How to Write Better with LLMs (Part 2)

chopinhauer

As mentioned before, there’s a series shaping up in our minds about Large Language Models (LLMs) and writing! We don’t pretend to have the answers for how writers should adjust to the era of LLMs. The answer probably won’t be clear until historians start studying us with 20-50 years’ worth of hindsight. But that’s no reason not to think through our moment of upheaval. So here are some thoughts that we’re likely to keep ruminating on.

Mingling Cognition and ⌜Cognition⌝

LLMs as shiny, abstract, weird ⌜intelligence⌝. Photo by Vinicius “amnx” Amano on Unsplash.

It’s important to remember that writing has always been commingled with technology. Strictly speaking, writing is technology: the recording of words mediated by tools: engraving tools, pens, pencils, parchment, paper, printing presses, computers, keyboards, and so on. History teaches us that there is nothing truly novel if you know enough historical parallels. Every time new writing-related technology emerged and caught on, it raised anxieties about the value of human thought. Famously, Socrates didn’t write anything down himself because he held that relying on written records undermined our memory and ability to reason. He wasn’t entirely wrong, but at the social level we have always adapted to technological upheavals—eventually.

The advent of LLMs fits this narrative quite well. LLM ⌜cognition⌝ can automate some low-level parts of the writing process, freeing up our cognition to focus on more high-level features of the writing. The promise of this technology is to keep us in the flow of ideas and high-level movement in writing, and to automate more repetitive and tedious parts of the process.

Of course, LLM ⌜cognition⌝ is a strange new technology, one that invites attributions of intelligence, although we think that’s quite premature. (Hence the corner quotes; they remind us that the LLM ⌜thinks⌝ differently than we do, and that we’re using these words in ways that stretch their natural meanings.) It will take time to sort out the capabilities and the pros and cons of LLM capabilities.

The endless AI hype train is not helping matters with its grandiose claims. Nuance tends to get buried in such hyperbolic discourse. LLMs have been shown to save a ton of time for some writing tasks, but they’ve also been found to be quite flat-footed in other tasks, to the point of needing very, very refined prompting—at which point, you might as well write the piece yourself.

The Once and Future Writer-Editor

We think the likeliest consequence of widespread LLM adoption is that the writing process will end up looking more like editing. Making piles of text on a topic will become almost trivial, so writers will naturally use more of their time in assessing the outputs of their LLMs. This means more carefully crafting their prompts, and wondering about whether to take the prompting in completely new directions. This is how human cognition will retain its leading role in its dance with LLM ⌜cognition⌝.

For example, you prompt ChatGPT with an idea you’ve had. It returns competent, somewhat generic, reasonably on-topic text. You have options. You can redo the prompt. Or you ask it to write in a different style, or to role-play as a particular writer. Perhaps you ask it to write the text in a more punchy, concise, direct, energetic, or conversational way. Or perhaps you ask it to cut paragraph 3 and expand paragraph 4. Maybe you fiddle with low-level details of the sentences yourself.

This is what editors do to their writers! Our suggestions are nudges for breaking writers out of sub-optimal habits at many different levels in writing. For some writing tasks, whether we’re nudging cognition or ⌜cognition⌝ may not matter that much.

The Future of Writing Skill

The skills of high-level discernment regarding LLMs will be in demand. Just as today “writing skill” means something like skill for effective language use, rather than excellent penmanship, in the medium-to-long term writing skill will become closer to arranging and drafting skill, rather than typing skill.

History can teach us here as well. In Ancient Greece and Rome, elite writers were trained in high-level features of how to compose text. They learned how to make writing compelling, logical, and persuasive, rather than just how to make sentences. This training emphasized writing as persuasion, and not merely the transcription of records—that’s what scribes did. Writers were actually writer-editors. So, as LLMs will take on more scribe-like roles, writers should occupy the more editor-like roles.

We hope for everyone’s sake that the transition is smooth, but only time will tell.

Structural Issues and Academic Work

admin

Here in Canada, May Day is a relatively neglectezd holiday because the labour movement is in a historic rut. But we think it’s important to recognize its contribution to improving working conditions for all sorts of people, including us.

It might seem strange that we, a small business, are saluting the labour movement. But this business wouldn’t exist without labour organizing at the University of Toronto. We were fortunate to participate in a protracted, month-long strike, and we experienced the difference it made. Our union participation is what gave us insight into the conditions of academic work beyond our own individual circumstances; that experience empowered us to offer our services.

