Journal tags: language

I got a nice email from someone regarding my recent posts about performance on The Session. They said:

I hope this message finds you well. First and foremost, I want to express how impressed I am with the overall performance of https://thesession.org/. It’s a fantastic resource for music enthusiasts like me.

How nice! I responded, thanking them for the kind words.

They sent a follow-up clarification:

Awesome, anyway there was an issue in my message.

The line ‘It’s a fantastic resource for music enthusiasts like me.’ added by chatGPT and I didn’t notice.

I imagine this is what it feels like when you’re on a phone call with someone and towards the end of the call you hear a distinct flushing sound.

I wrote back and told them about Simon’s rule:

I will not publish anything that takes someone else longer to read than it took me to write.

That just feels so rude!

I think that’s a good rule.

I just described prototype code as code to be thrown away. On that topic…

I’ve been observing how people are programming with large language models and I’ve seen a few trends.

The first thing that just about everyone agrees on is that the code produced by a generative tool is not fit for public consumption. At least not straight away. It definitely needs to be checked and tested. If you enjoy debugging and doing code reviews, this might be right up your street.

The other option is to not use these tools for production code at all. Instead use them for throwaway code. That could be prototyping. But it could also be the code for those annoying admin tasks that you don’t do very often.

Take content migration. Say you need to grab a data dump, do some operations on the data to transform it in some way, and then pipe the results into a new content management system.

That’s almost certainly something you’d want to automate with bespoke code. Once the content migration is done, the code can be thrown away.

Read Matt’s account of coding up his Braggoscope. The code needed to spider a thousand web pages, extract data from those pages, find similarities, and output the newly-structured data in a different format.

I’ve noticed that these are just the kind of tasks that large language models are pretty good at. In effect you’re training the tool on your own very specific data and getting it to do your drudge work for you.

To me, it feels right that the usefulness happens on your own machine. You don’t put the machine-generated code in front of other humans.

Back when the web was young, it wasn’t yet clear what the rules were. Like, could you really just link to something without asking permission?

Then came some legal rulings to establish that, yes, on the web you can just link to anything without checking if it’s okay first.

What about search engines and directories? Technically they’re rifling through all the stuff we publish and reposting snippets of it. Is that okay?

Again, through some legal precedents—but mostly common agreement—everyone decided that on balance it was fine. After all, those snippets they publish are helping your site get traffic.

In short order, search came to rule the web. And Google came to rule search.

The mutually beneficial arrangement persisted uneasily. Despite Google’s search results pages getting worse and worse in recent years, the company’s huge market share of search means you generally want to be in their good books.

Google’s business model relies on us publishing web pages so that they can put ads around the search results linking to that content, and we rely on Google to send people to our websites by responding smartly to search queries.

That has now changed. Instead of responding to search queries by linking to the web pages we’ve made, Google is instead generating dodgy summaries rife with hallucina… lies (a psychic hotline, basically).

Google still benefits from us publishing web pages. We no longer benefit from Google slurping up those web pages.

With AI, tech has broken the web’s social contract:

Google has steadily been manoeuvring their search engine results to more and more replace the pages in the results.

As Chris puts it:

Me, I just think it’s fuckin’ rude.

Google is a portal to the web. Google is an amazing tool for finding relevant websites to go to. That was useful when it was made, and it’s nothing but grown in usefulness. Google should be encouraging and fighting for the open web. But now they’re like, actually we’re just going to suck up your website, put it in a blender with all other websites, and spit out word smoothies for people instead of sending them to your website. Instead.

Ben proposes an update to robots.txt that would allow us to specify licensing information:

Robots.txt needs an update for the 2020s. Instead of just saying what content can be indexed, it should also grant rights.

Like crawl my site only to provide search results not train your LLM.

It’s a solid proposal. But Google has absolutely no incentive to implement it. They hold all the power.

Or do they?

There is still the nuclear option in robots.txt:

User-agent: Googlebot
Disallow: /

That’s what Vasilis is doing:

I have been looking for ways to not allow companies to use my stuff without asking, and so far I coulnd’t find any. But since this policy change I realised that there is a simple one: block google’s bots from visiting your website.

The general consensus is that this is nuts. “If you don’t appear in Google’s results, you might as well not be on the web!” is the common cry.

I’m not so sure. At least when it comes to personal websites, search isn’t how people get to your site. They get to your site from RSS, newsletters, links shared on social media or on Slack.

And isn’t it an uncomfortable feeling to think that there’s a third party service that you absolutely must appease? It’s the same kind of justification used by people who are still on Twitter even though it’s now a right-wing transphobic cesspit. “If I’m not on Twitter, I might as well not be on the web!”

The situation with Google reminds me of what Robin said about Twitter:

The speed with which Twitter recedes in your mind will shock you. Like a demon from a folktale, the kind that only gains power when you invite it into your home, the platform melts like mist when that invitation is rescinded.

We can rescind our invitation to Google.

I’m sure you’ve heard the law of the instrument: when all you have is a hammer, everything looks like a nail.

There’s another side to it. If you’re selling hammers, you’ll depict a world full of nails.

Recent hammers include cryptobollocks and virtual reality. It wasn’t enough for blockchains and the metaverse to be potentially useful for some situations; they staked their reputations on being utterly transformative, disrupting absolutely every facet of life.

This kind of hype is a terrible strategy in the long-term. But if you can convince enough people in the short term, you can make a killing on the stock market. In truth, the technology itself is superfluous. It’s the hype that matters. And if the hype is over-inflated enough, you can even get your critics to do your work for you, broadcasting their fears about these supposedly world-changing technologies.

You’d think we’d learn. If an industry cries wolf enough times, surely we’d become less trusting of extraordinary claims. But the tech industry continues to cry wolf—or rather, “hammer!”—at regular intervals.

The latest hammer is machine learning, usually—incorrectly—referred to as Artificial Intelligence. What makes this hype cycle particularly infuriating is that there are genuine use cases. There are some nails for this hammer. They’re just not as plentiful as the breathless hype—both positive and negative—would have you believe.

