Sue Smith

🎏 Writes code to help folk learn to write code

🪩 Smartarse swimmer
💅🏻 Still petty after cancer

Website: https://www.suesmith.dev
Bluesky: https://bsky.app/profile/suesmith.lol
Blog: https://www.sue.codes
Pronouns: She/Her

Public Key

npub1csg6adp29sd5zjnn7cn52u6mwvdnrggnsrdnaz4jtfeca2v3cs6q024dyv

NIP-05 Address

sue_at_glasgow.social@momostr.pink

Profile Code

nprofile1qqsvgydwks4zcx6pffelvf69wddhxxe35yfcpke732e95uuw4xgugdqprpmhxue69uhhyetvv9ujumt0d4hhxarj9ecxjmntza6gyg

Publishing to

relay.momostr.pink

Author Public Key

npub1csg6adp29sd5zjnn7cn52u6mwvdnrggnsrdnaz4jtfeca2v3cs6q024dyv

Show more details

Last Notes

2026-04-06 09:56:06 UTC

Absolutely wild how many men make patronising comments on linkedin posts and have the "open to work" flag on, to not even have the sense to think prospective employers might be women and will scan your activity on there is something else
tldr this is why I don't block them

2026-04-04 10:23:03 UTC

- reply

I wonder at what point the affected skills and understanding become business critical in different settings, because in practice that's when things might change, I joked about this on linkedin but I could envisage a situation similar to what happened with observability and monitoring tools as everything moved to the cloud but for codebase comprehension, when does it become urgent that your teams know how the fuck your software works

2026-04-04 10:18:27 UTC

- reply

The real "human in the loop" would be for the assignment of accountability lol, I find it absolutely bananas how many traditionally risk averse organisations have embraced this stuff without meaningful guardrails, but I do believe the legal shit, compliance, risk management etc will be the thing that enforces some regulation of these practices rather than the quality or reliability of the output itself

2026-04-04 10:14:16 UTC

- reply

Thinking about this again and I think when these tools encounter ambiguity, from the perspective of their logic when there's no way to verify correctness from the training data, instead of coming back to the human and asking them to make a judgment they keep cycling or hallucinate an "answer"
The traditional way of resolving this would be for the human to make a choice they take responsibility for, but since these tools are partially accountability sinks they're not designed for that

2026-04-03 11:11:41 UTC

- reply

The fun part is that I gave claude a few shots at building it, it's a low stakes project, I don't really care about the implementation and just need something that works, but it failed miserably because it's a relatively novel app using niche libraries that aren't well documented and will be almost non existent in training data, but it's a super simple system I could teach a beginner coder to build

2026-04-03 11:09:30 UTC

People keep talking about llms killing saas but we were quoted a ludicrous amount for a previously free platform during the week and I'm rustling up an alternative that's completely free aside from a few days of my own labour using the magical automation that is ✨bash scripting✨

2026-04-01 10:16:44 UTC

I see tech men have rebranded part-time working as "fractional" to distinguish it from the kind of thing women do

2026-03-30 22:27:58 UTC

- reply

yeah this is the more challenging part in reality eh

2026-03-30 19:58:33 UTC

- reply

Paging the replying men of mastodon

2026-03-30 19:58:11 UTC

People who encounter perspectives they mostly agree with, but focus on the part they disagree with or don't like, should know that's the literal opposite to how successful social movements are built

2026-03-30 11:14:57 UTC

"I warn my students:
- Yes, AI can generate the code for this assignment. Don’t let it. You have to write the code.
I explain that, if they don’t write the code, they will not be able to effectively read the code. The ability to read code is certainly going to be valuable, maybe more valuable, in an AI-based coding future."
This is 💯 the advice I'm giving young folk right now.
https://htmx.org/essays/yes-and/

2026-03-24 16:37:56 UTC

- reply

https://files.glasgow.social/media_attachments/files/116/285/127/340/214/620/original/c525cb08c91d9abe.jpg

2026-03-24 16:36:53 UTC

- reply

A couple of tech writing memes that didn't make my talk
https://files.glasgow.social/media_attachments/files/116/285/122/137/308/744/original/f78cc37ef6ed6920.jpg

2026-03-23 12:46:16 UTC

I've written up a rough overview of my #monkigras talk on developer learning and AI assisted coding from last week here, will likely be doing it + similar talks again later in the year
https://www.sue.codes/blog/fingerscode/

2026-03-12 17:01:33 UTC

Although I believe companies will have to dial down the claims that ai replaces their workers, I think it's important not to expect that to happen on the basis of productivity.
I say this because we already have a track record of productivity and other supposed indicators of business success being sacrificed in favour of control.
The cost of employee attrition, the endless studies showing how remote work and shorter working hours make people more effective etc.