So today, in honour of May Day, let’s focus on structural issues facing academics. The solution to all structural issues is simple but hard: get organized and demand better.

Structural Issues

Structural issues are complicated but intelligible. They are the big factors operating above the level of the individual that constrain us all. Difficult working conditions are one kind of structural issue; there are many others: repressive laws, uneven labour markets, poverty, racism, sexism, and many others. Individuals respond to structural issues by resisting, by moving, by internalizing the system, by grifting, or by organizing. Only organized resistance can address the structural issue at scale. Another term for structural issues is “collective action problems”, which nicely foregrounds what the solution is—collective action.

Our culture tends to place too much emphasis on individual factors. For example, difficulty at work tends to be discussed as laziness, lack of discipline, an attitude problem, etc. There’s something to this way of thinking, since structural issues never determine what an individual can do. But our experience as coaches has convinced us that most of those discussions aren’t very helpful.

An instance of academic structure! Photo by Jaeyoung Geoffrey Kang on Unsplash

Many people think that discussion of structural issues is somehow obscure, but this is usually a bad-faith argument. We all live in a society, having to navigate systems that constrain us; when the constraints are especially bad, that’s a structural issue. For example, complaining about traffic is complaining about a structural issue; traffic can be bad even if all the drivers you encounter are responsible and responsive.

Academic Structural Issues

The fundamental structural issue facing academia is overproduction of academics. There are simply not enough good positions for everyone to remain in academia and make a life of it. Because of this, academics usually overwork, overproduce, and compete with each other for scarce positions. This results in chronically chaotic, uncertain, and anxiety-fuelled working conditions.

There are also wider social structures around academia: the devaluation of liberal arts in favour of STEM, the narratives that academics are solitary figures, the myths around genius, the drive towards hyper-productivity, and all sorts of self-serving discipline-specific narratives.

As we’ve discussed before, deciding to be an academic, or deciding to stick it out in your degree, means accepting some of these structural issues as givens and working around them. But no sensitive human being should become fully adjusted to the system as it stands.

Interventions: Individual and Structural

Many academics learn to survive in the system, and some thrive. This process of learning usually depends on extensive mentoring support from other academics and the individual’s community. We consider our work to be part of this service. We work at both the individual and systemic level, as appropriate. Often, just naming and sketching the structural issue helps normalize the struggle and neutralizes it as fodder for self-blame. Knowing that you’re not alone in the struggle, and that the struggle is not of your own making, is usually very effective for getting academics to be a bit less anxious.

But individual and community support can only go so far. There have to also be periods of large-scale, organized pushback to the direction of the system, lest it fall apart completely. This is where associations and labour unions come in. There seems to be a general momentum in this direction in the past few years in academia, although it’s hard to make predictions. Academics are increasingly organizing and recognizing the structural issues that have beset their working conditions for decades. The only thing that has succeeded in making change is large groups of organized people. May workers of all sorts—academics included—get organized and demand what’s fair!

Attention, Writing, and Editing (Part 2): Nudging Attention into the Right Shape

admin

We’ve been thinking about attention, writing, and editing for quite a while now, either explicitly, or implicitly in a few places. Understandably, we’ve been pretty writing-focused. Today we’re going to focus on editing because this is where we’ve optimized our attention.

Understanding how attention works—and how it doesn’t—is the difference between good and bad editing. It’s also the difference between sustainable and exhausting editing. Much of what we’ll say applies to writing, but the writing process has its own motivational quirks on top of attention’s inherent quirks. We’ll get to that in future posts.

Attention is Complex

Psychology and cognitive science tells us a few very useful things about attention. First, it’s a limited resource. In attention research, there’s been a decades-long trend of finding new limitations. For example, our tracking abilities are actually quite limited, and our capacity for inattention is more extensive than previously thought. Second, the research shows that attention is a complex, layered structure. Simplistic metaphors of “shining a light” don’t capture its dynamics. If you need a metaphor, attention is more like musical performance, dancing, or martial arts than shooting an arrow at a target or driving from point A to point B. These metaphors imply that training attention is a combination of hard work, aptitude, and a bit of good luck.

This is the state of the art as it stands in research. If we want to write better, we need to take this research under advisement, but also not feel constrained by it. It turns out that attention is a complex habit that responds to how we make narrative sense of our lives.