When I was hosting the DiBi conference last week, there was a little section on generative “AI” tools. Matt Garbutt covered the visual side, demoing tools like Midjourney. Scott Salisbury covered the text side, showing how you can generate code. Afterwards we had a panel discussion.

During the panel I asked some fairly straightforward questions that nobody could answer. Who owns the input (the data used by these generative tools)? Who owns the output?

On the whole, it stayed quite grounded and mercifully free of hyperbole. Both speakers were treating the current crop of technologies as tools. Everyone agreed we were on the hype cycle, probably the peak of inflated expectations, looking forward to reaching the plateau of productivity.

Scott explicitly warned people off using generative tools for production code. His advice was to stick to side projects for now.

Matt took a closer look at where these tools could fit into your day-to-day design work. Mostly it was pretty sensible, except when he suggested that there could be any merit to using these tools as a replacement for user testing. That’s a terrible idea. A classic hammer/nail mismatch.

I think I moderated the panel reasonably well, but I have one regret. I wish I had first read Baldur Bjarnason’s new book, The Intelligence Illusion. I started reading it on the train journey back from Edinburgh but it would have been perfect for the panel.

The Intelligence Illusion is very level-headed. It is neither pro- nor anti-AI. Instead it takes a pragmatic look at both the benefits and the risks of using these tools in your business.

It has excellent advice for spotting genuine nails. For example:

Generative AI has impressive capabilities for converting and modifying seemingly unstructured data, such as prose, images, and audio. Using these tools for this purpose has less copyright risk, fewer legal risks, and is less error prone than using it to generate original output.

Think about transcripts of videos or podcasts—an excellent use of this technology. As Baldur puts it:

The safest and, probably, the most productive way to use generative AI is to not use it as generative AI. Instead, use it to explain, convert, or modify.

He also says:

Prefer internal tools over externally-facing chatbots.

That chimes with what I’ve been seeing. The most interesting uses of this technology that I’ve seen involve a constrained dataset. Like the way Luke trained a language model on his own content to create a useful chat interface.

Anyway, The Intelligence Illusion is full of practical down-to-earth advice based on plenty of research backed up with copious citations. I’m only halfway through it and it’s already helped me separate the hype from the reality.

I did an episode of the Clearleft podcast on innovation a while back:

Everyone wants to be innovative …but no one wants to take risks.

The word innovation is often bandied about in an unquestioned positive way. But if we acknowledge that innovation is—by definition—risky, then the exhortations sound less positive.

“We provide innovative solutions for businesses!” becomes “We provide risky solutions for businesses!”

I was reminded of this when I saw the website for the Podcast Standards Project. The original text on the website described the project as:

…a grassroots coalition working to establish modern, open standards, to enable innovation in the podcast industry.

I pushed back on that wording (partly because I’ve seen the word “innovation” used as a smoke screen for user-hostile practices like tracking and surveillance). The wording has since changed to:

…a grassroots coalition dedicated to creating standards and practices that improve the open podcasting ecosystem for both listeners and creators.

That’s better. It’s more precise.

Am I nitpicking? Only if you think that “innovation” and “improvement” are synonyms. I don’t think they are.

Innovation implies change. Improvement implies positive change.

Not all change is positive. Not all innovation is positive.

Innovation goes hand in hand with disruption. Again, disruption involves change. But not necessarily positive change.

Think about the antonyms of change and disruption: stasis and stability. Those words don’t sound very exciting, but in some arenas they’re exactly what you should be aiming for; arenas like infrastructure or standards.

Not to get all pace layers-y here, but it seems to me that every endeavour has a sweet spot for innovation. For some projects, too little innovation is bad. For others, too much innovation is worse.

The trick is knowing which kind of project you’re working on.

(As a side note, I think some people use the word innovation to describe the generative, divergent phase of a design project: “how might we come up with innovative new approaches?” But we already have a word to describe the practice of generating novel and interesting ideas. That word isn’t innovation. It’s creativity.)

I woke up today to a very annoying new bug in Firefox. The browser shits the bed in an unpredictable fashion when rounding up single pixel line widths in SVG. That’s quite a problem on The Session where all the sheet music is rendered in SVG. Those thin lines in sheet music are kind of important.

Browser bugs like these are very frustrating. There’s nothing you can do from your side other than filing a bug. The locus of control is very much with the developers of the browser.

Still, the occasional regression in a browser is a price I’m willing to pay for a plurality of rendering engines. Call me old-fashioned but I still value the ecological impact of browser diversity.

That said, I understand the argument for converging on a single rendering engine. I don’t agree with it but I understand it. It’s like this…

Back in the bad old days of the original browser wars, the browser companies just made shit up. That made life a misery for web developers. The Web Standards Project knocked some heads together. Netscape and Microsoft would agree to support standards.

So that’s where the bar was set: browsers agreed to work to the same standards, but competed by having different rendering engines.

There’s an argument to be made for raising that bar: browsers agree to work to the same standards, and have the same shared rendering engine, but compete by innovating in all other areas—the browser chrome, personalisation, privacy, and so on.

Like I said, I understand the argument. I just don’t agree with it.

One reason for zeroing in a single rendering engine is that it’s just too damned hard to create or maintain an entirely different rendering engine now that web standards are incredibly powerful and complex. Only a very large company with very deep pockets can hope to be a rendering engine player. Google. Apple. Heck, even Microsoft threw in the towel and abandoned their rendering engine in favour of Blink and V8.

And yet. Andreas Kling recently wrote about the Ladybird browser. How we’re building a browser when it’s supposed to be impossible:

The ECMAScript, HTML, and CSS specifications today are (for the most part) stellar technical documents whose algorithms can be implemented with considerably less effort and guesswork than in the past.

I’ll be watching that project with interest. Not because I plan to use the brower. I’d just like to see some evidence against the complexity argument.

