Vibe Coding and Bluesky

Ash:

But you feel bad if you relax, don't you? It's just it's

Ian:

not a done thing, is it?

Ash:

No. No. Because your brain's like, well, I could be doing this.

Ian:

Capitalism was all like, no. You must work more.

Ash:

Yeah. Exactly. It's like, why are you failing me?

Ian:

You must work like Elon even though you don't quite have his compensation.

Ash:

Yeah. And, apparently, I can't remember where where I read the article, but it was, like, basically CEOs. Apparently, it's the job that you get if you don't want to do any work. So none of them might go, like, and do anything. They just I don't know what else they do.

Ash:

But, apparently, they don't go to work.

Ian:

They steer. They steer.

Ash:

They they

Ian:

stop us from going off a cliff, normally. Yeah. Usually. Yeah. Well, sometimes.

Ash:

Just every few months issue a back to the office mandate.

Ian:

That's what we need.

Ash:

And, you know, the job's done then, isn't it?

Ian:

None of the namby pamby whatever. Work from home. Snowflake.

Ash:

Snowflake. Yeah. I I can't really speak in those because I I don't

Ian:

You can. You just have to think Yeah. In between each word.

Ash:

Yeah. The right wing spiel doesn't just doesn't come out.

Ian:

Well, yes. I think there's probably a lot of resistance Yeah. Yeah. To even ironically Well, yeah. Doing the right wing spiel.

Ash:

Yeah. I can do other spiels. You can, Ash.

Ian:

You can.

Ash:

I can.

Ian:

I think any listener to what a lot of things will know that you can do other spiels.

Ash:

The other spiels are available.

Ian:

Really? Really. We're starting to learn to speak clanger now, aren't we?

Ash:

Yeah. What's a clanger, Ian?

Ian:

Well, I'm glad you asked me that. It's a nineteen seventies TV character.

Ash:

Well, every now and then maybe we should do a Reboot.

Ian:

Refresh. Re rebooted in the February or February or something.

Ash:

Yeah. And then reimagined for what lot of things?

Ian:

Yes. Reenightmared. Yes. I think its main legacy is the loud whistling in between the segments of this show.

Ash:

Yeah. Yeah. I'm looking forward to explaining what a clanger is to the attendees of the Leeds testing atelier.

Ian:

Why would we be explaining it to them, Ash?

Ash:

Well,

Ian:

we story.

Ash:

There is a story. We might, as in we will, one way or another.

Ian:

One way or another. Something.

Ash:

Be recording a live episode.

Ian:

A live episode.

Ash:

Well, you if you're a podcast listener, unless you're there as well, you won't hear it live. It's not that live.

Ian:

I'm not going to, now volunteer to make it liver. Yeah. No. Tempting, though it in fact is.

Ash:

So it's with a live audience rather than it being a live episode. It's probably a more accurate.

Ian:

Well, we we'd like it to be accurate.

Ash:

Yes. We're keen on accuracy. Consistency, discipline, and accuracy. That's what the whole thing's the whole heart.

Ian:

The hallmarks of what a lot of things where we never have to take back anything I've said because I always remember all my anecdotes perfectly.

Ash:

Yeah. All those quotes that you don't know who said them. Yeah. But you know the quote.

Ian:

Yes. Like an authoritatively say the wrong person said them. Or we can discover that while everyone thinks they did say them, in fact, it's not true.

Ash:

There's quite a few of those, though, isn't there? Yes. That are attributed to, various figures of history for them.

Ian:

Oh, well, I've oh, my emails at the bottom, I had, Einstein's thing that he didn't say, which was, things should be as simple as possible, but no simpler. And I always enjoyed that very much, but who knows who said it? Yeah. Someone.

Ash:

Someone. It's I mean, it's it's wise words still.

Ian:

Einstein said something about, I don't know, quantum something.

Ash:

Quantum something something. Which What would Einstein have made of your quantum computing episode?

Ian:

I'm sure. I'm sure.

Ash:

I'm I'm completely like Yeah.

Ian:

It's nothing to do with to do with you, was it? He didn't say a word during that whole thing. I think he would have understood it better than either of us.

Ash:

I I would agree.

Ian:

In fact, we would have said to him that there was such a thing as a quantum computer, and then he would have said, oh, that's a good idea, and then he would have just figured it out for first principles. Yeah. Pretty much. So just in terms of not looking stupid, Einstein is never going to be invited onto this podcast.

Ash:

Yeah. What's the other quote? Oh, it's the Henry Ford one, isn't it, that he didn't say about

Ian:

A faster horse. Yeah. Yeah. Which Sorry. You were gonna say the whole thing.

Ian:

You should say the whole thing.

Ash:

No. No. It's okay.

Ian:

Because I've said the words a faster horse. I'm like

Ash:

It's a great Nearly

Ian:

everyone's like, oh, yes. I know that one. And then there were a few people who were like, what?

Ash:

What? Yeah. Henry Ford wanted everyone to get a faster horse. Right?

Ian:

Yes. Yes. You know he was a hay salesman before he took up. He sold special hay.

Ash:

Special hay.

Ian:

He was laced with cocaine. It made the horses run really quickly.

Ash:

Alright. Okay.

Ian:

Moving on. Moving on.

Ash:

Moving on from cocaine laced hay.

Ian:

It's not a a phrase I ever No. Thought of myself as being about to say at any particular point in my life.

Ash:

So apparently, Henry Ford wanted to he wanted to, give people a a better working week, shorter, with less days.

Ian:

What sort of capitalism was he?

Ash:

See, this is the interesting thing. He wanted he wanted a shorter week for people so they would come and work for him, but not work for anyone else. So he could maintain his monopoly. So it was still capitalism.

Ian:

Oh, okay. So should we do the then? The mister Burns. Wow.

Ash:

Yeah. So, yeah, I just found that really interesting because, you know, he he was definitely in the in the arch capitalist, space. Pantheon. Pantheon. Oh, yeah.

Ian:

There's a word that that's a word that's rising to the point where we now notice

Ash:

Yeah.

Ian:

Saying it after the last episode, which we can remember because it went live. I finished editing it last thing on Monday, and it went live 5AM on Tuesday

Ash:

Exactly.

Ian:

Giving Ash a lot of time for his listen to his pre release listen that he normally does. Yeah. Although to be fair, he doesn't normally then provide feedback along the lines of you've allowed a catastrophic sound to be

Ash:

So, Ian, could you make a catastrophic sound for me, please?

Ian:

That was that wasn't what I had in mind.

Ash:

Can you imagine if if one of the four horsemen of the apocalypse was called Kinevan. And that was the cry that broke open the sky when at the end of at the end of days.

Ian:

But which one would you change into it? Because obviously there's famine, pestilence, war, and death. So which one which one will have their scope slightly reduced to that of Cenevin?

Ash:

Let's get rid of famine. Well, I'm sure we could use the Cenevin framework for the end of days.

Ian:

Well, I'm sure it will be chaos. Yeah. Mind you, this might be the end of days.

Ash:

Yeah. It's the end of this day.

Ian:

Oh, yeah. It is the end of this day. I'm quite tired. I did a a run a workshop in the morning.

Ash:

Fertilization is tiring.

Ian:

It is tiring. It was design thinking, and it was remotely delivered to, teams in in India. So it was quite it was quite the epic adventure. It was really good, actually. They they're great proud.

Ash:

Design thinking, Please see

Ian:

That's our idea of birds. It there we did do an episode about it.

Ash:

We did.

Ian:

I'd probably know more. I feel like we should be going back over our back archives and listening to the things we said about

Ash:

stuff Wow. Yeah.

Ian:

And then redoing them right or better.

Ash:

Well, I think we made a we did the best what what what's the what's the retrospective prime directive? It's everyone does the best job that they can with, like, with what they know at any given time.

Ian:

Or what a their AI knows.

Ash:

What their AI knows.

Ian:

Or rather what their AI can express Yeah. With a in a very confident manner. Yeah. Speaking of AIs. Speak speaking of AIs.

Ian:

Oh, someone's gotta go first.

Ash:

Yeah.

Ian:

We've got to do things.

