From the mailbag #0 - Decisions, decisions

Sooner or later, when one writes professionally or semi-professionally, one runs into the issue where you have writers' block and so can't write, but you still have to write to keep the lights on and pay the bills. We all find different ways of getting around this problem: some people just push through it and develop a tolerance for publishing work that they think is a little mediocre, some people just deal with the lapse, some people try and write something out of the ordinary... you get the idea.

For my part, I recalled that in general, audiences like to engage with the people whose work they read rather than just passively absorbing it. This is something that I'm not very good at if I'm to be honest, but given that I've been struggling to get my latest pieces to work as I'd like, I had an idea.

What I wound up doing was setting up an AMA page on my blog where readers can submit questions. I pulled it together and threw it up online in a few hours last night, and come the morning I had some interesting questions to answer! They're on a variety of subjects, from writing fiction, to education, to the pervasive feeling that a lot of people have that they don't quite belong in tech. I've answered them here in what I hope will be an interesting way.

If you have a question that you want to ask me, you can do so here: depending on how many questions I get, I might make these a semi-regular occurence. Now, on to the questions!

The questions

Anonymous asks:

I'm a technical coach/trainer (and former software developer, in the industry for 35+ years) and my (former) clients have no need of me with all of the layoffs and now, especially LLMs. I dislike LLMs for many reasons, mostly ethical. Should I write a book (which nobody will read?), create a new workshop, or admit defeat and retire?

Well, first off, I'd not dismiss the idea of writing a book out of hand. You really never know what will or won't get read before you write it: I have quite a few readers, after all, and what I write is pretty recondite at best and downright strange at worst. You're highly unlikely to find that you have no readers at all, and if you can make life better for even ten or twenty people, that's no small thing. And who knows? It might prove a lot more popular than you think.

But that's a bit of a side-track from the main thrust of your question, which seems to be whether or not you should keep trying to make your coaching and training work at a time when there are massive layoffs and LLMs are reshaping what remains of the industry beyond recognition. My thoughts on this run two ways: if you don't find the work fulfilling or think it's pointless, of course you should stop and retire. There's no moral obligation for you to continue doing something that you find unpleasant or that you think is pointless, and if that's where you're at, doing something else might be a good idea.

The sense I get, though, is that you still enjoy the work and find it useful but are unsure whether you can make it pay in the current environment. If you want to keep doing this work, there are ways in which you'll have to adapt, but it's still very much possible and useful. LLMs do not, after all, eliminate the need to actually know how to do things without the tools, and trying to use the LLM to get results without that knowledge tends to end badly. And as you've noted, people no longer really need to know about the details of writing code as such.

What does this leave? If you're going to occupy the coaching/training space these days, you most likely want to be moving away from more vocational skills and towards areas where LLMs have had, at best, a muted impact. Architecture, for example, is one thing that LLMs remain resolutely bad at and that requires a lot of human judgement to execute on. For my part, I mostly teach statistics these days as it's a subject that a) still requires a great deal of human judgement and b) builds transferable skills, knowledge and judgement that are useful regardless of whether or not you ever use the actual skills being taught directly. In general, I think that if you teach things like that and market them effectively, it should be workable.

Whatever you decide to do going forward, I wish you every success!

Milk asks:

Do you have recommendations on fighting mediocrity in organizations and within oneself?

Within oneself avoiding mediocrity is simple, if difficult, and you've already done the most important thing because you don't want to be mediocre. If you don't want to be mediocre and you care enough about it to put actual effort and action behind it, mediocrity is unlikely. Now, you might still not be very good at what you're turning your hand to, or might be doing something in the wrong place, but you will steadily improve and you will, as sure as night follows day, become competent. The competence will probably initially be in quite restricted areas, but being competent in small matters and being mediocre are not the same thing. So... very simple, really. Just put the effort in, and make sure you learn something from every piece of work you do.

The organisational question is significantly harder, and I'm afraid I don't really have a good answer here: in my experience once mediocrity has become the norm in an organisation, it's almost impossible to get rid of it. While there are a few ways to get rid of a tendency towards mediocrity in an organisation, they all involve having more or less absolute power over the organisation to start with: with that being done, you can essentially disassemble the entire organisation and put it back together without the structures that cause mediocrity in it. It goes without saying that this is expensive, risky and often leads to the collapse of the organisation in its entirety.

Unless you can do something like that, your best option is usually to build small pockets of competence in your organisation wherever you can shield them from upper management or whoever else is responsible for maintaining the culture of mediocrity. In the right kind of organisation this can work quite well: you can create teams or even parts of departments that are competent in what are otherwise deeply mediocre organisations. However, these are always vulnerable to people in power fucking the whole thing up on a whim, and rather than seeing them as a way to reform an organisation, they're probably best used to give people the chance to leave and go somewhere more competent safely. In general, you're usually better off just finding a better organisation and supplanting the mediocre one rather than trying to fix it.

Jonas asks:

What role do you see sensitive people playing in technology? In your post “The problem is culture” you touched upon the differences in attitudes between the technological culture and the engineering culture. Recently, I keep thinking about one way these differences seem to manifest: communication styles. The pervasiveness of Bay Area tech culture comes with a certain style of communication. It honestly reminds me of chan culture. Conversations are not conversations but verbal duels. Sarcastic remarks and one-upmanship are ever present. Speakers are encouraged to spray manure everywhere and us listeners have to dig through the pile looking for a single nugget of gold that we are lucky to have bestowed upon us. As a sensitive person, I am tired of the mental effort all this takes. I am very tired of being told to grow thicker skin because I expect professionalism.