Meanwhile most other browser projects are building on the raised bar of a shared browser engine. Blisk, Brave, and Arc all use Chromium under the hood.

Arc is the most interesting one. Built by the wonderfully named Browser Company of New York, it’s attempting to inject some fresh thinking into everything outside of the rendering engine.

Experiments like Arc feel like they could have more in common with tools-for-thought software like Obsidian and Roam Research. Those tools build knowledge graphs of connected nodes. A kind of hypertext of ideas. But we’ve already got hypertext tools we use every day: web browsers. It’s just that they don’t do much with the accumulated knowledge of our web browsing. Our browsing history is a boring reverse chronological list instead of a cool-looking knowledge graph or timeline.

For inspiration we can go all the way back to Vannevar Bush’s genuinely seminal 1945 article, As We May Think. Bush imagined device, the Memex, was a direct inspiration on Douglas Engelbart, Ted Nelson, and Tim Berners-Lee.

The article describes a kind of hypertext machine that worked with microfilm. Thanks to Tim Berners-Lee’s World Wide Web, we now have a global digital hypertext system that we access every day through our browsers.

But the article also described the idea of “associative trails”:

Wholly new forms of encyclopedias will appear, ready made with a mesh of associative trails running through them, ready to be dropped into the memex and there amplified.

Our browsing histories are a kind of associative trail. They’re as unique as fingerprints. Even if everyone in the world started on the same URL, our browsing histories would quickly diverge.

Bush imagined that these associative trails could be shared:

The lawyer has at his touch the associated opinions and decisions of his whole experience, and of the experience of friends and authorities.

Heck, making a useful browsing history could be a real skill:

There is a new profession of trail blazers, those who find delight in the task of establishing useful trails through the enormous mass of the common record.

Taking something personal and making it public isn’t a new idea. It was what drove the wave of web 2.0 startups. Before Flickr, your photos were private. Before Delicous, your bookmarks were private. Before Last.fm, what music you were listening to was private.

I’m not saying that we should all make our browsing histories public. That would be a security nightmare. But I am saying there’s a lot of untapped potential in our browsing histories.

Let’s say we keep our browsing histories private, but make better use of them.

From what I’ve seen of large language model tools, the people getting most use of out of them are training them on a specific corpus. Like, “take this book and then answer my questions about the characters and plot” or “take this codebase and then answer my questions about the code.” If you treat these chatbots as calculators for words they can be useful for some tasks.

Large language model tools are getting smaller and more portable. It’s not hard to imagine one getting bundled into a web browser. It feeds on your browsing history. The bigger your browsing history, the more useful it can be.

Except, y’know, for the times when it just make shit up.

Vannevar Bush didn’t predict a Memex that would hallucinate bits of microfilm that didn’t exist.

Picture someone tediously going through a spreadsheet that someone else has filled in by hand and finding yet another error.

“I wish to God these calculations had been executed by steam!” they cry.

The year was 1821 and technically the spreadsheet was a book of logarithmic tables. The frustrated cry came from Charles Babbage, who channeled his frustration into a scheme to create the world’s first computer.

His difference engine didn’t work out. Neither did his analytical engine. He’d spend his later years taking his frustrations out on street musicians, which—as a former busker myself—earns him a hairy eyeball from me.

But we’ve all been there, right? Some tedious task that feels soul-destroying in its monotony. Surely this is exactly what machines should be doing?

I have a hunch that this is where machine learning and large language models might turn out to be most useful. Not in creating breathtaking works of creativity, but in menial tasks that nobody enjoys.

Someone was telling me earlier today about how they took a bunch of haphazard notes in a client meeting. When the meeting was done, they needed to organise those notes into a coherent summary. Boring! But ChatGPT handled it just fine.

I don’t think that use-case is going to appear on the cover of Wired magazine anytime soon but it might be a truer glimpse of the future than any of the breathless claims being eagerly bandied about in Silicon Valley.

You know the way we no longer remember phone numbers, because, well, why would we now that we have machines to remember them for us? I’d be quite happy if machines did that for the annoying little repetitive tasks that nobody enjoys.

I’ll give you an example based on my own experience.

Regular expressions are my kryptonite. I’m rubbish at them. Any time I have to figure one out, the knowledge seeps out of my brain before long. I think that’s because I kind of resent having to internalise that knowledge. It doesn’t feel like something a human should have to know. “I wish to God these regular expressions had been calculated by steam!”

Now I can get a chatbot with a large language model to write the regular expression for me. I still need to describe what I want, so I need to write the instructions clearly. But all the gobbledygook that I’m writing for a machine now gets written by a machine. That seems fair.

Mind you, I wouldn’t blindly trust the output. I’d take that regular expression and run it through a chatbot, maybe a different chatbot running on a different large language model. “Explain what this regular expression does,” would be my prompt. If my input into the first chatbot matches the output of the second, I’d have some confidence in using the regular expression.

A friend of mine told me about using a large language model to help write SQL statements. He described his database structure to the chatbot, and then described what he wanted to select.

Again, I wouldn’t use that output without checking it first. But again, I might use another chatbot to do that checking. “Explain what this SQL statement does.”

Playing chatbots off against each other like this is kinda how machine learning works under the hood: generative adverserial networks.

Of course, the task of having to validate the output of a chatbot by checking it with another chatbot could get quite tedious. “I wish to God these large language model outputs had been validated by steam!”

Sounds like a job for machines.

You know how when you’re on hold to any customer service line you hear a message that thanks you for calling and claims your call is important to them. The message always includes a disclaimer about calls possibly being recorded “for training purposes.”

Nobody expects that any training is ever actually going to happen—surely we would see some improvement if that kind of iterative feedback loop were actually in place. But we most certainly want to know that a call might be recorded. Recording a call without disclosure would be unethical and illegal.

Consider chatbots.

