David Whitney

Yeah that's pretty gross.

Nobody does a megastructure like the states. The discovery channel stuff was always mind-blowing as a kid.

100%

Person 1 "think I'm done, I won't offer the next round"
Person 2 *observes the breaking of the chain, slowly drinks up* etc
Everyone when finished, "right then"

People from other cultures "wtf just happened"

It's as amusingly British as it gets but we're conditioned out entire lives for subtext.

It's that - but make it everything :)

There's an anecdote I was told once by an American that moved to the UK and was struggling with pub culture and couldn't work out in the invisible conversation that happens when a group of people finish a round in a UK pub and silently decide to all leave via subtle social cues and finishing drinks.

I hate to be that European, but America as a whole, does not understand subtext or subtlety.

It's it's cultural trademark ;)

(I love America, sincerely, but it's objectively very funny)

Esssssh

Anti-Bribery Training

"You are approached by a partner to give them preferential treatment in a deal, should you:

1. Report it to your financial crime team.
2. Scream "hell yeah fuckos!" as you ride off into the sunset on your new yatch.

Correct answers only."

100% of corporate training material is this:

"Bobby and Jimmy are having a disagreement, should they:

1. Calmly discuss each other's opinions and feelings.
2. Get out a knife and have a deathmatch until one lies bloody on the floor
3. Jointly burn the office down.

80% pass rate required."

Hardest challenge for me is all the ghouls trying to remove the work of juniors and encode them into these tools. It's a pure capitalist lens of course.

There are no experts if you remove their avenue for practice. Which I guess is good for "me" personally but bad for everyone else.

Simultaneously one of the most interesting things in my three decades of programming and equally one of the most popularly and commonly misunderstood.

There's a reason why transformer models excel in domains where people can't trivially identify what good is - or where precison isn't required.

I get so much use from these tools - but I'm a 30+ year vet in my field and know when the tool is wrong so can discard it immediately.

But so much utility in triangulating answers, language shifting solutions - if you know the Lingua Franca if your domain you can navigate to good results fast.

When you understand that there are large categories of work where a general purpose computational approach for filtering through unstructured data is fundamentally an interesting and novel thing?

Very different problem space. Much more utility.

The marketing hype is all bullshit.

To understand the real world impact it's useful to forget what is being sold ("AI") and reframe as "Model Assisted Xyz" imo. If you get lost in the myth of precison "this will solve this thing" these tools will always fail because they're simply statistical models with integration points.

My personal opinion is that folks using these tools like electric screwdrivers see quite a lot of value - they're not outsourcing the thinking. (I have many examples of this in specific problem spaces).

The general genie in a bottle "make me this thing!" audience? It's a toy with negligible effect.

It's why the business world is struggling with ROI calculations especially - you can't isolate usage enough to know if there's any real casual relationship with perceived performance improvements, and even when there are it's a risk to attribute them to any one tool.

There have been numerous very bad studies that mostly just reflect confirmation bias (in both directions).

The problem with measuring these tools is that the kind of context isolation required for a study doesn't mesh very well with their average use.

And real world application is annecdoatal.

All this stuff blatantly backs onto green screen mainframe mulch.

All caps entry fields are often the giveaway. It's kind of gross how under invested in these critical systems are - and I'm sure that's nothing new to the technical people that are forced to hold the ceiling up.

Isn't this just pure over-exposure?

I can't think of any good reason to watch Marvel anything if the reviews aren't out of this world glowing.

Seen it all 20+ times, it's dull, pointless, junky at this point.

At the same time, the venn diagram of people that consume TV news in 2025 and people that watch GB News is one geriatric boomer circle.

Less representative than it seems.

I'm not paying a pound to read this, but no, AI models *literally* do not think or reason.

Could we not with speculative fiction as news content.

Yeah this stuff makes me yearn for Typescripts union types tbh.

Promise<Good | Error>

Is pretty tolerable. Shame about the obligatory Promise / Task wrap.

It's great, but my god those last two boss fights are a brutal difficulty spike upwards Vs everything before (if you got there).

Burn every church in the world xx

Audibly groaned at the finish pun.

Good job ;)

The Grease one in the park?

Don't get me wrong - I love my job, but I've been a programmer my entire life.

Spend a lot of my time doing leadership and management these days, such is my job.

However, spent all morning going to the mat with a team and doing deep technical work and my god I feel *so good* for it.

Easy to forget, but remember to do the thing you love.

I will never not be a programmer.

This is a little better but the presiding feeling is still "oh look, AspNet" rather than "this program does this".

I'm a huge fan of the DI by default approach, but prefer it all factored out.

Honestly, kinda wish there was just a container configuration API promoted by default.