Attention needs to be responsive to potentially any nuances of a situation. As with so many things, it’s not under full conscious control. So we have to think about attention a complex sedimentation of habits. These habits relate to how we shift between transparency and opacity, forest and trees, and figure and ground in any given situation, as we covered here. Attention is one of the most mercurial (slippery) things in our inner lives. It comes and goes like the wind, as every writer can attest. The moments of excellent performance arrive as if by accident most of the time.

Forest or trees? Photo by Arnaud Mesureur on Unsplash

Attention Responds to Nudges

It’s been imperative for us to get good at managing our attention. We’ve learned—from our experience, reflection, and the scientific literature—that attention is pretty chaotic, but that it responds to patient, persistent nudging.

That’s the bad news. The mercurial nature of attention means that nudges typically don’t work the first, third, or even tenth time. The good news is that with patient persistence (which can be very difficult), attention will eventually settle into the proper shape for the task, and once settled can sustain this shape for quite a while. This is, as we’ve already discussed, the starting point for deep work, flow, and all the best fruits of cognitive labour.

There’s an art to nudging attention. The nudge can’t be too forceful, otherwise it will backfire. Likewise, the nudge can’t be too lax, since attention will just keep whatever shape it already had. Training a puppy to sit still is an excellent analogy: you shouldn’t punish the puppy for failing to sit still, nor can you let the puppy do whatever it wants. You nudge the puppy’s behaviour with small rewards.

Practical Nudges for Editing

So, what’s the equivalent of a dog treat for our attention? Here are some nudges that we’ve found helpful.

There are some general things you should do to make sure you have basic energy to edit. First, divide writing and editing time into different blocks if possible. Attention takes slightly different shapes in writing and editing, and we should respect the differences. In general, you can do editing after the writing block. Second, the more sleep cycles you can work into an editing project, the better you’ll do. It’s far, far better to do two half-hour blocks on different days than one hour-long block.

There are also more subtle nudges within a block. First, if you’re copyediting, go through your paragraphs in reverse order. Avoiding the usual flow of the writing is a nudge from transparency (attending to what the words say) to opacity (attending to the words as markings on a page/screen). You will catch more errors this way. Second, for different types of editing you should approach the page/screen at different distances. The closer you are to the text, the easier it is to shift from transparency to opacity. Moving towards the text also shifts your focus from the proverbial forest to the proverbial trees.

If you’re doing structural or developmental editing, make the text smaller. If you’re doing copyediting or proofreading, make it bigger. For example, we sometimes do 150% zoom in Word for copyediting, 175% for dealing with more fiddly details such as references or punctuation, and 100% for dealing with paragraphs or developmental suggestions. Spatial cues like these are excellent nudges. Find what works for you!

How to Write Better with LLMs

chopinhauer

Despite the prevalence of hype, large language models (LLMs) are genuinely impressive. They performa as well as humans for some genres of writing. Given enough context and well-conceived prompting, they can measure up for more complex writing tasks. For spitting out grammatically impeccable text quickly, they’ve already left us humans in the dust.

We are considering developing a workshop on writing in the age of LLMs. Our goal would be to help with: (1) worrying about LLMS more efficiently, and (2) making good use of what the technology affords. It’s a big project. If you’d like to hear about it once it’s done, drop us a line. We’re also open to collaboration, since LLM technology affects all of us who make or improve words for a living.

In the meantime, here are some early thoughts in this direction.

How not to Give your Writing Away

We think the deep danger in using LLMs is succumbing to the temptation to give our writing away to the LLM.

This idea is drawn from academic writing, where authors sometimes make arguments that lean too heavily on the existing literature. These papers tend to read as a bit all over the place, tending to follow the breadcrumb trails of someone else’s thinking. The paper might be well-sourced and cite a great deal of literature, but it somehow doesn’t add up to an argument. The paper feels more like reporting others’ work rather than contributing.

We all do this from time to time, especially in early drafts. This tendency comes from an overall lack of confidence, unclear goals, or being overwhelmed. When we are not confident, we cling to anything that seems to have that confidence, which is usually the sources we’re reading to get up to speed. Similarly, when overwhelmed we tend to fixate on details (like citations or things we’ve read) as a way of finding some order in the chaos.

If we as writers (1) are feeling confident, (2) have a clear plan for what we want to say, and (3) are not overwhelmed, we’ll probably not give our writing away to the LLM. However, if any of (1)-(3) are missing, we need to check ourselves, otherwise we’re likely to mistake an LLM’s natural writing confidence for something more than it is. We are in no position to lead in the dance if we’re compromised.