2026-01-04 12:24:20 UTC

- reply

So aye, the usual propaganda
https://files.glasgow.social/media_attachments/files/115/836/802/848/309/002/original/aad81a27128179a3.jpg

2026-01-04 12:21:39 UTC

On the viral tweet about claude generating in an hour what a team at google had apparently built in a year, the OP provided more context which.. of course it was a misrepresentation of what had actually happened
https://files.glasgow.social/media_attachments/files/115/836/790/425/504/624/original/b370fdddadaf372b.jpg

2025-12-18 17:32:43 UTC

- reply

I remember the books but I don't think we had a tv version!

2025-12-18 17:14:36 UTC

- reply

🤣 I had not seen this

2025-12-18 17:10:11 UTC

You thought you'd seen innovation, well these slippers are also dusters 🤯
https://files.glasgow.social/media_attachments/files/115/741/671/422/970/839/original/017acddd2aadf2ca.jpg

2025-12-16 00:55:49 UTC

I had a cat who hated the smell of an air freshener we had, but would deliberately go up and smell it every time was near it, just to verify how appalling it was and be annoyed by it all over again. Well, that's how a lot of people appear to use social media.

2025-12-11 17:18:56 UTC

- reply

Above all else you need to learn how not to center yourself or your frustrations at things not being the way you think they should be, or people making decisions you think are mistaken. This is why I love Sidney Dekker's Field Guide to Understanding Human Error.

2025-12-11 17:16:28 UTC

- reply

I've managed to bring about a small amount of change in organisational behaviour over the years, certainly more than my level of seniority would lead you to suspect was feasible. In every single case I've done it by starting with conversations to ask why things are the way they are and taking a non-judgemental approach. You have to meet people where they are and assume there are things about their experiences that you don't know.

2025-12-11 17:12:52 UTC

- reply

Aside from everything else, if you actually want to persuade people to stop using LLMs, you need to start by trying to understand *why* they are using these technologies. Otherwise I have to conclude you don't want to influence change so much as judge folk because it makes you feel better about yourself.

2025-12-11 17:10:04 UTC

I mentioned on bluesky that several friends use LLMs to automate bureaucratic crap in their day jobs, and I get a guy confidently explaining that they clearly can't tell the difference between what matters and what doesn't.. These are mostly folk who work in non profits and public sector orgs, who do important work for low pay, and are automating timewasting, energy sapping nonsense to let them focus on the shit they actually care about, but this eejit is comfortable in his moral superiority.

2025-11-29 13:48:13 UTC

- reply

Over on bluesky where the conversation is *a lot more heated* I've seen folk telling engineering leads they're being unfair in expecting contributors to understand the code in a commit before it can be merged.. That is a recipe for harm without accountability.

2025-11-29 13:45:30 UTC

- reply

My less popular opinion is that LLM assisted coding can also be part of a pathway to accessing those opportunities, but not by having the human avoid the code!

2025-11-29 13:39:01 UTC

- reply

We also have an obligation to mitigate the harms software systems can cause, and to manage accountability. That also depends on people understanding code.
More code in the world is going to require *more* people with coding skills, gen AI makes it more urgent that we teach coding.

2025-11-29 13:37:10 UTC

I sometimes see people describing resistance to LLM assisted coding as gatekeeping and I get that, software engineering culture is absolutely rife with gatekeeping, I quite literally dedicated dedicated my career to developer education because I want more people to have access to these opportunities.
But here's the thing, in order to fully participate in the world of software you will still need to learn to code, giving people the impression that isn't the case is not empowering.

