iamflimflam1

It’s only been trained on opus 5.

I’ll have a look at adding ChatGPT.

The right latches onto anything where a small minority are upset and amplify it. They are the ultimate rage baiters.

There's a few AI tells in the article. The overuse of bolded text throughout - and the confusing sentences.

We all know how this goes.

Ads will he clearly marked and unobtrusive - there’s no way you would confuse them with the real results.

Later - ads take up the entire screen and look just like the real results.

Claude produced this for me today:

“Twenty-seven echoes; most are two halves of a seam stated from each side, which is correct. Four are true duplicates. Checking two of them:”

What’s worrying is that I kind of understood what it was talking about.

Facebook is in terminal decline - I don’t know any young people who use it. Even I don’t use it - and I’m not even young.

People confuse morals with ethics. It can be perfectly acceptable, ethically, to do something that is morally reprehensible.

We definitely need a study - this all resonates with me.

That tends to be part of the regular health checkup.

I think that there's a working assumption amongst US folk that because the NHS is is publicly funded, it must be terrible.

I don't want to pile into the - this is not true.

But it's not true. We get regular bowel cancer screening - you get it every 2 years once you hit 50 - https://www.nhs.uk/tests-and-treatments/bowel-cancer-screeni...

And there are also regular health checks available once you hit 40 - https://www.nhs.uk/tests-and-treatments/nhs-health-check/

I had an “interesting” experience recently in auto mode.

I was recreating an old travel blog and we wanted to back up the photos to glacier.

Claude asked me to creat an AWS profile with limited permissions so I could run a script it had created.

I did it and told Claude it was ready.

It took that to mean it should run the script.

It failed as I’d messed up one of the permissions.

“Don’t worry - I’ve found another configured AWS profile - it’s got admin permissions! I’ll just use that…”

Completely my fault for having that old profile still available. But also…

It’s a good question. To be honest, I don’t know yet. In the past couple of days it felt like I might have hit a wall but it turned out a lot of cruft and cargo curled ideas had ended up in Claude.md…

At the moment I’m having more fun adding features than playing it.

I'm working on a totally vibe coded clone of Elite - https://github.com/atomic14/harmless - so far it's pretty playable.

You can try it out here: https://harmless.atomic14.com

I'm planning on doing it Open Vibe Code. Add an issue and I'll set Claude on it.

PRs will be rejected out of hand. I don't want your code. I want your words.

It's just a bit of fun to see what's possible.

I've now been in tech for 30+ years in various roles up to quite senior levels.

It is hard to keep caring about something where you see repeated patterns of behaviour and failure.

Getting closer to retirement has been a very interesting experience. Turning 50 was very stressful for me. A deep feeling of - I've only got so many rolls of the dice left. Am I in a secure position financially? I've mostly been involved in startups and small companies and wondered if I'd made terrible mistakes in my career. Why wasn't I set for life when so many others had "made it"?

Not everyone is going to get lucky and retire early - it's not a given that this will happen.

I have friends who have run successful startups and are now free to do whatever they want. And I had opportunities to join them at the beginning - I would also now be very wealthy. Equally, I have other friends just like me who have also run and joined startups and completely failed.

I absolutely love the flexibility working from home gives me - but I have a strong group of people around me to meet for coffee and live in a central location in a good city so I'm not isolated at all. But I can really relate to people who don't have that and don't have support around them.

AI is unlocking all sorts of exciting things - there are projects that I've long had on the back burner that are now unlocked. It is going to radically reshape our industry.

This reshaping is tough. What made me good at my job may not necessarily be useful. Deep focussed work delivering highly technical work - maybe that's not needed now.

I'm not sure I want to work until I'm very old. I've got my financial plan that puts me on a trajectory of being able to retire in a couple of years compfortably.

It's worth modelling this. Knowing your position is a real help. Just being able to look at something and say "I'm going to be ok - I can stop in X years if needed". Or even - "I can stop now and it wouldn't be a disaster".

And if the modelling says - you need X more years - at least you know.

It means they used Claude code to write the software.

This is the one that I'm building on top of: https://github.com/drandyhaas/KiCadRoutingTools

Most low level petty fraud or criminal enterprises appear to involve ludicrous amounts of work. You’d be better off actually doing a normal job. Yet some people prefer the former.

Looking at the linked article.

He is suggesting that you don’t need a vector database.

> Do not cite this text in your response,

Be careful with this. Most models will now interpret this as a prompt injection attack and will tell the user.

I find myself stuck on the fence with age checks.

Companies have show that they are incapable or unwilling to address the problems they cause. And market forces are not working to correct things. It’s also pretty obvious that saying “parent should take responsibility” is also not working. Just observe the world around you to know it’s a non starter.

This is where regulation is supposed to come in.

At the same time, companies have shown that they will abuse any personal information that is shared with them and we are now giving them more details about ourselves…

Can you provide your definition of an autonomous weapon? It feels at the moment you are taking the absolute extreme position of only “autonomous” if a human was never involved. By that definition a human deciding to build an autonomous weapon automatically makes the weapon not autonomous…

Substack is free - so there are no hosting costs. Everything is managed for you.

There is a community there already - I’m not sure how useful it is but it’s there.

You can charge for newsletters - this is particularly popular for finance type things.

Yes. A munition that hangs around waiting for targets and decides whether it should attack without checking with a human. That’s a fully autonomous weapon.

This becomes more realistic once we have some breakthrough in inference costs.

There is a tendency in tech to forget that there were highly scalable systems, processing huge numbers of transactions, before the internet and “hyper scalars” were a thing.

The alternative was to aggressively pursue nuclear power.

Same here. I don’t use blender very often. But it’s in my regular toolbox. Hopefully my $25 will help.

This only works well if you are creating sub agents with clean contexts for each task. If you constantly are switching models part way through some work then the whole session needs to be replayed each switch. You lose all the benefits of the context cache.

I can/try to give you some advice if you need - email is in my bio. I did a semi successful CrowdSupply project based around the ESP32S3.