LLM Use will Depend on Delegation Skills

Even if we’re in a good headspace, LLMs might lead us astray if we don’t know how to delegate work.

If we think of LLMs as writing assistants, we can import all the wisdom about using assistants properly. Good assistant use is good delegation. We need to have some sense of the assistant’s capabilities. We need to have a clear vision of the overall goal which the assistant might not need to have. Given that clear vision, we can break the goal down into smaller tasks, some of which we can delegate to the assistant. To the extent that we’re good delegators, we’re likely to find LLMs useful supporting characters.

DeepMind’s visualization of LLMs.

Bad delegation introduces errors and confusion where there were none. And what do we expect when we give a big task to an assistant who’s not ready for it? We need to remember that LLMs, despite their polished presentation, are not all-knowing and all-competent. Just like some humans.

The Writer as AI Manager

Depending on the writing task, one might use several different AI assistants. We think that in the near-to-medium future writing will become a little more managerial under the influence of LLMs. We shudder at this possibility, but this seems to be the trend for the next 5-20 years.

To prepare for this, we all need to attend more deeply to the high-level structural aspects of writing. The writer of the future might manage a team of AI copyeditors, proofreaders, layout editors, and even developmental editors as part of their toolkit. In this context, the writer’s deepest responsibility will be, as always, clarity of overall vision. In a way, not much will change about why we write, but the methods of writing 20 years from now might look very different.

The ABD Morass, Academic Coaching, and Developmental Editing

chopinhauer

We started this business because during grad school we got an unusually close look at the struggles of hundreds of grad students at Canada’s largest university. We chose to specialize in academic editing and coaching because of own experience, that of our friends and colleagues, and the patterns we noticed that seemed near-universal for all sorts of people. One of our basic missions is to make the grad school journey a little easier for those who come after us.

So let’s talk about a universal feature of grad school: the ABD morass.

The ABD Morass

ABD (All But Dissertation) is an unofficial academic rank. It’s the stage where the PhD candidate has completed all other program requirements. These may include course work, a qualifying exam (or two!), a dissertation proposal, a language exam, and other program-specific demands. In the ABD stage there’s nothing left to do but write the dissertation! It’s where the PhD candidate finally gets to enjoy some well-earned freedom to do their deep research and contribute to their academic discipline.

In theory, the candidate has all they need to succeed. In practice, however, this is the stage where graduate students feel the most unprepared. They feel unprepared because the task is so big and ill-defined. A dissertation is, in essence, a book-length work that (1) makes a contribution to the academic field, and (2) demonstrates promise of further contributions. The problem is, there’s usually nothing in a grad student’s experience that helps them break this project down into manageable pieces. It’s the first experience of having to forge one’s own path. And it can be paralyzing.

Typically, most grad students spend the first year of the ABD stage floundering. They still work hard, write words, read things, run experiments, go to conferences, meet with their advisor, and so on, but many have the sense that all this activity isn’t adding up to anything. Many academics, even very successful ones, look back on this time with some dismay and shame.

“Ready to work, but lacking structure” should be the slogan of the ABD morass.

ABD Skills are New Skills

There is, of course, no need for shame. The ABD stage is a major transition in a grad student’s approach to academic work. Before the ABD stage, finely-honed student skills from undergraduate education are relevant and useful. In the ABD stage, the expectations shift to academic productivity and the size of the projects opens up a skills gap. The skills of writing a book do not come easily, or quickly, to anyone. It’s a hard transition because the task of writing a dissertation just is hard.

Ideally, dissertation committees exist to guide the grad student through this stage. But even well-meaning, supportive, and capable committees are often unable to provide intensive support to their grad students at this critical juncture. Academics are busy, and they never stop being busy. As a result, the ABD stage is almost universally one where grad students feel adrift and isolated. The struggle feels intensely personal, but it helps to realize that much of the difficulty is strangely universal and impersonal.

Developmental Editing and Academic Coaching

We offer developmental editing and academic coaching as a ways to make the ABD stage less difficult. We approach editing and academic coaching in an integrated way because in the ABD morass the distinctions between writing, pre-writing, discipline, and planning get pretty fuzzy.