Ash:

We do need to do things.

Ian:

That's why we're here.

Ash:

I'm quite excited about this thing as well.

Ian:

Well, I can't remember who went first last time. I think it was me.

Ash:

Yeah. Yeah.

Ian:

Because So because because the episode recording time ran for an hour and thirty eight minutes or something like that.

Ash:

And an hour and ten minutes of that was your thing. That's my thing. Yes.

Ian:

And, that was a more interesting thing than I had anticipated. But I had to edit out a massive chunk of it just to make it go down to under an hour and a quarter.

Ash:

That's the new benchmark now, isn't it?

Ian:

Well well

Ash:

Before the benchmark was an hour.

Ian:

Yeah. It's now an hour

Ash:

and a quarter. Yes.

Ian:

Wind forward.

Ash:

Yeah. We said we

Ian:

are never doing an an episode that's over six hours.

Ash:

Yeah. In 2050.

Ian:

Yeah. Yes. 2050. Scoped Creek will have overcome us.

Ash:

Yeah. Welcome back to this marathon episode. We we've stopped editing, and now we just broadcast.

Ian:

Yeah. That's that's literally the only way we can and even so, our, our time is getting away from us. And so we're operating constantly and with an ever decreasing amount in the past.

Ash:

So, basically, Ian goes to sleep for a bit during these marathon broadcasts, and then I just start a rant about, I don't know, Figma, story slicing. What

Ian:

Oh, whatever equivalent they're having those

Ash:

days. Exactly. And then Ian gets a couple of hours sleep. And then once my rant is finished, we can begin the podcast again.

Ian:

Yeah. The next episode. Oh, honestly, I think this is a lot of fun, but I'm not sure I would do it full time. Not literally full time anyway.

Ash:

It'd be quite a terrifying life, wouldn't it? Just being picking out things to have hot takes about continuously. Continuously.

Ian:

Well, I guess that's just being an influencer.

Ash:

Yeah. That does sound like being an influencer, doesn't it, actually? So, yeah, I don't wanna do that.

Ian:

No.

Ash:

No. So may I talk about a thing?

Ian:

Well, I suppose, grudgingly, that since I did talk about presentation tools for remarkably long time last time, that, it is right and proper that you should now talk about a thing. Yeah. But first

Ash:

But first.

Ian:

What is your thing?

Ash:

So my thing What?

Ian:

Sorry. Sorry. It's a little thing in my throat there.

Ash:

My thing is called vibe coding.

Ian:

Is that your thing? Really? Yeah.

Ash:

Although, I've never done any vibe coding.

Ian:

Have you not?

Ash:

No.

Ian:

I thought all coding with AI was vibe coding.

Ash:

Yeah. Yeah. So I think we'll get to that.

Ian:

Is that a popular misconception?

Ash:

Shall I explain the origin of vibe coding first?

Ian:

I think that's an important place to start from, actually Yeah. Because it what it's morphed into is is is it's supposed better to think of it in that Yeah. Way, isn't it?

Ash:

I mean, it's got more entertaining, but it wasn't the, the original intent

Ian:

I'm sorry.

Ash:

The phrase vibe coding.

Ian:

Okay. So let's let's do that then. What what? Vibe coding?

Ash:

So where did I first notice this? So there was a a TechCrunch article about a company that has received $20,000,000, I assume, in funding or maybe even pounds.

Ian:

20 million gold bars.

Ash:

Gold bars to fix problems caused by vibe coding, to which I thought, I guess the first thing I need to know is what is vibe coding, and why does it need to be fixed?

Ian:

Fair comment.

Ash:

What problems does it cause that require the fixing?

Ian:

Are people vibe coding firewall firmware?

Ash:

Well so I started to dig around vibe coding as one does in preparation for a lot of things. So a chap called Andre Karpathy, excuse my, pronunciation.

Ian:

He's a luminary of the AI world, isn't he?

Ash:

He is. He is. He's got some interesting thoughts.

Ian:

And he's author of some interesting papers as well or co author

Ash:

Yeah.

Ian:

In the history of modern AI. So he's quite an authority.

Ash:

Yeah. So on x Twitter, whatever, he posited this definition of vibe coding. I might not read it all, but it's quite fun. Oh. So there's a new kind of coding I call vibe coding, where you fully give in to the vibes, embrace exponentials, and forget the code even exists.

Ash:

It's possible because the LLMs for example cursor with sonnet are getting too good and I just talked to composer with super whisper so I barely even touch the keyboard. I asked for the dumbest things like decrease the panning on the side by by half because I'm too lazy to find it. I accept all always. I don't read the diffs anymore. When I get error messages, I just copy and paste them in with no comment.

Ash:

Usually, that fixes it. The code grows beyond my usual comprehension. I'd have to really read through it for a while. Sometimes the LMS can't fix a bug, so I just work around it or ask for random changes until it goes away. I like that one.

Ash:

Yeah. It's not too bad for throwaway weekend projects, but still quite amusing. I'm building a project or web app, but it's not really coding. I just see stuff, say stuff, run stuff, and copy and paste stuff, and it mostly works.

Ian:

Wow.

Ash:

How do you feel about vibe coding?

Ian:

I feel very comfortable with vibe coding by that definition.

Ash:

If

Ian:

he had said I see stuff, say stuff, run stuff, copy paste stuff, and then push it to, mission critical production. Then just push it

Ash:

to the air traffic control system.

Ian:

I would have more reservations about it. Yeah. But that just sounds like fun. Okay. Yeah.

Ian:

I mean, it is kind of fun, isn't it? Yeah. I it is sort of miraculous watching an AI produce some amazing UI just out of nothing.

Ash:

Yeah. So have you ever done vibe coding?

Ian:

Only miss only accidentally.

Ash:

Oh, interesting. So it's possible just to not realize that you're doing

Ian:

it right. Well, what happened what so in my instance, I recently I'm I don't think I quite have vibe coded it, but I wrote a simple app Yeah. For the Ilkley Live busking festival

Ash:

Mhmm.

Ian:

Because people, who shall remain nameless, said to me, we are launching five busking spots in Ilkley. Yep. And we need we want to have a festival on a day to launch them, and we want a booking system for this festival. Is there something we could do with Google Forms or something?

Ash:

Wow. Looking at you.

Ian:

Well, possibly having some kind of eye contact. And I said, okay. Well, I'm sure we can do something with Notion or Notion forms or something like that. Yeah. So I went away and had a go.

Ian:

Give it a go. Give it a go. Go. Go. I I had a go.

Ian:

Theme with

Ash:

with AI, isn't it? That's the

Ian:

national anthem of vibe coding.

Ash:

We'll have a go.

Ian:

And, basically, after an hour and a half of messing around with Google Form, not it was Notion and some other things, I thought to myself, basically, this is gonna require full time hoots human supervision to deconflict people who've booked the same spot at the same time. Yeah. And I can't find a way to easily manage that Okay. Through the software without writing software. And then I thought, oh, writing software.

Ian:

And so I got my Claude code out. And about another hour and a half later, I had an app that did this that I could deploy to Vercel Yeah. With a database that cost me a dollar a month or something. And I was delighted with that, and then I spent a really long time polishing it.

Ash:

So you lost the vibe.

Ian:

Well, I the vibe coding is great Yeah. Providing that you don't have any opinions at all about how your application should work. Yeah. But if you have opinions about how it should work, then vibe coding is a, shall we say, extended discussion about whether you whether it's actually done your way of doing it or not.

Ash:

Yeah.

Ian:

So this is probably the closest thing I've done to making an app that was vibe coded, if you like. Yeah. But the things where it went wrong and caused problems was when I did what he said here. So not reading the diffs.

Ash:

Yeah.

Ian:

So or not reading it properly or not reading it and saying, well, hang on a second. We already do something else over here Mhmm. That we should reuse here Yeah. Or something like that. If you can ignore all of that and your app's not that important and doesn't have any real security

Ash:

Yeah.

Ian:

Or privacy concerns, you're not storing people's data

Ash:

Mhmm.

Ian:

Then it's probably fine to entirely vibe code something like that.