If you’re having a text-based (or maybe even voice-based) interaction with a customer service representative that doesn’t disclose its output is the result of large language models, that too would be unethical. But, at the present moment in time, it would be perfectly legal.

That needs to change.

I suspect the necessary legislation will pass in Europe first. We’ll see if the USA follows.

In a way, this goes back to my obsession with seamful design. With something as inherently varied as the output of large language models, it’s vital that people have some way of evaluating what they’re told. I believe we should be able to see as much of the plumbing as possible.

The bare minimum amount of transparency is revealing that a machine is in the loop.

This shouldn’t be a controversial take. But I guarantee we’ll see resistance from tech companies trying to sell their “AI” tools as seamless, indistinguishable drop-in replacements for human workers.

We use metaphors all the time. To quote George Lakoff, we live by them.

We use analogies some of the time. They’re particularly useful when we’re wrapping our heads around something new. By comparing something novel to something familiar, we can make a shortcut to comprehension, or at least, categorisation.

But we need a certain amount of vigilance when it comes to analogies. Just because something is like something else doesn’t mean it’s the same.

With that in mind, here are some ways that people are describing generative machine learning tools. Large language models are like…

• …a blurry JPEG of the web,
• …fly-by-wire avionics,
• …drum machines,
• …an octopus.

I’ve mentioned before that I’m not a fan of initialisms and acronyms. They can be exclusionary.

It bothers me doubly when everyone is talking about AI.

First of all, the term is so vague as to be meaningless. Sometimes—though rarely—AI refers to general artificial intelligence. Sometimes AI refers to machine learning. Sometimes AI refers to large language models. Sometimes AI refers to a series of if/else statements. That’s quite a spectrum of meaning.

Secondly, there’s the assumption that everyone understands the abbreviation. I guess that’s generally a safe assumption, but sometimes AI could refer to something other than artificial intelligence.

In countries with plenty of pastoral agriculture, if someone works in AI, it usually means they’re going from farm to farm either extracting or injecting animal semen. AI stands for artificial insemination.

I think that abbreviation might work better for the kind of things currently described as using AI.

We were discussing this hot topic at work recently. Is AI coming for our jobs? The consensus was maybe, but only the parts of our jobs that we’re more than happy to have automated. Like summarising some some findings. Or perhaps as a kind of lorem ipsum generator. Or for just getting the ball rolling with a design direction. As Terence puts it:

Midjourney is great for a first draft. If, like me, you struggle to give shape to your ideas then it is nothing short of magic. It gets you through the first 90% of the hard work. It’s then up to you to refine things.

That’s pretty much the conclusion we came to in our discussion at Clearleft. There’s no way that we’d use this technology to generate outputs for clients, but we certainly might use it to generate inputs. It’s like how we’d do a quick round of sketching to get a bunch of different ideas out into the open. Terence is spot on when he says:

Midjourney lets me quickly be wrong in an interesting direction.

To put it another way, using a large language model could be a way of artificially injecting some seeds of ideas. Artificial insemination.

So now when I hear people talk about using AI to create images or articles, I don’t get frustrated. Instead I think, “Using artificial insemination to create images or articles? Yes, that sounds about right.”

I just finished watching The Billion Dollar Code, a German language miniseries on Netflix. It’s no Halt and Catch Fire, but it combines ’90s nostalgia, early web tech, and an opportunity for me to exercise my German comprehension.

It’s based on a true story, but with amalgamated characters. The plot, which centres around the preparation for a court case, inevitably invites comparison to The Social Network, although this time the viewpoint is from that of the underdogs trying to take on the incumbent. The incumbent is Google. The underdogs are ART+COM, artist-hackers who created the technology later used by Google Earth.

Early on, one of the characters says something about creating a one-to-one model of the whole world. That phrase struck me as familiar…

I remember being at the inaugural Future Of Web Apps conference in London back in 2006. Discussing the talks with friends afterwards, we all got a kick out of the speaker from Google, who happened to be German. His content and delivery was like a wonderfully Stranglovesque mad scientist. I’m sure I remember him saying something like “vee made a vun-to-vun model of the vurld.”

His name was Steffen Meschkat. I liveblogged the talk at the time. Turns out he was indeed on the team at ART+COM when they created Terravision, the technology later appropriated by Google (he ended up working at Google, which doesn’t make for as exciting a story as the TV show).

His 2006 talk was all about Ajax, something he was very familiar with, being on the Google Maps team. The Internet Archive even has a copy of the original audio!

It’s easy to forget now just how much hype there was around Ajax back then. It prompted me to write a book about combining Ajax and progressive enhancement.

These days, no one talks about Ajax. But that’s not because the technology went away. Quite the opposite. The technology became so ubiquituous that it no longer even needs a label.

A web developer today might ask “what’s Ajax?” in the same way that a fish might ask “what’s water?”

When I was braindumping my thoughts prompted by last week’s UX Fest conference, I wrote about dark patterns.

Well, actually I wrote about deceptive dark patterns. That was a deliberate choice.

The phrase “dark pattern” is …problematic. We really don’t need to be associating darkness with negativity any more than we already do in our language and culture.

This is something I discussed with Melissa Smith after her talk on this topic. The consensus in general seems to be that the terminology is far from ideal, but it’s a bit late to change it now (I’m sure if Harry were coining the term today, he would choose a different phrase).

The defining characteristic of a “dark” pattern is that intentionally deceptive. How about we shift the terminology to talk about deceptive patterns?

Now, I get that inertia is a powerful force and it would be confusing to try do to a find-and-replace on all the resources that already exist on documenting “dark” patterns. So here’s a compromise:

From here on out, let’s start using the adjective “deceptive” in addition to the existing adjective “dark.” That’s what I did in my blog post. I only used the phrase “deceptive dark patterns.”

If we do that consistently, then after a while we’ll be able to drop one of those adjectives—“dark”—and refer to “deceptive patterns.”

Personally I’d love it if we could change the terminology overnight—and I’m quite heartened by the speed at which we changed our Github branches from “master” to “main”—but being pragmatic, I think this approach stands a greater chance of success.