Developmental editing is where we offer feedback on the high-level structure of a work in progress. Here, we focus on the overall argument and offer specific paths to a cleaner argument. Working iteratively and intensively, sometimes ad nauseam, is the best way to develop the ideas and structure that will sustain a large-scale research project.

Academic coaching is where we develop and refine our clients’ general approach to the problem. Here we offer empathy, normalization, perspective (if appropriate), and help work out SMART goals as a way to avoid the overwhelm of the ABD stage. Coaching can happen either prior to or alongside developmental editing, depending on context.

Through the Dark Wood of the ABD Phase

Getting some momentum in the ABD phase with this combination of editing and coaching also helps with grad students’ relationship with their committees. All too often, the lack of progress leads into a downward spiral of avoidance, negative self-talk, and stress that feeds back into the lack of productivity that started the spiral. Finding some structure, even if imperfect, gives both the grad student and their committee something to work with and focus on.

Our services help because nobody is an expert at changing their work habits in a pinch, and it’s very hard to both do that and form a groundbreaking academic work. We are here to support the struggle towards new heights of efficacy in forging a way out of the morass.

Announcing ClarityGPT: Fully Automated Full-Stack Academic Editing and Coaching

chopinhauer

We’re thrilled and proud to announce that today we are launching ClarityGPT (cGPT), a large language model that automates what we do as editors and academic coaches at all levels. ClarityGPT is trained on all our editing work over the past 12 years, as well as video and audio data from meetings with hundreds of grad students. Sign up today! Through the ClarityGPT API, you will get:

Image shows a robot sitting on a bench with a group of roofing tiles laid out flat behind them to the left. The robot is typing on a keyboard. It's a surreal image.
The LLM will see you now. Photo by Andrea De Santis on Unsplash
  • Near-instant structural feedback on your work
  • Near-instant copyediting of your work at all levels, in compliance with all relevant style guides—just select from a drop-down menu!
  • “In-person” support over Zoom with our deepfake avatars that will offer basic emotional support and the occasional anecdote or wry witticism about grad school

With cGPT, you benefit directly:

  • Rush jobs are just jobs now
  • Write 10 papers per week!
  • Arrange a meeting whenever you want—time zones are just times now
  • You can use cGPT to train your supervisor (conditions apply)
  • No more fiddly human errors in writing. Only pervasive algorithmic biases from now on.

We are looking forward to sitting back, fiddling with high-level cGPT parameters, and developing our long-term passion project, ClarityCoin,* which will use blockchain technology to optimize editing and payment, somehow.

Subscribe now! Get in on the ground floor! Special rate expires today at noon. Contact us here.

Writerly Akrasia

chopinhauer

We have written about writers’ difficulties in quite a few places because our clients often face these challenges. This is an enormous topic, spanning culture, psychology, theories of education, social science, disability studies, politics, and more! Perhaps one day we’ll write a book! But for now we’ll keep exploring writing difficulties from different angles. Today we’ll try to drill down to the root of the many ways we get stuck in writing.

The root is simple: all of us, regardless of skill level or dedication, are less disciplined than we’d like to be. And this is one of the most human things about us. We need to be kind with our inability, because we’re up against a fundamental limitation of all sensitive, intelligent, ambitious creatures: akrasia.

Akrasia

Akrasia is a term from ancient Greek philosophy that refers to an internal conflict where our better judgment settles on a course of action, but we fail to follow through. In other words, akrasia involves both (1) knowing what you need to do, and (2) doing something else (or nothing at all). We think “akrasia” is more precise than “anxiety,” and precision is essential for breaking below the apparent but unhelpful way this stuff appears to us. That said, a lot of akrasia manifests as anxiety.

We all experience mild forms of akrasia: eating that extra piece of cake, dipping into social media when we should be studying, or making that dubious purchase. Note that we are all capable of pulling ourselves out of akrasia in many situations. Akrasia is a feature of life and need not be the end of the world.

Writerly akrasia is stronger: it’s what we might call an akrasia storm. This is when we get stuck, which triggers various feelings that d undermine our focus, which gets us more stuck, and tired, and the whole process snowballs, taking on a life of its own.

At the extreme end, we are in the landscape of anxiety disorders of all sorts, where akrasia storms persist beyond specific situations, or strike so hard as to be debilitating, rather than merely difficult. If this is your situation, we recommend seeing a therapist, if possible. They can help immensely with management of what seem like insurmountable inner difficulties.