Ash:

Yeah. Yeah.

Ian:

But I wanted to put people's phone numbers and email addresses in my database. And in order to do that, it wasn't really practical. I just let the AI decide how to cure it.

Ash:

Not, like, vibe your way through it.

Ian:

No.

Ash:

So one of the things that interests me is at the moment at work, they're doing, like, a rollout of Copilot.

Ian:

Is that the one that everybody says sucks?

Ash:

Yeah. But Okay. I mean, regardless of the tool Yes. What you said about slipping into vibe coding

Ian:

Well, exactly.

Ash:

So I guess that's like the danger, isn't it? Is that you might you might start with the intentions and opinions about how things should be. But then in your weaker moments, you might slip into just going with the vibe.

Ian:

Well, the thing is that you just think, oh, you decide to trust it and you you can't really ever trust it. And you can't trust it with your software architecture design. No. And and that that's the thing. So I I, for example, when I I'm slowly building my list of things that I do when I'm making Next JS apps Yeah.

Ian:

With React and server components and, and server actions and what have you. But I've built a library that I use in all my apps for passing state back and forth between forms

Ash:

Okay.

Ian:

And the server. I want to do that. I want to use that. Yeah. I don't want to I don't want it to invent a new way of doing form every time.

Ian:

Yeah. And half the time it says right well we'll just install react hook form now and I'm like no you won't. Right. Okay. And then then it sometimes it it says oh I'm just gonna make up a type Yeah.

Ian:

For passing the data into it. Okay.

Ash:

Yeah. Yeah.

Ian:

Which doesn't do all the things that I want And getting it to do the thing I want is quite difficult.

Ash:

Mhmm.

Ian:

And, actually, what I've found is I have to use templates. So I'll make a template server action and a template form that have scaffolded Yeah. How I want it to be. And then I put those in place, and then I say, make this have these fields and do this. And it and it kind of builds it up from the foundation that I want to be there.

Ian:

But you can't let it just go ahead and do the form because it will do something completely random.

Ash:

Yeah. Yeah. It's interesting the distinction there between that. Say if you just I don't know. Adding some error handling to an API endpoint, say, versus, like, maintaining your the wider design patterns.

Ash:

So you got, like, code versus design patterns argument where it's like, well, maybe you can vibe code something, but you can't vibe design something.

Ian:

No. And I think there may be a thing about context windows here.

Ash:

Mhmm.

Ian:

And some AIs are now starting to have very big context windows where you could almost say, oh, we're just gonna send the whole repository to them. Yeah. And in that case, it might be a lot better about following the rules and the patterns that you've set up. Yeah. But in general, it's it's a bit messy.

Ian:

And and I totally get this idea of vibe coding. I think it's fun. And if you're just messing about, go nuts. But professional software development is a whole different Yeah. Game than that.

Ian:

Yeah. And I think you're right to highlight this idea that a developer with the best of intentions will become tired, start trusting it, clicking okay. And I find a lot of the time I'm saying I'm stopping it in mid flow and then saying that thing I just approved actually Right. Okay. Because I see it after I've already noticed things about it, after I've already pressed enter Yeah.

Ian:

Or pressed approve. And that, you know, you sort of think, how many times have I not done that? How many times have I not noticed? And I went you know, you sort of you and you can end up with entire APIs with no authentication.

Ash:

Yeah.

Ian:

Yeah. It can be very dangerous. It's not a way to produce production grade code. Yeah. But you can produce production grade code with AI.

Ian:

You just can't vibe code it. Right. That's the thing that kind of winds me up now.

Ash:

Yeah. Yeah.

Ian:

So people say vibe coding is equated to all coding that involves AI. It's like, no. It isn't. The kind of coding where you're constantly arguing with it and making it adhere to a professional design and making it Yeah. Yeah.

Ian:

That is is not vibe coding. There's no vibe going on there. Apart from the slight vibe of irritation in it yet again Well, I guess that doesn't do your form really.

Ash:

Mean, that is a vibe, isn't it? You know what I mean?

Ian:

I don't think it's the vibe that Andre Karpathy is No. Is talking about in his, in his iconic and, nearly monumental Nearly

Ash:

monumental description.

Ian:

Description. Yeah.

Ash:

Yeah. So one other thing that I spotted as well, I often haunt the Thoughtworks tech radar.

Ian:

Does that mean that you lurk around it and then you jump out at people?

Ash:

Yeah. And say, oh, look at that. Look what they've added.

Ian:

Yes. Yes. And you say

Ash:

Oh, that's in the wrong quadrant. That's in the wrong yeah.

Ian:

You're wrong.

Ash:

So one of the things that they weren't wrong about or was at least interesting could possibly turn out to be wrong, they started talking about fuzz testing, which is quite an old technique of just supplying random data to, say, an API endpoint and making sure that you get the responses that you you think you're gonna get and you don't.

Ian:

Is this like, the the tester going into the bar and ordering a pint Yeah. And then ordering 998,473,716.3 pints

Ash:

Yeah.

Ian:

And minus two pints?

Ash:

Yeah. Pretty much. So all of which, probably, apart from the first one, get you annoyed.

Ian:

Bartender. Bartender. Yes. Rightly so. Yeah.

Ian:

Don't don't do that, kids.

Ash:

So but they started talking about fuzz testing as a way to cope with complacency in AI generated coding.

Ian:

Oh, is complacency the thing where you just let it do the

Ash:

Well, you see, this is the thing. Is complacency the vibe? And I like the way complacency with AI generated code is also on the tech radar in the hold status as it don't do this. So, yeah, fair enough.

Ian:

But yeah. You see, this is what this is the this this is where we get I There. This is where I started with a problem because vibe coding is not that. There's no place for complacency in vibe coding because vibe coding is just fun. It's not Yeah.

Ian:

It's not real work.

Ash:

Yeah. But it's already morphed. So here it says, the emergence of vibe coding, where developers let I a a I

Ian:

Did you know they they do some really great

Ash:

I understand that. Presenter and writer. Where developers let AI generate code with minimal review. And, like, basically, on here, on the Thoughtworks tech radar, it basically says that that's what pipe coating is. It doesn't mention fun or weekend projects or anything like that.

Ian:

Well, I link to an article that Simon Willison, who I often quote

Ash:

Oh, yeah.

Ian:

Because I really like his his blog. Anyone who's interested in AI should certainly be reading Simon's blog, if you're a nerd anyway. And he made a blog post entitled semantic diffusion.

Ash:

I prefer vibe coding.

Ian:

I learned about this term today while complaining about how the definition of vibe coding is already being distorted to mean any time an LLM writes code as opposed to the intended meaning of code I wrote with an LLM without even reviewing what it wrote. So semantic diffusion is a term that means the way that other terms because semantic diffusion has not suffered from itself, I don't think. No. Well That would be really interesting. Yeah.

Ash:

Yeah. Definitely.

Ian:

Semantic diffusion refers to the way in which the term vibe coding has turned into code I wrote with an LLM.

Ash:

Yeah. So to be fair to Thoughtworks, they they back it up with, apparently, GitClear's latest code quality research that the emerging trends are four times more code cloning, copy and paste, exceeds moved code for the first time in history. Oh, dear. Oh, dear. Yeah.

Ash:

So, basically, they posit that the AI, coding assistants are already having this influence this vibe. Sorry. Not influence vibe.

Ian:

Oh, we're using vibe in so many different

Ash:

ways now

Ian:

that it just lose all meanings.

Ash:

Yeah. Pretty much. Do you

Ian:

ever have

Ash:

that? Symantec diffusion for vibe.

Ian:

Yeah. Do you ever have that thing happen to you where you say a word a few times, then you can't remember what it means? It kind of it it starts to look weird. And you sort of think this is a very common word that I use all the time but

Ash:

Like what?

Ian:

Why does it look funny? Hang on. I'm typing in the search. The experience of a word's sound starting to be weird after saying it a few times. The experience of a word sounding or appearing strange after repetition is a psychological phenomenon called semantic satiation.

Ash:

Of course it is.

Ian:

Repeating a word causes it to temporarily lose its meaning for the listener. The brain shifts focus to the sound of the word rather than its meaning making it sound like gibberish.