Who’s with me?

I work with words. Sometimes they’re my words. Sometimes they’re words that my colleagues have written:

One of my roles at Clearleft is “content buddy.” If anyone is writing a talk, or a blog post, or a proposal and they want an extra pair of eyes on it, I’m there to help.

I also work with web technologies, usually front-of-the-front-end stuff. HTML, CSS, and JavaScript. The technologies that users experience directly in web browsers.

I think a lot about design principles for the web. The two principles I keep coming back to are the robustness principle and the principle of least power.

When it comes to words, the guide that I return to again and again is George Orwell, specifically his short essay, Politics and the English Language.

Towards the end, he offers some rules for writing.

1. Never use a metaphor, simile, or other figure of speech which you are used to seeing in print.
2. Never use a long word where a short one will do.
3. If it is possible to cut a word out, always cut it out.
4. Never use the passive where you can use the active.
5. Never use a foreign phrase, a scientific word, or a jargon word if you can think of an everyday English equivalent.
6. Break any of these rules sooner than say anything outright barbarous.

These look a lot like design principles. Not only that, but some of them look like specific design principles. Take the robustness principle:

Be conservative in what you send, be liberal in what you accept.

That first part applies to Orwell’s third rule:

If it is possible to cut a word out, always cut it out.

Be conservative in what words you send.

Then there’s the principle of least power:

Choose the least powerful language suitable for a given purpose.

Compare that to Orwell’s second rule:

Never use a long word where a short one will do.

That could be rephrased as:

Choose the shortest word suitable for a given purpose.

Or, going in the other direction, the principle of least power could be rephrased in Orwell’s terms as:

Never use a powerful language where a simple language will do.

Oh, I like that! I like that a lot.

One of my roles at Clearleft is “content buddy.” If anyone is writing a talk, or a blog post, or a proposal and they want an extra pair of eyes on it, I’m there to help.

Sometimes a colleague will send a link to a Google Doc where they’ve written an article. I can then go through it and suggest changes. Using the “suggest” mode rather than the “edit” mode in Google Docs means that they can accept or reject each suggestion later.

But what works better—and is far more fun—is if we arrange to have a video call while we both have the Google Doc open in our browsers. That way, instead of just getting the suggestions, we can talk through the reasoning behind each one. It feels more like teaching them to fish instead of giving them a grammatically correct fish.

Some of the suggestions are very minor; punctuation, capitalisation, stuff like that. Where it gets really interesting is trying to figure out and explain why some sentence constructions feel better than others.

A fairly straightforward example is long sentences. Not all long sentences are bad, but the longer a sentence gets, the more it runs the risk of overwhelming the reader. So if there’s an opportunity to split one long sentence into two shorter sentences, I’ll usually recommend that.

Here’s an example from Chris’s post, Delivering training remotely – the same yet different. The original sentence read:

I recently had the privilege of running some training sessions on product design and research techniques with the design team at Duck Duck Go.

There’s nothing wrong with that. But maybe this is a little easier to digest:

I recently had the privilege of running some training sessions with the design team at Duck Duck Go. We covered product design and research techniques.

Perhaps this is kind of like the single responsibility principle in programming. Whereas the initial version was one sentence that conveyed two pieces of information (who the training was with and what the training covered), the final version has a separate sentence for each piece of information.

I wouldn’t take that idea too far though. Otherwise you’d end up with something quite stilted and robotic.

Speaking of sounding robotic, I’ve noticed that people sometimes avoid using contractions when they’re writing online: “there is” instead of “there’s” or “I am” instead of “I’m.” Avoiding contractions seems to be more professional, but actually it makes the writing a bit too formal. There’s a danger of sounding like a legal contract. Or a Vulcan.

Sometimes a long sentence can’t be broken down into shorter sentences. In that case, I watch out for how much cognitive load the sentence is doling out to the reader.

Here’s an example from Maite’s post, How to engage the right people when recruiting in house for research. One sentence initially read:

The relevance of the people you invite to participate in a study and the information they provide have a great impact on the quality of the insights that you get.

The verb comes quite late there. As a reader, until I get to “have a great impact”, I have to keep track of everything up to that point. Here’s a rephrased version:

The quality of the insights that you get depends on the relevance of the people you invite to participate in a study and the information they provide.

Okay, there are two changes there. First of all, the verb is now “depends on” instead of “have a great impact on.” I think that’s a bit clearer. Secondly, the verb comes sooner. Now I only have to keep track of the words up until “depends on”. After that, I can flush my memory buffer.

Here’s another changed sentence from the same article. The initial sentence read:

You will have to communicate at different times and for different reasons with your research participants.

I suggested changing that to:

You will have to communicate with your research participants at different times and for different reasons.

To be honest, I find it hard to explain why that second version flows better. I think it’s related to the idea of reducing dependencies. The subject “your research participants” is dependent on the verb “to communicate with.” So it makes more sense to keep them together instead of putting a subclause between them. The subclause can go afterwards instead: “at different times and for different reasons.”

Here’s one final example from Katie’s post, Service Designers don’t design services, we all do. One sentence initially read:

Understanding the relationships between these moments, digital and non-digital, and designing across and between these moments is key to creating a compelling user experience.

That sentence could be broken into shorter sentences, but it might lose some impact. Still, it can be rephrased so the reader doesn’t have to do as much work. As it stands, until the reader gets to “is key to creating”, they have to keep track of everything before that. It’s like the feeling of copying and pasting. If you copy something to the clipboard, you want to paste it as soon as possible. The longer you have to hold onto it, the more uncomfortable it feels.

So here’s the reworked version:

The key to creating a compelling user experience is understanding the relationships between these moments, digital and non-digital, and designing across and between these moments.

As a reader, I can digest and discard each of these pieces in turn:

1. The key to creating a compelling user experience is…
2. understanding the relationships between these moments…
3. digital and non-digital…
4. and…
5. designing across and between these moments.