...I would gently suggest that there's no reason you need to let some of the worst, most socially maladjusted people in the world set the terms of the discussion. The same people whom you describe as communicating in the way you describe gave us thousands of cryptocurrency scams, NFTs, social media in its current, rather awful forms, and in the extreme, an actual, literal CSAM and Nazi propaganda generator in the form of Grok. The same kind of empathy for their fellow people that would have prevented them from talking the way they do would have prevented this, and the failure to moderate in how they speak is reflected in a failure to moderate in what and how they build.

All things considered, I don't know what kind of role sensitive people have to play in technology, or if you'd even be considered unusually sensitive in a better tech industry. What I can say is that there's really no place in tech for people who communicate in the way you describe. We all deserve to be treated professionally and with kindness, and honestly, I doubt that much good comes from the people who haven't been properly socialised and can't behave in a minimally civil way in the workplace. If they have done any good, it's in spite of how they behave and certainly not because of it.

So, what role do I see sensitive people playing in technology? All of them, ideally.

The Mana Pool asks:

Do you have many opinions on writing fiction, as opposed to your current styles of writing? Do you see yourself treading in this area much, or in creative writing more generally? I understand they're extremely distinct in their stylings and topics, but am still interested in hearing your thoughts as a creative writer myself!

Well, as you've noticed, I am an essayist first and foremost, and honestly I'm generally kinda shit at writing fiction (I love reading it, but that's another matter entirely). I've written the occasional story or bit of verse that was kinda OK (imitation Eliot, alas) , but in general writing fiction is something that I really struggle with. So I don't have much to say on that front, I fear.

My thoughts on writing in general are pretty easy to summarise though. First off, you can't write well unless you read well and widely: ex nihilo nihil fit, after all, and if you don't have raw matter to work with, it's very, very easy to fall into cliche when writing. Moreover, you usually form a voice of your own by reading a lot of other people's writing, seeing their voices, understanding them and borrowing what you like from them. So, making time to read and really focus on what you're reading is always a good idea.

Secondly, deliberate practice is always important. The commonplace advice is that you should write frequently is very true, but not enough: if you just write the same thing over and over without deliberately reviewing and aiming to improve, you will improve, but slowly. In general, it's a good idea to try and improve one specific thing that annoys you about your writing every time you write something new, and it's also a good idea to try writing in new styles and formats when you can (which is partly what this article is). Writing in a breadth of different ways will help you develop your writing skills much faster than sticking to one format all the time.

For similar reasons, you should aim to publish something for a public audience as soon as you can. Without having published something, writing can be really rather theoretical: it's something you've just kinda written. Once you've actually written and built a bit of an audience, though, you start getting a sense of where your writing is good and where it falls down that you just didn't before by virtue of interacting with your audience. This isn't to say that your audience is always right: sometimes they're horribly, horribly wrong, and when that happens it's important that you catch that and don't do what your audience says. Of course, you can't really develop a sense for that without having an audience in the first place. All in all, it's pretty boring, simple stuff: you just have to actually do it, and you'll know as well as anyone that that's kinda hard to do.

If it's a question of what kind of fiction/creative writing I like, as opposed to thoughts on process? I adore Thomas Pynchon, and Eliot is a favourite poet: my tastes in general run pretty modernist to postmodern. Hunter S. Thompson represents an interesting blend of journalism and fiction that I could see myself trying to work with, and I'm also pretty fond of Vonnegut. I like Iain Banks a lot as well (both the literary fic and the sci-fi), and Charles Stross's Laundry Files are a lot of fun.

Curtis asks:

Hi Iris, My question relates to your experience teaching statistics and math topics to people new to the domain, such as executives and those in software-adjacent roles (e.g., product managers and designers). What strategies do you employ to introduce non-intuitive concepts like probability distributions and Bayes' Theorem? For context, I'm working on an introductory course for privacy-enhancing technologies (such as differential privacy), and I get mixed advice for how much or how little math content to include in the course material. Thank you for your writing, I love the blog!

A good question, I think. Our job is somewhat easier than it might otherwise be because a lot of people whom we teach to have already developed abstraction skills more generally, and the single hardest thing when it comes to teaching stats and maths is trying to build up abstraction capabilities from nothing, which we mostly don't have to do.

Given that, my usual approach is pretty mathematical: I begin with building up the definition of probability using the probability measure from the baseline of set theory, then develop random variables and build into probability distributions from that. For professionals, I think that's the best approach even though some people will find it challenging: struggling with the concepts and then eventually understanding them tends to make things stick a lot better than the situation when the concepts are presented easily enough that no effort is required.

Probability does have some advantages when doing this though, because you can perform statistical experiments with cards and dice and stuff, and those are things that most people have the skills to reason about. Even when there's a struggle, you can get people to actually physically run experiments themselves. That makes it comparatively easy to get in, and it helps that the mathematical infrastructure underlying probability (set theory, set functions, boolean logic) are pretty intuitive and easy to grasp, even for people who've not had to deal with them before. If your field assumes a knowledge of algebra or calculus, you're going to struggle a bit more.

So in general, I'm not against the idea of using quite a lot of maths, but it has to be the right kind: set theory and logic are easier than algebra or calculus. If you can actually get people to physically run experiments or simulations to demonstrate the concepts you want them to learn, that's also a really good idea.

Sign-off

That's all the questions for this episode of From the mailbag. I don't know when we'll do another of these (much depends on how many questions I get and how interesting they are), but for now, thank you to everyone who's sent questions in and I hope you got the answers you wanted!