Ash:

Oh, alright. Okay.

Ian:

So So we've got semantic, diffusion, and also satiation. Mhmm.

Ash:

So is vibe coding starting to sound strange and and gibberish like?

Ian:

This whole conversation, though. I'm sure we were going somewhere, and then I think my semantic satiation And then sidetracked may have made me forget what we were what we were doing.

Ash:

So someone I know went to work for a company who are making, do you know this people like to buy what they call motor homes?

Ian:

RVs?

Ash:

RVs. So they're making, like, like a management and monitoring software for RVs. But the chap who set up the company was was nontechnical and just just vibe coded, like, the whole thing. And it works in a way. It's not very extensible, which I suppose speaks to your, like, design comment from earlier on.

Ash:

But it does work, and he has sold it to people. And then he's hired a developer, and they've gone, woah. Look at the innards of this. All inheritance for everything comes from, like, one class, and you could have a catastrophic change in that class. And that's kind of interesting because the according to Y Combinator, one quarter of the startups in its latest batch relies on AI to generate 95% of their code base.

Ash:

So, basically, this whole generation of startups whose code is largely generated by an LLM. But I suppose within that, it depends, like, are they doing it by vibe, or are they doing it as professional developers guiding the LLM to come up with a sensible code base? I suppose it depends, doesn't it? I bet I bet it's vibe.

Ian:

Well, it's probably somewhere in between, isn't it? Yeah. So there's a guy on Twitter. I probably should call it x, really, shouldn't I? And I keep seeing screenshots of this guy's tweet, so I've just been looking for it to try and find the original of it, and it is still there.

Ian:

And it's it says this, March, leo j r '90 '4 underscore.

Ash:

Okay. Snappy.

Ian:

Ever since I started to share how I built my software as a service using cursor, random thing are happening. I I like him already.

Ash:

What a lot of.

Ian:

Yeah. Maxed out usage on API keys, people bypassing the subscription, creating random stuff on the DB. As you know, I'm not technical, so this is taking me longer than usual to figure out. For now, I will stop sharing what I do publicly on x. There are just some weird people out there.

Ash:

There are some weird people out there. It's true.

Ian:

He's rolled all API keys and moved them to environment variables. That's very sensible. That's that's excellent practice. Yeah. He's implemented authentication to API endpoints.

Ian:

He's still there the next morning. And then FML, not sure what that stands for. I should have just kept it quiet. Anyways, thanks everyone who's been trying to help or at least not throw me under the bus. I'll learn from my mistakes.

Ian:

I suppose me reading out his tweets in a sarcastic voice probably isn't helping his

Ash:

Oh, no.

Ian:

That's his psychological well-being.

Ash:

Another form of being thrown under the bus, isn't it?

Ian:

It it is. And then underneath immediately, there's an ad for gearastic kitchenware. So that's the wonder that is x.

Ash:

Twitter. So, yeah, I just thought that was such an interesting introduction of a term, and then it morphing into something else. Via semantic diffusion. Via semantic diffusion. And, also, I wanted to bring it to you, Ian, because I know that you're a fan of AI augmented development.

Ian:

I'd like to thank you for not saying vibe coding.

Ash:

See? See? Yeah. Gwen was like, just call it vibe coding than what Ian does.

Ian:

Yeah. Yeah. Cheers, Gwen.

Ash:

And I thought it would be an interesting one from, like, your point of view Yeah. Because, you know, you've embraced this for a little while now. Not vibe coding. Sorry. Just to be clear.

Ian:

Well, not intentional vibe coding.

Ash:

No. I mean, I guess unintentional vibe coding has probably been around for a long time. Right?

Ian:

I remember the the day before Claude Code came out, we released an episode including, AI pair programming as one of the things. Not my thing I seem to remember. Aside from it being funny that Cloud Code came out the day after we released a bloody episode.

Ash:

That's just our timing.

Ian:

Right? Yes. Yes. Insofar as we interact with time, it screws us. I've entirely forgotten what I'm doing.

Ian:

Sorry.

Ash:

I think that Vibe Coding's been around for a long time as a as a career software tester. So hear me out. Right. So I've worked on projects where, like, suddenly, it's the most important thing in the world to deliver it as fast as humanly possible. And then suddenly, all the error handling around all code just starts to thin away and eventually just disappear, and then things like authentication are forgotten.

Ash:

So

Ian:

Well well, that's it, isn't it? And you start to compromise your crown jewels of of data. Yeah. Because you only need one serious breach to be very publicly embarrassed. Yeah.

Ian:

Because you have to tell people it's the law. Yeah. You can't just cover it up and move on.

Ash:

Yeah. So I think that in its way, Vibe Coding's been around for a little while.

Ian:

Well, something like it. Yes.

Ash:

Yeah. Because you have to be vigilant whether or not it's some you know, another developer has written the code and you're reviewing it or the AI has written the code.

Ian:

Well, no. Hang on. Hang on. Hang on. You can't be dragging that into the definition of vibe coding.

Ian:

The semantic diffusion has gone far enough.

Ash:

I don't think it can ever go

Ian:

far enough. Vibe coding is where you write code with AI without checking it. And the essential features are you don't check it, an AI writes it, and it's you that does it. The subject, the person who's doing the vibe coding, does it.

Ash:

So not just vibes in general then?

Ian:

No. No. No. I feel

Ash:

like you're keen to ensure that the definition of vibe coding is protected I I for any further exploration of the subject.

Ian:

Well, I like the definition of vibe coding. I think is where where where that comes from. I like that it kind of makes it okay to just sit and play about and Yeah. See what happens. And coding, it can be fun.

Ian:

It can be fun, and doing it in in that kind of way is an interesting way to explore and see what the AI is capable of. Yeah. But it's not the way to write production ready code that you can release and that people can trust that their data will be safe. It just isn't, and it's important to have that distinction. What's interesting particularly for me in the various things you've brought together

Ash:

Yeah.

Ian:

Is the Thoughtworks side of it. Despite the fact I now probably will have PTSD about you leaping out on people in the Thoughtworks radar and disagreeing with them. But it's very stressful. But not the leaping out, obviously, then just the disagreement. It's just socially very terrifying.

Ian:

That idea of, actually, it doesn't require a developer to set out to do vibe coding. It just requires that they relax their village their villages. It just requires that they

Ash:

take a village.

Ian:

They it requires that they relax their vigilance Yeah. For a a short while while they're doing this, while they're coding with an AI, and the AI just does something Yeah. That they weren't expecting. And that, I think, is a real risk, and I know it is because it's happened to me.

Ash:

Yeah.

Ian:

And I tend to independently review the code at the end now. Sometimes I even get an AI to help with that. So that, ah, that was the link back. Finally, I can finish my my thoughts. The thing we talked about during the AI pair programming episode was Simon Willison's LLM command.

Ian:

Yeah. How you can basically pipe your git repository's code into the LLM command, omitting, node modules and other folders of libraries because the LLMs are only so big. Yeah. But if you put those things into an LLM, you can ask it to do a security review and it will review for have all my APIs got suitable authentication. Yeah.

Ian:

Because I've done it. I did that. I did that before before I even really did much with AI coding just to test. Some of my code was pen tested, which was quite a scary no one's ever done that to me before and it was alright actually. I only got one medium.

Ian:

There were no criticals.

Ash:

K.

Ian:

And my medium was that my code allowed itself to be displayed in an iframe. Mhmm. So I've told it never be displayed in an iframe. Okay. And and now it it's happy and and and I've solved my my my boat.

Ian:

Interesting. But, you know, I'd before that test, I'd piped that repository into the, AI and said, you know, review the security model of the API calls, and it did. And it it said, yes. They all seem to have authentication apart from the ones that seem deliberately not to. Yeah.

Ian:

And, yeah, it was good. And it but it it made gave me a warm feeling that, yeah, it was gonna be alright, which it was. But I think that kind of taking a second look after you've done whatever it is getting a person Yeah. Or some, dare I say, even testing, but

Ash:

Who does that?

Ian:

Crazy people. But even

Ash:

No one wants to know how anything really works.