Maybe I should’ve suggested “between these digital and non-digital moments” instead of “between these moments, digital and non-digital”. But then I worry that I’m intruding on the author’s style too much. With the finished sentence, it still feels like a rousing rallying cry in Katie’s voice, but slightly adjusted to flow a little easier.

I must say, I really, really enjoy being a content buddy. I know the word “editor” would be the usual descriptor, but I like how unintimidating “content buddy” sounds.

I am almost certainly a terrible content buddy to myself. Just as I ignore my own advice about preparing conference talks, I’m sure I go against my own editorial advice every time I blurt out a blog post here. But there’s one piece I’ve given to others that I try to stick to: write like you speak.

I think my co-workers are getting annoyed with me. Any time they use an acronym or initialism—either in a video call or Slack—I ask them what it stands for. I’m sure they think I’m being contrarian.

The truth is that most of the time I genuinely don’t know what the letters stand for. And I’ve got to that age where I don’t feel any inhibition about asking “stupid” questions.

But it’s also true that I really, really dislike acronyms, initialisms, and other kinds of jargon. They’re manifestations of gatekeeping. They demarcate in-groups from outsiders.

Of course if you’re in a conversation with an in-group that has the same background and context as you, then sure, you can use acronyms and initialisms with the confidence that there’s a shared understanding. But how often can you be that sure? The more likely situation—and this scales exponentially with group size—is that people have differing levels of inside knowledge and experience.

I feel sorry for anyone trying to get into the field of web performance. Not only are there complex browser behaviours to understand, there’s also a veritable alphabet soup of initialisms to memorise. Here’s a really good post on web performance by Harry, but notice how the initialisms multiply like tribbles as the post progresses until we’re talking about using CWV metrics like LCP, FID, and CLS—alongside TTFB and SI—to look at PLPs, PDPs, and SRPs. And fair play to Harry; he expands each initialism the first time he introduces it.

But are we really saving any time by saying FID instead of first input delay? I suspect that the only reason why the word “cumulative” precedes “layout shift” is just to make it into the three-letter initialism CLS.

Still, I get why initialisms run rampant in technical discussions. You can be sure that most discussions of particle physics would be incomprehensible to outsiders, not necessarily because of the concepts, but because of the terminology.

Again, if you’re certain that you’re speaking to peers, that’s fine. But if you’re trying to communicate even a little more widely, then initialisms and abbreviations are obstacles to overcome. And once you’re in the habit of using the short forms, it gets harder and harder to apply context-shifting in your language. So the safest habit to form is to generally avoid using acronyms and initialisms.

Unnecessary initialisms are exclusionary.

Think about on-boarding someone new to your organisation. They’ve already got a lot to wrap their heads around without making them figure out what a TAM is. That’s a real example from Clearleft. We have a regular Thursday afternoon meeting. I call it the Thursday afternoon meeting. Other people …don’t.

I’m trying—as gently as possible—to ensure we’re not being exclusionary in our language. My co-workers indulge me, even it’s just to shut me up.

But here’s the thing. I remember many years back when a job ad went out on the Clearleft website that included the phrase “culture fit”. I winced and explained why I thought that was a really bad phrase to use—one that is used as code for “more people like us”. At the time my concerns were met with eye-rolls and chuckles. Now, as knowledge about diversity and inclusion has become more widespread, everyone understands that using a phrase like “culture fit” can be exclusionary.

But when I ask people to expand their acronyms and initialisms today, I get the same kind of chuckles. My aversion to abbreviations is an eccentric foible to be tolerated.

But this isn’t about me.

There are three ways—that I know of—to associate styles with markup.

External CSS

This is probably the most common. Using a link element with a rel value of “stylesheet”, you point to a URL using the href attribute. That URL is a style sheet that is applied to the current document (“the relationship of the linked resource it is that is a ‘stylesheet’ for the current document”).

<link rel="stylesheet" href="/path/to/styles.css">

In theory you could associate a style sheet with a document using an HTTP header, but I don’t think many browsers support this in practice.

You can also pull in external style sheets using the @import declaration in CSS itself, as long as the @import rule is declared at the start, before any other styles.

@import url('/path/to/more-styles.css');

When you use link rel="stylesheet" to apply styles, it’s a blocking request: the browser will fetch the style sheet before rendering the HTML. It needs to know how the HTML elements will be painted to the screen so there’s no point rendering the HTML until the CSS is parsed.

Embedded CSS

You can also place CSS rules inside a style element directly in the document. This is usually in the head of the document.

<style>
element {
    property: value;
}
</style>

When you embed CSS in the head of a document like this, there is no network request like there would be with external style sheets so there’s no render-blocking behaviour.

You can put any CSS inside the style element, which means that you could use embedded CSS to load external CSS using an @import statement (as long as that @import statement appears right at the start).

<style>
@import url('/path/to/more-styles.css');
element {
    property: value;
}
</style>

But then you’re back to having a network request.

Inline CSS

Using the style attribute you can apply CSS rules directly to an element. This is a universal attribute. It can be used on any HTML element. That doesn’t necessarily mean that the styles will work, but your markup is never invalidated by the presence of the style attribute.

<element style="property: value">
</element>

Whereas external CSS and embedded CSS don’t have any effect on specificity, inline styles are positively radioactive with specificity. Any styles applied this way are almost certain to over-ride any external or embedded styles.

You can also apply styles using JavaScript and the Document Object Model.

element.style.property = 'value';

Using the DOM style object this way is equivalent to inline styles. The radioactive specificity applies here too.

Style declarations specified in external style sheets or embedded CSS follow the rules of the cascade. Values can be over-ridden depending on the order they appear in. Combined with the separate-but-related rules for specificity, this can be very powerful. But if you don’t understand how the cascade and specificity work then the results can be unexpected, leading to frustration. In that situation, inline styles look very appealing—there’s no cascade and everything has equal specificity. But using inline styles means foregoing a lot of power—you’d be ditching the C in CSS.