2025-11-23 12:05:25 UTC

"Writing code is easy. Once you have a solution in mind, and have mastered the syntax of your favorite programming language, writing code is easy. Having an LLM write entire functions for you? Even easier. But the hard part isn’t the writing. It’s the reading. It’s the time it takes to load the mental model of the system into your head. That’s where all the cost really is."
https://idiallo.com/blog/writing-code-is-easy-reading-is-hard

2025-11-07 22:39:32 UTC

- reply

When you read the book and read anything about what Mary Shelley was going through this conclusion becomes inevitable

2025-11-07 22:38:38 UTC

I like Guillermo del Toro but I also think men should stop making versions of Frankenstein

2025-11-07 15:02:12 UTC

- reply

that's the thing, I think we need to get a lot more intentional about good practices for using these tools *in the context of team collaboration* both in terms of how it fits into the puzzle when pairing happens between more and less experienced folk, code review, etc etc, it's all a bit chaotic at the moment

2025-11-07 14:53:13 UTC

- reply

yeah you don't need to read it all lol, I'm just glad to see more of this thinking out there

2025-11-07 14:41:58 UTC

"The more Copilot handles, the less devs feel responsible for the code they ship."
A lot of interesting stuff in here
https://dev.to/dev_tips/copilot-is-gaslighting-developers-and-were-all-pretending-its-fine-51j9

2025-11-07 11:45:39 UTC

My conviction that most LLM coding tooling is aimed at avoiding learning rather than enabling it just keeps getting stronger. Today I'm looking at debugging tools for LLM generated code and you'll never guess what everything I'm finding uses, why it's more LLMs of course. 🤡

2025-10-11 10:44:48 UTC

- reply

If anything I would say the glorification of specialist skills correlates with a lot of the toxicity in dev communities. It goes hand in hand with a refusal to consider perspectives beyond your own.

2025-10-11 10:42:44 UTC

- reply

Often I see software developers refer to generalists in a disparaging way, and for sure economic dynamics are forcing teams to do more with less which reduces the potential for specialism and that can be a problem, but there is tremendous value in the perspectives that come with a generalist approach in my experience.

2025-10-11 10:40:17 UTC

Something I think is important to understand when it comes to what we call specialists vs generalists is that the distinction isn't between being skilled or unskilled.. Generalists use a different set of skills, at a different abstraction level within their field or organisation or whatever. Once again, tis not a zero sum game.

2025-09-28 17:58:36 UTC

A very solid quality of life multiplier I can recommend is getting the full Columbo series on DVD should you have the capacity to play them

2025-09-27 22:07:35 UTC

Saw this shit in a hotel bar I'll be taking all my friends who like books to, just to annoy the absolute fuck out of them
https://files.glasgow.social/media_attachments/files/115/278/526/957/804/912/original/93e757102ffef9cb.jpg

2025-09-23 17:15:49 UTC

- reply

CPP - covert person prompt
Need to workshop names

2025-09-23 17:13:17 UTC

- reply

Trying to pin down the characteristics of these comments / questions designed to prompt human content generation 🥴
* They are generic and don't refer to a specific project the person is working on
* They use stock phrases you'd see in listicle type articles (asking about options and tradeoffs etc)
* They can be described as requests for content (RFCs lolol)
I've been checking the poster profiles as well, if they do a lot of similar comments etc

2025-09-23 13:18:58 UTC

- reply

I mean, you can go far with content up to the point in time when all this started, but the utility of that will fade pretty fast surely

2025-09-23 13:15:56 UTC

- reply

This has kind of blown my mind, we've gone so quickly from the storyline that LLMs will be generating all the content in future to folk trying to trick humans into generating more content to feed the LLMs.. It's all unravelling so fast, if we are actually able to build bot scraping protection that covers a decent chunk of the web in the short term this business model is surely going to just collapse.. The part that is predicated on stealing other people's content without consent anyway

2025-09-23 10:26:29 UTC

- reply

Bowfin! 🤮

2025-09-23 10:25:41 UTC

- reply

And you're the kind of person who is trying to keep community spaces alive hence you posting on these platforms in the first place.. 💀

2025-09-23 10:24:22 UTC

- reply

The horrible thing about this is that it's one of those situations where there's nothing obviously untoward but your brain goes eeehhh something's no right here and unless you pay attention to it you'll likely just respond, especially folk who share tutorials and the like freely by default.

2025-09-23 10:08:31 UTC

- reply

"Why do you ask?"
I think I'll be responding to questions with this a lot more

2025-09-23 09:55:34 UTC

- reply

So when you've cannibalised all of the existing content on the web what do you do to get more? Trick people into giving you it I guess.