Ian:

Even if it's just an AI going through and looking for pattern breaches in your APIs, that's just a second look, isn't it?

Ash:

Yeah. Which is the, obviously, the antithesis of vibe coding. Yes. Because you're not supposed to look or even be interested in looking.

Ian:

No. You're just playing you're just doing a little dance with the machine. Yeah. It's, an activity done for pleasure, not for work.

Ash:

Exactly.

Ian:

That's it, isn't it? That's the core difference. Yeah. If someone's paying you, don't vibe code.

Ash:

Don't have any fun.

Ian:

Have fun. Just don't vibe code. No. A semantic diffusion on the word fun is taking place.

Ash:

So that's, that's vibe coding. Oh. Well, or at least one discussion about vibe coding. I'm sure there'll be many more.

Ian:

What a great thing. I enjoyed that thing immensely much more than I'm probably gonna enjoy my thing.

Ash:

Well, that sounds like the time for an interlude, doesn't it?

Ian:

It does. I wonder how how long it will continue to be amusing to me that I put the music on when it's time for an interlude.

Ash:

Forever.

Ian:

Nice. It's nice, isn't it?

Ash:

Start the relaxation. Oh, no. Don't. Because you're failing your capitalist masters.

Ian:

Oh, sorry. Yep.

Ash:

Turn off. You got you got really nervous when I mentioned your capitalist master.

Ian:

I did. Yes.

Ash:

And you're like, oh, I went to pieces and pressed all the wrong buttons.

Ian:

Ash was looking at me as I stabbed violently at the buttons on the on the road caster trying to make it stop. Well, actually, what I was doing was turning on and off the stupid deep voice effect.

Ash:

And the squeaky voice effect, which I know to be your favorite. It is good.

Ian:

She's becoming incapacitated with laughter due to the squeaky voice. I kind of wish I could just do it to him, but unfortunately, my powers are limited to to only doing it to myself. Maybe if I switch the microphone channels without without saying, and then I can just sit there and make you all squeaky.

Ash:

So I've got a couple of, interlude things.

Ian:

I'm glad you have because I've failed dismally to have any interlude.

Ash:

This is this is my interlude now given I've brought the things, two things. Interlude things.

Ian:

Things with a small t.

Ash:

Things with a small t.

Ian:

Honestly, this the semantic diffusion associated with the word thing in this podcast is is is dramatic.

Ash:

Is semantic diffusion gonna become a theme of the podcast?

Ian:

Only if I have anything to do with it.

Ash:

So, yes, it will.

Ian:

My joy at making Ash laugh when he's got a mouthful of water is tempered by the way in which he's pointing at a lot of technology that I own.

Ash:

So interlude things. Are you ready? Just seems

Ian:

a bit structured. Well It seems like we should be able to sort of flow into the interlude things.

Ash:

Wow. But we're kind of But I've put them in a in a bulleted list, so it's it is structured.

Ian:

It is structured.

Ash:

Yes. One of them is a question for you, Ian.

Ian:

What? It's your favorite color.

Ash:

So say if you'd received some designs. What?

Ian:

Like like, for example, Figma designs.

Ash:

In a popular tool

Ian:

In a butcher's tool.

Ash:

Style of the time. And say if you were building part of those designs, and then there was a button on them which took you to a thing that didn't exist yet, as in it's we think it's gonna exist in the future. So would you add the button that takes you to the thing that doesn't exist?

Ian:

Is is that the equivalent to, in the physical world? As I'm driving along, a road sign says to my destination pointing over a bridge, and then the bridge is only half built, and I fall in the water in my car.

Ash:

No.

Ian:

Well, it's it's the same. It's

Ash:

Yeah. Yeah.

Ian:

Slightly different in in a it's just different scale of things.

Ash:

Yeah.

Ian:

I don't know. A button's nowhere. One of my favorite Simpsons things was the escalator to no clue. Yeah. Yeah.

Ian:

Yeah. True.

Ash:

So I guess it does have

Ian:

It was just a throwaway joke at the end of the episode about the monorail. Yeah. Some other things, basically, the writers have thought of.

Ash:

Yeah. I've just noticed that there's there's there's generally, like, two types of developers. The ones

Ian:

There are 10 types of developers. Yeah. The ones that understand binary and the ones that don't.

Ash:

Yeah. Pretty. Well, to me, there's the ones that retrieve a design, build it, but realize that there's a button that doesn't go anywhere yet, so they just don't add it.

Ian:

And then there's

Ash:

the ones who just follow the design and have the button to nowhere.

Ian:

You see, I I quite I would like to add the button to nowhere Yeah. Because I would think it's funny. I wouldn't think it was the right thing to do. Yeah. But I would think it was funny, which is also much more important.

Ash:

No. That's fair enough. That has answered my question, to be frank.

Ian:

In fact, I would probably go so far as to build something at the other end of the button to give you a more amusing error page when you got there. Yeah.

Ash:

That's true, actually. I I would accept that more than just

Ian:

Here's the phone number of the designer. Please ring them and ask what should be here.

Ash:

Yeah.

Ian:

I'm at this page, and it says I should ring you up to ask what should be

Ash:

on it.

Ian:

Yeah. Yeah.

Ash:

Yeah. So I'd accept that more than just, you know, you press it. It gets into a confused set of redirects and then just goes to the error page. So, yeah, I was just wondering, really.

Ian:

Yeah. Yeah. That's it's almost like, there's a problem with the design tool in that it allows you to, I'm

Ash:

not sure what you mean.

Ian:

To create buttons to nowhere.

Ash:

Buttons to nowhere.

Ian:

We press the button to nowhere. More Australian bands.

Ash:

And then the other thing, interlude thing.

Ian:

Wow. We're really motoring through these interlude things.

Ash:

Yeah. Yeah. I'm I'm intrigued. So I saw that Sam Altman, rather than saying that

Ian:

We're quoting billionaires now.

Ash:

Yeah. Rather than saying that LLMs are gonna replace developers, now he's changed tack and said they're gonna make them 10 x developer. So and all along, he said, you know, we just wanted to make them better, not replace them.

Ian:

Well, I think there's been a lot of that kind of noise around. I remember at IBM years ago when Watson came along, and we were just digesting the idea that someone had made an app that could beat humans at jeopardy. I say an app, it was like a massive machine room full of servers.

Ash:

I was gonna say the very glib description there.

Ian:

Yeah. It wasn't running on anyone's phone. Put it that way.

Ash:

It's actually a 10 story building. Yes.

Ian:

And and that was IBM was sort of saying a long time, you know, back then, AI will help humans not replace them. Yeah. But I also think that companies are making assumptions that they can replace humans with AIs. I'm fairly certain that's happening. I don't know about 10 x in developers.

Ian:

I find this idea of a 10 x developer quite hilarious, to be honest. Yeah. It's just like but even so, yes, it will empower developers

Ash:

Yeah.

Ian:

Or at least vigilant developers. All the ones that aren't vigilant will be fired when they expose the company data through a slightly, misplaced reliance.

Ash:

Yeah. But it won't be their fault. It'll be the AI's.

Ian:

No. It'll be their fault. That's the thing.

Ash:

You're just not with the vibe in.

Ian:

Yes. I am. And that isn't the vibe.

Ash:

Oh, dear. Tell me what the vibe is?

Ian:

I won't tell you what the I'm starting to lose semantic coherence.

Ash:

Oh, I like it.

Ian:

We're talking about your thing again. This this second thing about 10 x ing developers was just a secret. It's like the button to nowhere has come out of your thing and landed us in the interlude still talking about vibe coding.

Ash:

I can jab at you again about vibe coding.

Ian:

You're not jabbing at me. I feel no jabs. I'm completely

Ash:

Jab free.

Ian:

Jab free. No. I'm not. I've been vaccinated against all all the good news. Diseases.

Ian:

But I guess what I what I mean is that yeah, I'm just gonna stick with vibe coding means what Andrei Capafi said it means and nothing else. Yeah. Yeah. I think AI can really help developers.

Ash:

Yeah. Some of the less charitable commentators on Sam Altman's words are based

Ian:

Surely. There are no uncharitable commentators.