A common technique for web performance is to favour embedded CSS over external CSS in order to avoid the extra network request (at least for the first visit—there are clever techniques for caching an external style sheet once the HTML has already loaded). This is commonly referred to as inlining your CSS. But really it should be called embedding your CSS.

This language mix-up is not a hill I’m going to die on (that hill would be referring to blog posts as blogs) but I thought it was worth pointing out.

The latest episode of the Clearleft podcast is out. If you’re a subscriber, it will magically appear in your podcast software of choice using the power of RSS. If you’re not a subscriber, it isn’t too late to change that.

This week’s episode is all about design ops. I began contructing the episode by gathering raw material from talks at Leading Design. There’s good stuff from Kim Fellman, Jacqui Frey, Courtney Kaplan, and Meredith Black.

But as I was putting the snippets together, I felt like the episode was missing something. It needed a bit of oomph. So I harangued Andy for some of his time. I wasn’t just fishing for spicy hot takes—something that Andy is known for. Andy is also the right person to explain design ops. If you search for that term, one of the first results you’ll get is a post he wrote on the Clearleft blog a few years back called Design Ops — A New Discipline.

I remember helping Andy edit that post and I distinctly recall my feedback. The initial post didn’t have a definition of the term, and I felt that a definition was necessary (and Andy added one to the post).

My cluenessness about the meaning of terms like “design ops” or “service design” isn’t some schtick I’m putting on for the benefit of the podcast. I’m genuinely trying to understand these terms better. I don’t like the feeling of hearing a term being used a lot without a clear understanding of what that term means. All too often my understanding feels more like “I think I know it means, but I’m not sure I could describe it.” I’m not comfortable with that.

Making podcast episodes on these topics—which are outside my comfort zone—has been really helpful. I now feel like I have a much better understanding of service design, design ops and other topical terms. I hope that the podcast will be just as helpful for listeners who feel as bamboozled as I do.

Ben Holliday recently said:

The secret of design being useful in many places is not talking about design too much and just getting on with it. I sometimes think we create significant language barriers with job titles, design theory and making people learn a new language for the privilege of working with us.

I think there’s some truth to that. Andy disagrees. Strongly.

In our chat, Andy and I had what politicians would describe as “a robust discussion.” I certainly got some great material for the podcast (though some of the more contentious bits remain on the cutting room floor).

Calling on Andy for this episode was definitely the right call. I definitely got the added oomph I was looking for. In fact, this ended up being one of my favourite episodes.

There’s a lot of snappy editing, all in service of crafting a compelling narrative. First, there’s the origin story of design ops. Then there’s an explanation of what it entails. From around the 13 minute mark, there’s a pivot—via design systems—into asking whether introducing a new term is exclusionary. That’s when the sparks start to fly. Finally, I pull it back to talking about how Clearleft can help in providing design ops as a service.

The whole episode comes out at 21 minutes, which feels just right to me.

I’m really pleased with how this episode turned out, and I hope you’ll like it too. Have a listen and decide for yourself.

I compared design systems to dictionaries. My point was that design systems—like language—can be approached in a prescriptivist or descriptivist manner. And I favour descriptivism.

A prescriptive approach might give you a beautiful design system, but if it doesn’t reflect the actual product, it’s fiction. A descriptive approach might give a design system with imperfections and annoying flaws, but at least it will be accurate.

I think it’s more important for a design system to be accurate than beautiful.

Meanwhile, over on Frank’s website, he’s been documenting the process of its (re)design. He made an interesting comparison in his post Redesign: Gardening vs. Architecture. He talks about two styles of writing:

In interviews, Martin has compared himself to a gardener—forgoing detailed outlines and overly planned plot points to favor ideas and opportunities that spring up in the writing process. You see what grows as you write, then tend to it, nurture it. Each tendrilly digression may turn into the next big branch of your story. This feels right: good things grow, and an important quality of growth is that the significant moments are often unanticipated.

On the other side of writing is who I’ll call “the architect”—one who writes detailed outlines for plots and believes in the necessity of overt structure. It puts stock in planning and foresight. Architectural writing favors divisions and subdivisions, then subdivisions of the subdivisions. It depends on people’s ability to move forward by breaking big things down into smaller things with increasing detail.

It’s not just me, right? It all sounds very design systemsy, doesn’t it?

This is a false dichotomy, of course, but everyone favors one mode of working over the other. It’s a matter of personality, from what I can tell.

Replace “personality” with “company culture” and I think you’ve got an interesting analysis of the two different approaches to design systems. Descriptivist gardening and prescriptivist architecture.

Frank also says something that I think resonates with the evergreen debate about whether design systems stifle creativity:

It can be hard to stay interested if it feels like you’re painting by numbers, even if they are your own numbers.

I think Frank’s comparison—gardeners and architects—also speaks to something bigger than design systems…

I gave a talk last year called Building. You can watch it, listen to it, or read the transcript if you like. The talk is about language (sort of). There’s nothing about prescriptivism or descriptivism in there, but there’s lots about metaphors. I dive into the metaphors we use to describe our work and ourselves: builders, engineers, and architects.

It’s rare to find job titles like software gardener, or information librarian (even though they would be just as valid as other terms we’ve made up like software engineer or information architect). Outside of the context of open source projects, we don’t talk much about maintenance. We’re much more likely to talk about making.

Back in 2015, Debbie Chachra wrote a piece in the Atlantic Monthly called Why I Am Not a Maker:

When tech culture only celebrates creation, it risks ignoring those who teach, criticize, and take care of others.

Anyone who’s spent any time working on design systems can tell you there’s no shortage of enthusiasm for architecture and making—“let’s build a library of components!”

There’s less enthusiasm for gardening, care, communication and maintenance. But that’s where the really important work happens.

