The accurate wedding was taking place upstairs at the same time.

There are multiple wedding venues in that particular Turkish Centre.

“Where is the shovelware?”… It’s Coming!

Changing domain to writing and images and video you can see LinkedIn is awash with everyone generating everything by LLMs. The posting cadence has quickened too as people shout louder to raise their AI assisted voice over other people’s.

We’ve all seen and heard the AI images and video tsunami

So why not software (yet but soon)??

Firstly, Software often has a function and AI tool creations cannot make that work. Lovable/Bolt etc are too flakey to live up to their text to app promises. A shedload of horror debugging or a lottery win of luck is required to fashion an app out of that. This will improve over time but the big question is, by enough?

And secondly, like on LinkedIn above: perhaps the standards of the users will drop? LinkedIn readers now tolerate the llm posts, it is not a mark of shame. Will the same reduction in standards in software users open the door to good-enough shovelware?

MS Windows has excellent multi window management with Alt Tab Win Tab etc. Far superior to others.

I have all my terminals with distinct icons and background colours to tell them apart. The operating system (Windows) does the heavy lifting.

i tried Mac for about five years but missed MS Windows “every window can be alt tabbed to”. Mac has “every app can be command tabbed to and therein each app has its own subwindow management”

Add to the list:

https://x.com/rowancheung - Rowan Cheung: “Daily” updates and insider access

Short answer: Not yet

Longer answer: It can do an okay job if you prompt it certain specific ways.

I write a blog https://generative-ai.review and some of my posts walk through the exact prompts I used and the output is there for you to see right in the browser[1]. Take a look for some hand holding advice.

I personally tackle AI helpers as an 'external' internal voice. The voice that you have yourself inside your own head when you're assessing a situation. This internal dialogue doesn't get it right every time and neither does the external version (LLM).

I've had very poor results with One Stop Shop builders like Bolt and Lovable, and even did a survey yesterday here on HN on who had magically gotten them to work[2]. The response was tepid.

My suggestion is paste your HN comment into the tool OpenAI/Gemini/Claude etc, and prefix "A little bit about me", then after your comment ask the original coding portion. The tool will naturally adopt the approach you are asking for, within limits.

[1] https://generative-ai.review/2025/05/vibe-coding-my-way-to-e... - a 3D scene of ancient pyramid construction .

[2] https://news.ycombinator.com/item?id=44513404 - Q: Has anyone on HN built anything meaningful with Lovable/Bolt? Something that works as intended?

Q: Has anyone on HN built anything meaningful with Lovable/Bolt? Something that works as intended?

I’ve tried several proof of concepts with Bolt and every time just get into a doom loop where there is a cycle of breakage, each ‘fix’ resurrecting a previous ‘break’

Simon’s blog is high-signal, low-noise deep dives. The very opposite of blog spam

Thoughts all the way down?

I’m working on a review website for Generative AI. Targeted at people who want to use GenAI to build stuff.

e.g.: Following up on one of my HN comments on OpenAI ImageGen gpt-image-1 quality: Side by side comparison of more challenging prompts at Low/Medium/High:

https://generative-ai.review/2025/04/apple-a-dog-how-quality...

Now that really would be Kafka-esque

For the curious I generated the same prompt for each of the quality types. ‘Auto’, ‘low’, ‘medium’, ‘high’.

Prompt: “a cute dog hugs a cute cat”

https://x.com/terrylurie/status/1915161141489136095

I also then showed a couple of DALL:E 3 images for comparison in a comment

Precisely. The tools often hallucinate: including in its instructions higher up even before your prompt portion. Also the behind the scenes stuff not show to the user during reasoning.

You see binary failures all the time when doing function calls or JSON outputs.

That is… “please call this function” … does not call function

“calling JSON endpoint”… does not emit JSON

so from the article the tool generates hallucinations that the tool has used external stuff: but that was entirely fictitious. it does not know that this tool usage was fictitious and then sticks by its guns.

The workaround is to have verification steps, throw away “bad” answers. Instead of expecting one true output, expect a stream of results which have a yield (agriculture) of a certain amount. say 95% work, 5% garbage. never consider the results truly accurate, just “accurate enough”. Verify always

cool, that would work too. my preference is leading separators so the separators are all in a visual column. being in a visual column allows the eye to discount the separators easily.

typically names are different lengths and the commas are hard/harder to spot

Your suggestion:

  SELECT
     first_column,
     second_column_wider_a_lot,
     (third + fourth_combined_expression),
  1 as pad FROM

vs my current preference:

  SELECT 1 as pad
     , first_column
     , second_column_wider_a_lot
     , (third + fourth_combined_expression)
  FROM

> I would fire and forget any developer who tried to pull that stunt.

A tad bit harsh there?

it is a trade off.

clarity and ease during design time versus slightly uglier but still consistent code as a work around. miniscule energy overhead

it is a trade-off

reordering, insertions and deletions much easier within the IDE as you’re manipulating lines as a whole. focus maintained while you are in the zone or tired is the biggest gain

at the cost of slightly uglier code and a miniscule… truly miniscule wire overhead.

other decisions will have much larger wire impact, say choosing a column wider than necessary

as mentioned elsewhere, i personally introduce a pad element to get fire and forget consistency

  SELECT 1 as pad
    , a
    , b

I do this as well.

on top i often do a pad entry so that the elements are all on their own line

  SELECT 1 as pad
    , a
    , b

then i can reorder lines trivially without stopping to re-comma the endpoints or maintain which special entry is on the line of the SELECT token

what would be helpful is both LEADING and trailing commas

so I am suggesting:

  SELECT
    , a
    , b

would be permissible too. The parsing step is no longer the difficult portion.

Developer ease leads to less mistakes is my conjecture.