2025-09-23 09:53:36 UTC

- reply

Turns out I am not the only person paranoid enough to have asked this question https://www.reddit.com/r/gamingsuggestions/comments/1n37n8a/are_we_getting_farmed_by_some_bot_questions/

2025-09-23 09:35:24 UTC

- reply

The response to that would be a dead giveaway I think if there was one

2025-09-23 09:34:50 UTC

- reply

I guess you could add a wee bit of human passable friction like "could you explain more about why you ask this? what are you working on" or something

2025-09-23 09:33:47 UTC

- reply

In which case, I am certainly not taking the bait and providing more free content to help kill the spaces this is cannibalising.. but the risk of not engaging with genuine folk is always there, it's a judgement call for every post

2025-09-23 09:31:45 UTC

- reply

Which leaves me wondering, I now see a lot of posts in places on the web that look like prompts, so is part of the AI bot traffic a method for stimulating human content generation to feed the LLMs by essentially posting prompts in places where people provide content..? Github issues, developer tutorial comments, forums etc..

2025-09-23 09:28:13 UTC

- reply

If it was a person using LLMs to polish their writing I'd happily respond, but these feel like questions you'd use to seed content if that makes sense, it's shit like "what are the tradeoffs" etc that I must assume a human commenter on a developer tutorial would not care about, they sound like prompts for generating articles basically

2025-09-23 09:26:01 UTC

- reply

It's the kind of thing that wouldn't be caught in spam filters or whatever unless the user is posting too frequently or getting flagged a lot, but the content looks plausible, it just feels off, stock phrases etc that now make me think it's not a question a human being wrote..

2025-09-23 09:22:25 UTC

Normally when I post tutorials I make a point of being responsive to comments but these days so many of them "feel" AI generated and I'm inclined not to respond, I hate what this shit is doing to online interaction

2025-09-13 10:03:25 UTC

- reply

It's another way in which LLMs reflect the misprioritisation in dev learning, itself a reflection of the economic dynamics.. we always focused too much on writing code over reading it, and this production line mentality is why anyone would believe a machine that generates code could replace a developer
We're having to fully articulate the skills in building software for the first time

2025-09-13 09:55:41 UTC

This is why I believe we're going to need better tools for reading and exploring codebases
"Writing code is easy. Once you have a solution in mind, and have mastered the syntax of your favorite programming language, writing code is easy. Having an LLM write entire functions for you? Even easier. But the hard part isn’t the writing. It’s the reading. It’s the time it takes to load the mental model of the system into your head. That’s where all the cost really is."
https://idiallo.com/blog/writing-code-is-easy-reading-is-hard

2025-08-09 16:14:53 UTC

Fooling people into thinking you know about wine when you've actually just seen the Carsini Columbo episode twelve times
https://files.glasgow.social/media_attachments/files/114/999/690/959/678/485/original/c4ed6aca2c314414.jpg

2025-08-04 13:28:28 UTC

- reply

There's a guide in here on teaching schoolchildren not to anthropomorphise AI and if we could have journalists attend the same classes that would be great thx

2025-08-02 15:39:40 UTC

"I vibe coded enterprise grade software in four days"
I'm gonnae go out on a limb and say naw, ye didnae
The primary reason such software normally takes longer than this isn't because the programmers can't type fast enough ffs

2025-08-01 14:36:31 UTC

- reply

Of course I should have guessed this but the raspberry pi foundation have some great guides to coding pedagogy in the age of ai which include computational thinking and anthropomorphism https://www.raspberrypi.org/blog/pedagogy-quick-reads-turning-abstract-ideas-into-classroom-practice/

2025-08-01 14:08:34 UTC

Another tip I'm seeing about coding with LLMs is that it's most effective if you break a project into smaller tasks, so... Decomposition
Again, I believe computational thinking skills are going to remain essential

2025-07-14 15:36:58 UTC

- reply

Not exactly an abstraction but another glitch affordance I found so valuable for education purposes was the learner not having to download or install anything on their computer, it was a great enabler but everything being in the cloud is far from ideal from a cost, energy, or connectivity standpoint.. Curious about platforms like stackblitz that do it by running containers locally in the browser..