Ash:

Yeah. Indeed. But they were basically saying that, the amount of, compute power needed in order to do that actual job, might be something that's a bit too, much and expensive for, the world to bear. So

Ian:

Well, it is expensive. Yeah. And anthropic, particularly, the Claude APIs are really quite expensive.

Ash:

Mhmm.

Ian:

So Claude code, I could quite if I spent a whole day really focused on using it to do something, plus all the do agains it has to do because it gets it wrong, I could easily spend $50, 60, 70, 80 dollars. That's a lot cheaper than another human to help me. Mhmm. But, you know, it's not it's it's a noticeable sums of money.

Ash:

Yeah. Yeah.

Ian:

I don't know. But I just think all this 10 x full stack developer Yeah. Ten years experience in something invented yesterday. It's just it's just recruitment industry called mad, isn't it?

Ash:

Yeah. I guess it's just the kind of the sign of a, you know, another set of semantically diffuse terms. Yes.

Ian:

Yes. We all say that, but none of us know what it means. Yep. Or we all know what it means, but everyone's version is different.

Ash:

So that that's what I had. No. Terms of interlude things.

Ian:

I can't believe oh, I've got an interlude thing. I've just thought of it. Because I was driven to reflect on your four item WIP limit that you have in your life.

Ash:

Oh, yeah.

Ian:

Because the other day, I was trying to figure out why I was so busy. If you look behind you sorry. Not you. I mean, Ash, in the other direction, there's a list of all the things I'm doing at the moment, and it's not four. When I did it, it was seven.

Ian:

There's lots of things.

Ash:

But that's the first stage. Well, it is. I'm being genuine.

Ian:

Yes.

Ash:

Yeah. It's just showing how much work in progress you actually have.

Ian:

Have a problem.

Ash:

Yeah. Exactly.

Ian:

They're going to be a a number of steps. Oh, I know I have to

Ash:

do step program.

Ian:

Now I've admitted I have a problem.

Ash:

Yep. Then you can begin to adopt Stop

Ian:

doing all the things.

Ash:

Then you can begin to adopt the patented Ash Winter system.

Ian:

Will you sell me some stationery? Because I need stationery before I go

Ash:

to the is the is step seven. Step

Ian:

Is that to give yourself some time to make some?

Ash:

Oh, yeah. Exactly.

Ian:

Put it on sale for an abnormally large amount of money. Yeah. So it's quite interesting. So I have allowed my work in progress to build up to quite a high level. And dismally, only two of the seven things are being paid for.

Ash:

That does sound like you, Ian, to be fair.

Ian:

It does a bit, doesn't

Ash:

it? But,

Ian:

you know It's

Ash:

the first step.

Ian:

It's the first step.

Ash:

To a better life.

Ian:

To a different life. Different. Oh, dear. Yes. So I I do often think about that.

Ian:

I I think I admire your your discipline of doing it, which isn't to say I think you do it perfectly because presumably, there are all manner of very enticing opportunities to fall off the the wagon Yeah. Of of adding new things to your

Ash:

Yeah. Things come along, and you have to be very, very diligent.

Ian:

Yes. You can just be riding along on your nice day, and you look behind you, and there's, like, seven things on your to do list.

Ash:

Yeah. Yeah. Then you have to try and ignore them as best you can.

Ian:

Hope they go away.

Ash:

A lot of things do go away.

Ian:

That is true.

Ash:

So but not that's not to say you should ignore everything. Things for the amount of things that actually go away, there's any probably, like, a similar amount of things that don't go away and become other things.

Ian:

Yeah. Well, that's the thing. It's tell telling the difference between the things that will go away and the things that will multiply if you don't have a

Ash:

Yeah. Yeah. There's something about judgment in there as well, which is a harder

Ian:

Is that judgment as in showing good judgment or judgment as in hard. I find you guilty of having too many things on your wit list.

Ash:

Well

Ian:

You you must feel bad about yourself.

Ash:

Now you put it like that. Kind of a bit of both.

Ian:

Yeah. Yeah. Yeah.

Ash:

Yeah. Yeah. Because, you know, everyone's got too much work in progress. Right?

Ian:

I feel that's a reasonable hypothesis. Yeah. There's nothing worse than than than having not enough work in progress. I don't like that at all.

Ash:

Just need to find the right amount. So far, it's a nice amount. Yeah? Yeah. Keeps me from being too crazy.

Ian:

Yes. Yes. Very successful strategy if I if I may say so.

Ash:

Absolutely. Agreed.

Ian:

Ash winter, ladies and gentlemen, not too crazy.

Ash:

Not too crazy.

Ian:

Oh, dear. We can't take the interlude stress anymore.

Ash:

No. Well, maybe we should do another thing.

Ian:

I need to do another thing. Yeah. This thing is a sort of antidote to last time's thing, which was so weighty and unexpectedly, traumatic that I feel the need for something vapid

Ash:

Oh, excellent.

Ian:

And relatively safe. Is anything truly safe? No. No. I thought not.

Ian:

My thing Is? Blue sky.

Ash:

Back on the, social media train. Well When I first saw this in here, I was like, is this just gonna be Ian asking me to go on blue sky? Is that gonna be your thing? Because the answer is still no.

Ian:

I I feel it could become part of it maybe. I actually think you should go on Mastodon not blue sky because I think you'd be more likely to interact with people that you'd find interesting to interact with on there. So Mastodon, you'll generally find that you've got smaller group of people, but they're more interesting. Mhmm. So one thing about Blue Sky is that a few episodes ago, I accused it of being owned by billionaires.

Ash:

Right.

Ian:

Because I thought Jack Dorsey owned it because he was he started it. But actually it and it is owned by VCs but not billionaires so Jack Dorsey is not involved with it anymore.

Ash:

Alright. Okay.

Ian:

And the other thing about it is that it's had a massive uptick recently.

Ash:

Alright. Okay.

Ian:

Because people who were on another social network, the kind of social network where your account name gets truncated because there aren't enough characters for it have left that social network. And I think one might speculate that it's something to do with the owner and his Oh, yeah. Salutes and political leanings.

Ash:

Yeah.

Ian:

And a lot of people have landed on Blue Sky. And it's interesting to me because they didn't land on Mastodon. They landed on Blue Sky. Yeah. And I think the reason is that when you land on Mastodon, you have to choose a server.

Ian:

I think the friction of that is too much for people.

Ash:

Well, yes. It doesn't have the simplicity of just signing up to the service.

Ian:

No. But yeah. So a lot of people have landed on Blue Sky. So I think people like Ian Dunt Mhmm. Who's a political writer who swears a loss on Twitter, he's now on there instead.

Ian:

And that's good because before, I had left him behind because he was on Twitter. Yeah. But now I can read his stuff again. And so there's a lot of a lot of interesting people on there now. And the other thing is that it's built on this kind of open protocol Yeah.

Ian:

Called the AT protocol. And if you're gonna ask me what AT stands for, I don't.

Ash:

Okay.

Ian:

There's a lot of opportunity for people to moderate their own feeds and things. It's got a lot of, sort of plug in stuff for that. The CEO is a woman called Jay Graber. Yeah. And she says the blue sky tea does team tea.

Ian:

The blue sky tea. That sounds nice, doesn't it? The team does moderate content, but it offers via the open protocol. Users can have the ability to create their own moderation systems

Ash:

Okay.

Ian:

If they want to. And you can have modules that screen for particular things or or politics or whatever and use it to create your own kind of filtered view. So I think there's a potentially a lot more of that that people will will do. That's the kind of thing that I think people will take advantage of.

Ash:

Yeah. You could do that on on Twitter to some extent as well, couldn't you?

Ian:

Yeah. I mean, people are people use Twitter lists and things, didn't they? Yeah. Yeah. Another interesting blue sky thing is that you can make a kind of a thing called a a welcome pack or something like that

Ash:

Alright. Okay.

Ian:

Where you can just recommend a group of users to follow that new users can just click and follow them all. So if you just found someone if you were to sign up to Blue Sky and found a tester Yeah. Who had done that, made a list of that world, you would be very ease easily able to then jump back into those conversations that you're currently missing out on. It must feel terrible to be missing out on on on conversations.