The Causes of Akrasia Storms

As with actual storms, many factors typically come together to produce akrasia storms. Many different stressors can contribute. What might be trivial to handle for one person might just steamroll another person.

Sometimes it really is like that!

Here are some common stressors that contribute to writerly akrasia storms: facing a deadline, being overcommitted, receiving harsh comments, being told to massively revise or rewrite something we’re attached to, interpersonal turmoil, insufficient sleep, and the infinite scroll of social media. Academics in particular face a few special ones: writing for a dissertation or tenure committee, imagining committee feedback, and office politics.

Whether a stressor is important depends on general tendencies to self-doubt, self-flagellation, self-criticism, time pressure, general level of ambition, personal standards and, of course, the stakes of a specific project. A lot of this—stressors, tendencies, and stakes—are context-sensitive.

This is not an exhaustive list, of course, but it tracks important themes of our experience as writers and coaches. The key point is: akrasia storms are complex knots of many causes that feed off each other.

How to Unravel an Akrasia Storm

if you’re stuck in an akrasia storm, you have to do the most counterintuitive thing in the world: patiently, kindly unravel the knot so you can get back to work. You need to get clear and discerning about what’s going on below the surface. Akrasia storms present as a generalized “ugh!” feeling and the sudden need to do anything else. But the only way through is to find a thread to loosen, and then to carefully loosen it. Finding the thread takes discernment; loosening the thread takes kindness. Neither is especially available when we’re stuck.

Yet, we can move toward kindness even here. Sometimes a “fake it ’till you make it” attitude helps, and so repeating affirmations can help loosen things up. At other times breathing and movement can help to at least distract us long enough so that we don’t spiral further. Sometimes, personifying your akrasia as a silly character can really lighten things up. Sometimes just stepping away from the situation for a bit is enough. Context is all-important here; you may have to try a few different things.

If kindness is not available, you can also try reframing the situation. Write down your goals an then break them down into SMART goals. This helps with discernment regardless of kindness—that can come later. As we said, akrasia storms are complex, and the solution to them is complex, context-sensitive, and requires creativity.

Take heart! Every writer goes through these difficult moments. We are united in our hidden, slightly shameful battles with akrasia. As long as we don’t over-identify with the difficulties and develop a kind, patient discernment, we will be all right.

How to Keep Trying in the Stuck Places

admin

Writing is hard. Learning to write is hard, and the more we learn the more sensitive we get to ways we could improve. Learning and growing is magical and wonderful, but it often invites a sense of futility. Let’s explore how to navigate this!

The opening of Dante’s Inferno cuts to the root of it:

In the middle of the journey of our life I came to myself within a dark wood where the straight way was lost. Ah, how hard a thing it is to tell what a wild, and rough, and stubborn wood this was, which in my thought renews the fear!

All who aim at mastery and excellence must hit many plateaus on the way. These are the places where the low-hanging fruit has been plucked but the goal is still out of reach. At these stages we tend to make great efforts that feel futile. While the growth is agonizingly slow and subtle, the frustrations are obvious, and it feels like they are winning.

Those who achieve excellence in anything share a certain mulish stubbornness in the face of such plateaus. Natural talent and “genius” are quite secondary to this main virtue of stubbornness.

How do we live in this space? We don’t think there’s anything mystical about it. Dealing with the plateau is a learnable skill. We just have to get clear on our goals and keep trying. That’s the hard part.

Yoda and Grace

Consider Yoda’s maxim: “Do or do not. There is no try”. This is a great truth, but during the plateau stage we couldn’t disagree more! Yoda is skipping steps. His words ring true only after the plateau.

The opposite of Yoda’s great truth is another great truth: we surmount the plateau through nothing but trying.

Grace Greenleaf puts it well. Since Greenleaf is not as well-known as Star Wars, we need some context. Grace is an apostate Christian who has just arrived back to her megachurch-operating family. They’re all at a tense family dinner. Grace’s mother is grilling her about whether she believes in anything at all. Grace responds thus:

I guess I believe there’s a part of everything that tries. You know, like plants try to grow. Animals try to survive. People try to better themselves to get ahead. Everything tries to do something, and I guess I think that Christianity is just one way that the trying part of people tries to connect with the trying part of everything else.

In the dark wood, on the plateau, we need to be more like Grace and less like Yoda.

There are two things to point out here. First, notice how much more tentative Grace is than Yoda. Second, notice how tender and open this attitude is to the possibility of failure. Everything tries, but not everything succeeds. Regardless of outcome, we can connect to the truth of trying as a way of staying with the process.