In her article, Debbie cites Ethan’s touchstone:

In her book The Real World of Technology, the metallurgist Ursula Franklin contrasts prescriptive technologies, where many individuals produce components of the whole (think about Adam Smith’s pin factory), with holistic technologies, where the creator controls and understands the process from start to finish.

(Emphasis mine.)

In that light, design systems take their place in a long history of dehumanising approaches to manufacturing like Taylorism. The priorities of “scientific management” are the same as those of design systems—increasing efficiency and enforcing consistency.

Humans aren’t always great at efficiency and consistency, but machines are. Automation increases efficiency and consistency, sacrificing messy humanity along the way:

Machine with the strength of a hundred men
Can’t feed and clothe my children.

Historically, we’ve seen automation in terms of physical labour—dock workers, factory workers, truck drivers. As far as I know, none of those workers participated in the creation of their mechanical successors. But when it comes to our work on the web, we’re positively eager to create the systems to make us redundant.

The usual response to this is the one given to other examples of automation: you’ll be free to spend your time in a more meaningful way. With a design system in place, you’ll be freed from the drudgery of manual labour. Instead, you can spend your time doing more important work …like maintaining the design system.

You’ve heard the joke about the factory of the future, right? The factory of the future will have just two living things in it: one worker and one dog. The worker is there to feed the dog. The dog is there to bite the worker if he touches anything.

Good joke.

Everybody laugh.

Roll on snare drum.

Curtains.

Years ago, the world of web standards was split. Two groups—the W3C and the WHATWG—were working on the next iteration of HTML. They had different ideas about the nature of standardisation.

Broadly speaking, the W3C followed a specification-first approach. Figure out what should be implemented first and foremost. From this perspective, specs can be seen as blueprints for browsers to work from.

The WHATWG, by contrast, were implementation led. The way they saw it, there was no point specifying something if browsers weren’t going to implement it. Instead, specs are there to document existing behaviour in browsers.

I’m over-generalising somewhat in my descriptions there, but the point is that there was an ideological difference of opinion around what standards bodies should do.

This always reminded me of a similar ideological conflict when it comes to language usage.

Language prescriptivists attempt to define rules about what’s right or right or wrong in a language. Rules like “never end a sentence with a preposition.” Prescriptivists are generally fighting a losing battle and spend most of their time bemoaning the decline of their language because people aren’t following the rules.

Language descriptivists work the exact opposite way. They see their job as documenting existing language usage instead of defining it. Lexicographers—like Merriam-Webster or the Oxford English Dictionary—receive complaints from angry prescriptivists when dictionaries document usage like “literally” meaning “figuratively”.

Dictionaries are descriptive, not prescriptive.

I’ve seen the prescriptive/descriptive divide somewhere else too. I’ve seen it in the world of design systems.

Jordan Moore talks about intentional and emergent design systems:

There appears to be two competing approaches in designing design systems.

An intentional design system. The flavour and framework may vary, but the approach generally consists of: design system first → design/build solutions.

An emergent design system. This approach is much closer to the user needs end of the scale by beginning with creative solutions before deriving patterns and systems (i.e the system emerges from real, coded scenarios).

An intentional design system is prescriptive. An emergent design system is descriptive.

I think we can learn from the worlds of web standards and dictionaries here. A prescriptive approach might give you a beautiful design system, but if it doesn’t reflect the actual product, it’s fiction. A descriptive approach might give a design system with imperfections and annoying flaws, but at least it will be accurate.

I think it’s more important for a design system to be accurate than beautiful.

As Matthew Ström says, you should start with the design system you already have:

Instead of drawing a whole new set of components, start with the components you already have in production. Document them meticulously. Create a single source of truth for design, warts and all.

The latest episode of Ariel’s excellent Offworld video series (and podcast) is all about Close Encounters Of The Third Kind.

I have such fondness for this film. It’s one of those films that I love to watch on a Sunday afternoon (though that’s true of so many Spielberg films—Jaws, Raiders Of The Lost Ark, E.T.). I remember seeing it in the cinema—this would’ve been the special edition re-release—and feeling the seat under me quake with the rumbling of the musical exchange during the film’s climax.

Ariel invited Rose Eveleth and Laura Welcher on to discuss the film. They spent a lot of time discussing the depiction of first contact communication—Arrival being the other landmark film on this topic.

This is a timely discussion. There’s a new book by Daniel Oberhaus published by MIT Press called Extraterrestrial Languages:

If we send a message into space, will extraterrestrial beings receive it? Will they understand?

You can a read an article by the author on The Guardian, where he mentions some of the wilder ideas about transmitting signals to aliens:

Minsky, widely regarded as the father of AI, suggested it would be best to send a cat as our extraterrestrial delegate.

Don’t worry. Marvin Minsky wasn’t talking about sending a real live cat. Rather, we transmit instructions for building a computer and then we can transmit information as software. Software about, say, cats.

It’s not that far removed from what happened with the Voyager golden record, although that relied on analogue technology—the phonograph—and sent the message pre-compiled on hardware; a much slower transmission rate than radio.

But it’s interesting to me that Minsky specifically mentioned cats. There’s another long-term communication puzzle that has a cat connection.

The Yukka Mountain nuclear waste repository is supposed to store nuclear waste for 10,000 years. How do we warn our descendants to stay away? We can’t use language. We probably can’t even use symbols; they’re too culturally specific. A think tank called the Human Interference Task Force was convened to agree on the message to be conveyed:

This place is a message… and part of a system of messages… pay attention to it! Sending this message was important to us. We considered ourselves to be a powerful culture.

This place is not a place of honor…no highly esteemed deed is commemorated here… nothing valued is here.

What is here is dangerous and repulsive to us. This message is a warning about danger.

A series of thorn-like threatening earthworks was deemed the most feasible solution. But there was another proposal that took a two pronged approach with genetics and folklore:

1. Breed cats that change colour in the presence of radioactive material.
2. Teach children nursery rhymes about staying away from cats that change colour.

This is the raycat solution.