2025-07-13 18:10:53 UTC

- reply

So tldr it's just a big fucking cocktail of people who don't respect other people's skills 🤡

2025-07-13 18:08:31 UTC

- reply

Another fun repeated theme in these spaces is folk who are coders and are managing to churn out completed apps but then realising things like product market fit, sales and marketing in general, are actually important.. will it make them respect coworkers who do these things in their day jobs perhaps, I wonder lol

2025-07-12 11:31:23 UTC

- reply

What you've done there is set up a doomed consultancy

2025-07-12 11:30:29 UTC

- reply

There's also an ai product from ex faang employees designed to "finish the job" for your llm generated apps and their deal is if their tool can't do it they'll do it via professional services which is peak "oh mate" lol

2025-07-12 10:27:55 UTC

- reply

I'm still on vibe coding reddit and there's a recurring pattern in the replies to posts like this
"I'm a non-coder who loves the idea of creating software that works the way I want it to. [...] but apart from very simple HTML/CSS projects, I've never managed to complete something that actually worked. It always ends in an endless loop of the Al trying to fix bugs."
People will say oh but the models are getting better, that they've been successful *but they already know how to code*

2025-07-10 17:53:26 UTC

- reply

What I'm most concerned about as an educator is the paths to *becoming* a senior software engineer collapsing, and lo I will be exploring learning experiences that start from an existing app

2025-07-10 17:49:05 UTC

- reply

Gonnae be a lot more red diffs

2025-07-10 17:47:23 UTC

- reply

I guess one way to put this would be what happens if you ask a senior engineer whether the existence of lots more code in the world makes their involvement more or less necessary lol

2025-07-10 14:53:34 UTC

- reply

And the typical stacks these vibe coding tools generate also tend to use the most ludicrously overcomplicated frameworks the web has been characterised by in recent years which is somewhat ironic lol. If the generated app was using vanilla shit it would be a hell of a lot easier to dig into the code but instead learners will be getting flung in at the deep end if they start this way.

2025-07-10 14:50:42 UTC

- reply

It makes me think about other abstractions in software and the deal you make when you commit to one.. One of the abstractions in glitch that made it a great enabler at one time was the removal of a deployment step.. Your app was instantly online which was a hugely valuable automation but as the web changed became more of a problem for various reasons

2025-07-10 14:47:04 UTC

- reply

A huge part of the smoke and mirrors at work is that you very quickly get what looks closer to a *real thing* than it is if you don't know how software actually works, it appears like you're on the home stretch almost immediately when you've actually just heard the gun go off

2025-07-10 14:43:13 UTC

- reply

I'd also double down on my earlier belief that these tools are primarily going to be used most effectively by software engineers, both in terms of the specification step relying on understanding of computational processing, and that the bulk of the work is not in generating code

2025-07-10 14:40:11 UTC

- reply

There are also multiple posts from folk realising they are going to need to learn to code or hire a developer to fix an issue, get their app to the next stage, etc etc..
With the platforms also ramping up charges I think it's safe to say that even with improvements around debugging it's becoming clearer that LLMs will only take you a small portion of the way to building a *real thing* with code

2025-07-10 14:21:52 UTC

I've started browsing reddit for vibe coding chat to get a better sense of the state of things and it has been eye opening.. For example one today titled "How do I give the website to the client"
...
Let's just say there is going to be plenty of work for developers, the jobs might dry up temporarily but someone's going to need to clean up the mess, and anyone who's business is tangled up in these apps will need to be ready to pay a premium to sort their shit out

2025-06-19 11:46:42 UTC

One of the things I love about these protests is that they demonstrate how *misogynist* the associated policies and expectations are, they are simultaneously feminist protests and as a side effect these women are showing how utterly contradictory TERFism is ♥️
https://www.out.com/gay-athletes/trans-woman-swims-topless-england

2025-05-27 17:43:10 UTC

Always love to see debates about the supposed benefits and drawbacks of remote work because if it wasn't an option I'd simply need to leave the tech industry

2025-05-22 09:30:02 UTC

Looking at the microsoft github conversations everyone is laughing about has me wondering about the suggestion that AI assisted coding with an agent is just a new abstraction level.. The devs trying to wrangle copilot are having to dig in at a level of detail that I think implies that analogy is wrong
https://www.reddit.com/r/ExperiencedDevs/comments/1krttqo/my_new_hobby_watching_ai_slowly_drive_microsoft/