Honour your Trying

It’s always tricky to give practical advice when dealing with the plateau. Whether the advice hits home or appears simplistic can shift hour-by-hour and day-by-day. That said, here are some hopefully inspiring generalities.

Inside this lavender there is something that tries.

It’s easy to get stuck in achievement or productivity mode because getting things done is easy to get a handle on, especially in difficult times. The problem is: so much of productivity-mode is busywork. Individual trying is often de-emphasized in our culture. But it matters! Find little ways of enjoying the work you do! Find tweaks, adjustments, and ways to make the work less of a slog. Much of that involves ceremonializing your own trying in ways that feel appropriate.

Separate Process and Product

There are many ways to do this. Sometimes just breaking down a goal into smaller sub-goals is enough to get us out of product-mode and into process-mode. Sometimes recognizing that the product is mainly for others, but there are pieces of the work that are individual, ineffable, and just for you will do the trick. Perhaps reminding ourselves that we are most in touch with ourselves when doing creative work will help. Nobody can walk the path for us, or replicate the path we’ve walked. And that matters.

Just keep Trying

This one is most likely to be misunderstood. It doesn’t mean stiffening up with effort or powering through tiredness and resistance. Finding ways to keep trying when stuck demands creativity, discernment, and sensitivity, not force. If force were all you needed, you’d be out of the dark wood already, since nobody wants to dwell there for long.

Finding ways to keep trying involves finding little bubbles of kindness for yourself in tough situations. While the particulars are unique and individual, finding kindness doesn’t have to be lonely. Reading about other people’s process is often useful in sparking something in us. Of course, meeting with an academic coach might help as well.

We hope some piece of this work has been helpful to you. You may be finding the work hard because the work is hard. We’ve lived this life for a long time, and want to help.

Artificial Intelligence, Writing, and Editing (Part 6): Mediocre Computing and Excellent Writing

admin

The discourse around large language models (LLMs) continues to pick up steam, and shows no signs of slowing down. Just this week, we have the launch of GPT-4, which will mean we all have to update our sense of what LLMs can do yet again!

There’s a week-by-week overview of AI developments here. These takes about LLMs and literary fiction are interesting and probably right. This book will help you worry more precisely about the near-future of AI technology. There is too much to read, but we aren’t slowing down!

This week, we’re being optimistic. We think that once things settle, LLMs will push writers towards excellence because there will be no other natural home for human writing. In other words, LLMs will supplant mediocre human writing of all genres, and this will probably be good. Mediocre writers will have little reason to exist, so human writers should aim for excellence—perhaps with the help of editors.

Let’s dive in!

LLMs and Mediocre Computing

The advent of LLMs inugurated what Venkatesh Rao calls mediocre computing. Whereas previous impressive AI breakthroughs like Deep Blue or AlphaGo were examples of excellent computing, LLMs are impressive because they are mediocre.

Mediocre computing is computing that aims for parity with mediocre human performance in a realish domains where notions of excellence are ill-posed.

Excellent computing is computing that aims to surpass the best-performing humans in stylized, closed-world domains where notions of excellence are well-posed.

AlphaGo and Deep Blue reached excellent performance in stylized domains. LLMs have reached mediocre performance in realist domains.

Realish domains are enmeshed with the real world’s messiness and complexity. What makes them realish rather than real is that some of this complexity is codified and simplified. The urban road system is realish whereas off-road driving is a more real domain. We live most of our lives in realish domains, so an AI challenge to our performance here hits harder.

Skills learned in one realish domain transfer and leak into other realish domains. Competent language use by LLMs requires coping with the many realish domains that our language tries to model. Language is an open-ended system that can be adapted to model any realish or real domain, and so the impact of LLMs is both impressive and open-ended. Which is why there’s so much discourse about them.

LLMs Tend to Mediocrity

The LLM training process has been covered in many places. Put briefly, LLMs see huge amounts of example data and using vast computational power extract patterns in the data at all sorts of scales. LLMs end up having models of the many ways language is used. With these models, the LLM determines the likely continuation of any given string of words (really “tokens”) in a way that’s responsive to the meaning of the words.

THe LLM spiderweb continues to expand!