Ash:

You know what? It doesn't.

Ian:

It doesn't. Yeah. You're right.

Ash:

I'm not I'm

Ian:

You're right.

Ash:

I'm not against these tools. I think I've just left it behind a little bit. I'm just not I find myself less interested in social media than I ever have been. Yeah. Yeah.

Ash:

It's it's not a I'm not fundamentally against the idea. You know?

Ian:

Well, I enjoy putting these options before you, but I'm not tremendously optimistic that you're able to take them up. But I I feel as though I feel as though social media is missing out on you rather than necessarily the other way around.

Ash:

Well, yeah. Maybe. But I don't know. I think I maybe I was oversaturated previously. And then now

Ian:

You're just in detox rehab. Yeah. You're on social media view.

Ash:

Yeah. Yeah. I think so. I think so. So Jack Dorsey Yes.

Ash:

In the notes, it says he left and deleted his account.

Ian:

I'm taking my ball off. I'm going

Ash:

Yeah. I'm interested in the petulance of, of this. So because he because then he says he complained about Blue Sky's moderation approach.

Ian:

I mean, that it had one.

Ash:

Yeah. Yeah. And basically then said, this is literally repeating all the mistakes Twitter made as a company.

Ian:

Free speech, man.

Ash:

Yeah. Yeah. That's really it's a common thing with the billionaire class, isn't it? This kind of odd odd views on free speech and not moderating things and just letting, you know, people be as toxic as possible. And and then any form of moderation seems to be like a, you know, an imposition.

Ian:

Well, the thing is that free speech is a good thing.

Ash:

Yeah.

Ian:

But you sort of think, okay. So the first thing that people do when they're trying to form a dictatorship is to start shutting down Yeah. Free speech. So I think free speech is a vital freedom, but then there you get all these questions about where the lines are to be drawn. And when someone's freedom to do something is in conflict with someone else's freedom to be not have done to them.

Ash:

So they seem to like, here, Jack Dorsey, it seems to be a very sort of childish view of what free speech is.

Ian:

Well, I think that they've got there's this whole idea of being a free speech absolutist. Yeah. And it's basically saying it's never the right thing to do Yeah. To limit free speech. And, actually, the cure for bad speech is more speech.

Ian:

If people are wrong, then you should argue with them.

Ash:

Just endlessly.

Ian:

Not not ban them. Yeah. And I, you know, I don't think that's a completely meritless point of view. The thing is you can't you you can't do it. It's like an ideal.

Ash:

Yeah. Yeah. Because it doesn't work in practice.

Ian:

In practice, you have to draw a line, then you get into the argument of where should the line be Yeah. All the way to one side of, I think free speech is very important unless it's disagreeing with my opinion.

Ash:

Yeah. And I feel like that's what this generally is.

Ian:

Yes. And I think that's Elon's Twitter version of free speech.

Ash:

Yeah. It's like freedom for me, but not for you.

Ian:

Yeah. Through to the other side where you just say you can say anything, and then people are using their speech to hurt other people Yeah. Or to to whip up

Ash:

mobs or that kind of thing. To target them to But it box them to whatever it is.

Ian:

It is very difficult to find where to draw that line

Ash:

Yeah.

Ian:

Because you can move the line around, and different people are saying, no. You're wrong. No. You're wrong. Yeah.

Ian:

You know, I don't think ethics professors are necessarily a million miles ahead of anybody else in

Ash:

this kind of

Ian:

in this you know, people who study ethics. I'm sure they have a view that's more of more high quality than mine. Yeah. But I just kinda run out of steam on free speech as it's complicated, but I feel like we should probably in the direction of freedom. Yeah.

Ian:

It can't be all the way.

Ash:

Yeah. It's complicated, but you have to try to make sure people don't get hurt.

Ian:

Yes. You But what does your dirt mean?

Ash:

Yeah. Exactly.

Ian:

Are we talking about snowflakes and their poor feelings?

Ash:

Yeah. Or the other

Ian:

I have some sympathy with snowflakes and their poor poor feelings.

Ash:

Well, we've all got feelings. Yes. So But

Ian:

are we being hurt if our feelings are hurt? And should we stop people from saying anything that hurts anyone's feelings? Because you can't really do that because anyone can say their feelings are hurt just to stop anyone else from saying anything. So

Ash:

Again, I'm not saying it's easy.

Ian:

Complicated. I'm saying

Ash:

but you can't just abdicate. If you're gonna set up a platform such as this, you cannot just abdicate all responsibility and just engage in absolutism.

Ian:

Well, not anymore because you've left and deleted your account Well

Ash:

in the case of Jack. But that is what a child does when faced with a complicated problem. So I think blue sky in the long term might benefit from not having mister Darcy around.

Ian:

Well, they are doing some interesting things about the moderation problem. It's an interesting area. Yeah. But, yeah, I'm kind of behind it at the moment even though it's VC backed, which

Ash:

Well, unfortunately, in the in the world

Ian:

suggests in shitification will arrive eventually.

Ash:

Well, eventually, you'll be paying your subscription. But

Ian:

Yeah. That's that's fine. It won't be even noticed amongst all the other subscriptions, though, like

Ash:

In the forest of subscriptions. Yes. The If another one falls Yes. Does anyone hear it?

Ian:

Does anyone hear it? Yeah. But at the moment, they seem to be doing some interesting things. Yeah. And, actually, in compared with being on Twitter now, it's quite a pleasant experience.

Ash:

Yeah. So what about the modular moderation? That's kind of interesting as well, isn't it? So you could build your own, like, personalized moderation.

Ian:

Well, they say it's, I I did try and look into it a bit. I couldn't figure out how to do it.

Ash:

Oh, right. Okay.

Ian:

It's one of those things where I think if I had read the documents for another hour, I probably would have got a handle of it. But, you know, I have these seven things on my to do list, so I didn't do it. Yeah. But it sounds like it's sort of pluggable modular kind of thing where you can write little bits of code Okay. That can look at your Blue Sky feed and report back to Blue Sky, no.

Ian:

Don't show them that.

Ash:

Right. Okay.

Ian:

And Jay Graber says, you can think of these moderation pieces as Lego blocks. So I already like her because she didn't say Legos because Americans say Legos, and you just can't say that.

Ash:

Yeah.

Ian:

But you can say Lego blocks.

Ash:

Lego blocks. Yeah.

Ian:

And a user can assemble them, I e modules that screen for profanity or graphic language. You mean people talking about Figma or the word politics

Ash:

I don't screen that out.

Ian:

To to create their own

Ash:

Give me all of it.

Ian:

Bespoke experience. Maybe your bespoke experience can be you only get Figma. You can just follow the Figma hashtag. And then

Ash:

Just keep adding buttons to nowhere.

Ian:

And then and then argue with people. Yeah. Yeah. That's but you could

Ash:

That that could be my

Ian:

Oh, yeah.

Ash:

My reason for being on

Ian:

the on the

Ash:

blue sky. Soon as someone mentions Figma.

Ian:

Well, I'm just thinking I think

Ash:

I would be getting filtered out by the moderation tools.

Ian:

Yes. You probably would. Yeah. I feel as though if you went on blue sky, then we'd be on episode 60 or something. I say, you haven't mentioned Figma for ages.

Ian:

And you'd be like, yes. I have. It's just not on here.

Ash:

Maybe it doesn't need another outlet.

Ian:

Yeah. Yeah. Yes. Add another thing. Hashtag add another thing.

Ian:

So, yes, I as I said, that's a bit more of a vapid Sure. Lightweight thing. But I do think it's interesting, and I feel that the open protocol gives it some

Ash:

Yeah.

Ian:

Some hope. I feel that I think if you wanted to run your own, you probably could.

Ash:

Yeah. Because with because I remember when I had a could

Ian:

vibe code a a server.

Ash:

Oh, yeah. Totally. Yeah. So when I had a Twitter account, I often filtered out words like Trump and other words like that. Just because, you know, you get sick of seeing it, I realized that that's a privileged position to be in.