2025-05-04 00:20:12 UTC

My take on the - computer science educator only teaches computer science and not politics - discourse.. is that it very much constitutes telling on yourself as an educator if you are incapable of creating the conditions for true learning, which means supporting inquiry as learners develop their understanding of *literally any subject* within a social context, and idk maybe find another vocation because it doesnae sound like teaching is for you sweetheart

2025-05-03 09:36:57 UTC

A post on another platform where someone has a "stronger than cancer" t-shirt on because they've survived it and I... other people don't die from cancer because they are not strong enough, tae fuck with this shite

2025-04-24 16:12:43 UTC

While I hesitate to encourage folk to run unnecessary queries, I do think the "made up phrase" + "meaning" google search is a great exercise in helping people understand what generative AI actually does.. I've found it hard to explain to folk outside tech that its purpose is literally just to generate text but this seems a more intuitive demo that might make them question the output.
https://files.glasgow.social/media_attachments/files/114/393/810/678/175/711/original/1c0b9b4b4e866c4e.jpg

2025-02-19 09:38:07 UTC

DHH going full mask off racist is a good time for a reminder that him and his wee pal wrote all that organisational management 💩thought leadership💩 based on running a company with *a few dozen employees* and is basically the manager of your local supermarket posting on facebook about how you can't say anything these days

2025-02-15 17:22:29 UTC

- reply

Still so pissed off at William Gibson for making me think the tech fueled dystopia was going to be cool and not cringey as fuck

2025-02-15 17:20:45 UTC

Mind when tech job titles had "ninja" and shit in them, truly the most embarrassing industry

2025-02-13 11:16:26 UTC

- reply

To go back to the Microsoft research https://glasgow.social/@sue/113980780460436876
They found that "users with access to GenAI tools produce a less diverse set of outcomes for the same task ... This tendency for convergence reflects a lack of personal, contextualised, critical and reflective judgement of AI output"
So we're working towards your output always falling within the bounds of what is defined as "refined" or whatever lol based on the model, what could possibly go wrong etc

2025-02-13 10:47:45 UTC

- reply

"We've modelled our AI assistant on an editor who hates you" cool cool

2025-02-13 10:45:44 UTC

- reply

I just want to know who came up with "refine" as the name for the action here... Something tells me this isn't the work of people who value their coworkers lolol

2025-02-13 10:43:06 UTC

Google workspace AI offers to "refine" your writing... You've got the human contribution being framed as unrefined, you've even got an unintentional severance reference, it's all there! Bargain basement scifi shit.

2025-02-04 11:28:52 UTC

Something that really annoys me with text editing apps that give you intrusive suggestions as you type (AI or other) is that a lot of the time the writing is part of my thought process. The goal is more complex than getting the words on the page, it's part of me figuring the thing out and these prompts interfere with that, aggressively so. 🤬

2025-02-03 18:42:58 UTC

- reply

What an incredible privilege to have access to all of this without worrying about how much it costs, seriously fuck anyone who messes with the NHS

2025-02-03 17:37:48 UTC

- reply

The anaesthetic nurse + student walked me along, said "there are going to be a lot of people in here, they're all here for you" and there they all were just like in the picture lol, you're like hiya.. Then you lie down and they all descend on you, in a nice way 🤣
https://files.glasgow.social/media_attachments/files/113/941/172/063/701/548/original/84d3a2d79e7cc5c9.jpg

2025-02-03 17:31:40 UTC

- reply

Walking into the operating theatre was not something I expected, they did it because they'd decided not to give me a spinal injection but it was quite a surreal moment

2025-02-03 17:30:45 UTC

This is the consultant who did my treatment planning, and the operating theatre looks just like it did when they got me to walk in there, with a load of folk standing waiting lol ♥️ https://www.glasgowlive.co.uk/news/glasgow-news/glasgow-pensioner-worried-after-bleeding-30919401

2025-01-31 19:12:54 UTC

My massive expensive telly is acting the pig, needless to say the cheap one we had before it lasted almost twenty years before even the slightest malfunction 💩

2025-01-16 09:39:50 UTC

This pattern of toxic CEOs posting from corporate social media accounts, forcing highly visible product changes based on their own personal drama etc is certainly revealing what joke organisations populate the tech industry lol