Ash:

But it's this doesn't enhance my life, so I want to remove it. But, again, that was that wasn't an open protocol, was it? That was what Twitter had decided

Ian:

you would be allowed to do.

Ash:

Yes. Exactly. So, you were still exposed to and then often, you know, you would still get it in your feed, but just in a kind of a roundabout way. You know what I mean? Because there'd obviously be gaps in the implementation where articles that are about the particular person

Ian:

Without actually mentioning it.

Ash:

Yeah. Exactly. So there was no way to do that, like, completely. So maybe having, like, a more open version would be a bit more personalized to your own needs.

Ian:

Well, that's what you need.

Ash:

Yeah. But in terms of intensification, it's hard to see how a tool how a service like Blue Sky won't necessarily won't eventually end up in that situation.

Ian:

I like to be optimistic about these things.

Ash:

I don't know. It's difficult, though, isn't it? I like to

Ian:

be optimistic. Didn't say I was.

Ash:

Yeah. Yeah.

Ian:

Last time, we talked about testing, and we tried to get chat GPT to find out if it knew the table of contents of your book. And and I was trying to find out the because remember, it made up a whole book.

Ash:

But it was like, well, I've already done that. So

Ian:

And there was a particular kind of testing that I'm trying to find the name of, but I can't remember it, that you said came from James Buck. And I was gonna mention it and say, is that one of the words you would filter out? But, basically, I just can't find it.

Ash:

I'm trying desperately not to say vibe testing.

Ian:

Well, I have a feeling that that is not something that you're gonna manage to sustain.

Ash:

It's already being talked about in online, though.

Ian:

You heard it here first. See, that would be much more effective if people have realized I was kind of whispering into my microphone secretively. But it just sounded like I was talking in a funny voice.

Ash:

Yeah. Are you doing both?

Ian:

I always do both.

Ash:

And then is it different enough from Twitter to have a chance of being okay? Well That's an interesting question too.

Ian:

I think it needs to be different from Twitter

Ash:

now,

Ian:

not different from Twitter Yeah. Yeah. It's like five or ten years ago.

Ash:

Time distortion field there, isn't there?

Ian:

We're we're constantly dodging them around here.

Ash:

Yeah. Because I there was a time now when Twitter was a really nice place to be.

Ian:

It was. Yeah. And that

Ash:

And quite a positive influence.

Ian:

And that's kind of why I'm doing this kind of mock campaign to persuade you because I feel like you might like it. Yeah. But because some of the particularly mastered on I know I keep going on about it, but it it particularly, it has that

Ash:

Yeah. Yeah.

Ian:

That vibe where most of the people on it are decent people who you would actually want to talk to. And it is it does it is reminiscent of that. Yeah. But, yeah, I mean, obviously, Twitter today is just this sort of toxic soup.

Ash:

Another toxic soup.

Ian:

Want to swim around in it. No. But Blue Sky is is quite nice.

Ash:

Mhmm.

Ian:

I like it. And I'll keep liking it as long as it keeps not being in shitified, I suppose.

Ash:

Yeah. Yeah. And hopefully, it can that can go on for a little while.

Ian:

Yeah. And there's always Mastodon, which is not owned by anybody and therefore is least inshittifiable. I mean, there's definitely things that can go wrong with it.

Ash:

Yeah.

Ian:

But it's one of the things that goes wrong isn't going to be VCs monetizing it.

Ash:

Well, there's a will. There's a way.

Ian:

There's no one called Will involved in it to my knowledge.

Ash:

To my knowledge.

Ian:

Sorry, Will. Whoever you are. I didn't mean to imply you weren't involved with it when you are if you are. Wow. So that was my lightweight thing.

Ash:

What free speech was your lightweight thing?

Ian:

Yeah. That's my best lightweight thing, though. That's the thing I come up with when I want to carry a feather around for a bit.

Ash:

Free speech is a light word for it.

Ian:

Was not my thing. My thing was blue sky.

Ash:

Blue sky.

Ian:

We may or may not have messed

Ash:

up a container of speech. Fair. The idea of free speech.

Ian:

Yes. Yep. Cool. Well, we're doing quite well compared to normal. It's only been an hour and twenty five minutes.

Ash:

Of a treat.

Ian:

We might be coming in under ninety minutes, which will be,

Ash:

Gradually bringing it back.

Ian:

Yeah. That whole hundred and whatever it was minutes last time, that was that was very painful.

Ash:

So how do you get in touch, Ian?

Ian:

You signal.

Ash:

With? You. How does the listeners get in touch?

Ian:

Oh.

Ash:

Yeah. Them ones.

Ian:

Why didn't you say that? Yeah. I should've said that. Honestly, these imprecise you're just you're just asking vibe questions. Vibe questions.

Ash:

But that means I don't

Ian:

care about them. Fuzz questions. Well, you can reply on Blue Sky to this episode and our Blue Sky account, which actually is not a most of the accounts are in a domain of bsky.app. Mhmm. But if you want to, you can confirm ownership of a domain that you own, and then your Blue Sky account can be your domain.

Ash:

Oh.

Ian:

So we are whatalotofthings.com on Blue Sky. And if you reply to posts by whatalotofthings.com, they appear on the episode page.

Ash:

Nice.

Ian:

And if you do that, then Ash will have no choice but to sign up to Blue Sky so that he can talk to you about them. So say something infuriating to Ash

Ash:

About Figma and how much you love it.

Ian:

Yeah. How Figma is perfect.

Ash:

Revolutionized how you work.

Ian:

Has never been misused by anybody.

Ash:

And how you add buttons to nowhere.

Ian:

How buttons to nowhere are a feature, not a bug. Then Ash will finally get on blue sky to come and argue with you.

Ash:

Maybe the page should have, like, five five buttons, four buttons that don't go anywhere, and only one submits the form.

Ian:

This sounds like a nineteen seventies quiz show. Behind door number one, it's a duck. Yeah. Behind door number two, the Ferrari, but you've missed out on that because you've got the duck.

Ash:

I like duck.

Ian:

It took forever for us to shuffle them around behind before we open the doors. There. But, yes, I feel that we just need a juicy conversation underneath our episode on blue sky to to trigger a whole new ash social media revolution.

Ash:

Could it could happen.

Ian:

Could could could happen. Not completely impossible. The chance is not great. Not not great. If if you would feel cheated by the lack of response from Ash or Blue Sky, then probably don't do it.

Ash:

We also have a LinkedIn group.

Ian:

We have a LinkedIn group.

Ash:

Which I am actually on. You

Ian:

we're both on it.

Ash:

Yep. So I can be contacted.

Ian:

Ash Ash posted a hot take on the most recent episode, which was fantastic. Ever we we love to read the Ash winter hot take on the on the episodes of content. Hot takes. It's very good. It's a it's a fine fine hot take.

Ian:

Obviously, a nice refreshing counterpoint to the show notes.

Ash:

I enjoy contradicting the show notes

Ian:

as well.

Ash:

Yes. I do that on purpose.

Ian:

Yes. I I thought so.

Ash:

Yeah. And, also, we have an email address.

Ian:

And that is ianandash@whatalotofthings.com.

Ash:

That's the

Ian:

one. So if none of those three ways of contacting us work, then we have failed you, and we're sorry. But they do work.

Ash:

They do.

Ian:

So you so use them. Ash is ceremonially pressing a button on his laptop, which I think is supposed to symbolically induce me to press the button on my recorder.

Ash:

I'd just try and do it.

Ian:

Oh, no. It's gone over an hour and thirty. That's why

Ash:

I wanted you to do it.

Creators and Guests

Ash Winter
Host
Ash Winter
Tester and international speaker, loves to talk about testability. Along with a number of other community minded souls, one of the co-organisers of the Leeds Testing Atelier. Also co-author of the Team Guide to Software Testability.
Ian Smith
Host
Ian Smith
Happiest when making stuff or making people laugh. Tech, and Design Thinking. Works as a fractional CTO, Innovation leader and occasionally an AI or web developer through my company, craftscale. I'm a FRSA.
Vibe Coding and Bluesky
Broadcast by