LLMs are trained on all the language that’s fit to print, post, or otherwise appear on the internet. This means that most of what the LLM sees is language use in its compete mediocrity. For every hard-hitting, well-crafted poem, LLMs see thousands of sloppily written poems. For every piece of excellent prose, it sees millions of chat logs, forum posts, pointless arguments, and self-indulgent purple prose. It’s statically certain that the LLM defaults to mimicking these patterns of language use.

Prompt Refinement

To be sure, people are getting better at tweaking prompts to be specific, to give the LLM context, to help it stay in a particular style, and many more desirable qualities. For example, the LLM “knows” about Milton’s Paradise Lost, so you can get it to write in that style. Note that getting good at prompt engineering is a human skill. A beautiful piece of LLM-generated prose will, we think, take as much human skill to extract from the LLM as a comparable piece of human-written prose. We leave it to writers to figure out how and when LLM use will benefit them. Notice that this mode of LLM use doesn’t trigger the usual fears that the machines will supplant us. Instead, our labour mixes with their power. Using LLMs will be akin to programming, except programming that people trained in the humanities will have some advantages in.

Prompt refinement is good both for mediocre and excellent writing. For most mediocre uses of writing, there practically no downside to getting an LLM to do it. Nobody will miss spending extra time tying out mediocre language. For excellent uses of language, the intrinsic worth of the project will guide us; LLMs can enter into the workflow as assistants in many ways, some of which we’ve explored here.

So, we think writers should be cautiously optimistic. For writers aim at excellence, we think LLMs will help with structure. Writers will become more editor-like, and we editors will continue being editors.

Attention, Writing, and Editing

admin

Following from last week’s post on deep work, we’ve been thinking a bit more about attention. Skill in writing is, among other things, skill with subtleties of attention. So let’s explore that!

Attention: Some Nuances

Cultural scripts around attention are somewhat broken. The intuitive metaphor for attention is something like a spotlight, which decades of cognitive science have shown to be a simplistic way to think. So let’s give a slightly more nuanced analysis.

Attending involves at least three distinct kinds of shifts: figure-ground, feature-gestalt, and transparency-opacity. When we talk about moving attention, we are really talking about one or more of these shifts.

Figure-ground shifting tracks what you foreground and what gets relegated to the background. For example, imagine talking to a person at a loud party. You have to exert some effort to foreground their voice and background all the other sounds. Often we fail to sustain our attention in this way; for example, someone may call out your name and suddenly you’re foregrounding something else.

Feature-gestalt shifting tracks whether you attend small structures or large ones—whether we are noticing the proverbial forest or trees. We all know what it’s like to attend more widely or more narrowly.

So much to attend to! Leaves or branches? Tree or sky? Form or content?

Transparency-opacity shifting tracks whether we focus on or through some layer of our attention. Imagine looking at a storefront’s pane of glass. You can attend to your partial reflection in the glass, or you can attend to what’s behind the glass. Similarly, when writing a sentence you can focus on the meaning you intend to convey or you can focus on the specific phrasing. In the former case, the words you use are more transparent to you than in the latter case. Attention is intricately layered, and it’s surprising what we can make transparent and opaque to ourselves with a bit of practice!

Habits of Attention

These three shifts often happen together. Usually, when we focus on the gestalt, we’re also making a lot of stuff transparent to us, and when we focus on the features, we’re making a lot more stuff opaque to us. Both moves involve subtle shifts in what is foreground and what is background. There are probably also other sorts of moves that haven’t been described yet. Attention is complicated!

That said, we think that this rough scheme makes editing a bit more transparent (ha!) as a skill set. Different phases of writing draw on different attentional moves. Editing involves taking the natural attentional moves writers make and dialing them up further.

Forest and Trees

Writers have to move from the features to the gestalt in all sorts of contexts: laying out plots, judging whether there is enough tension in a piece of writing, keeping things consistent across chapters, making sure writing flows, or making sure arguments make sense. The reverse move from forest to trees is also important: laying out the specific of dialogue, setting the stage properly, finding just the right adjective, and so on.

Developmental editing is a process of assessing a piece of writing as whole. Skilled developmental editors tune their attention far out into the gestalt direction, so that the forest of the piece is foremost in their mind. Editorial suggestions in developmental editing serve to make the feature-gestalt shift easier for the writer as well, since writers have to balance between the features and the gestalt in a way that editors don’t.

Transparency and Opacity

Writers move back and forth between these all the time as well. For example, you know where a paragraph is going, but now you have to move through it sentence by sentence. To do this, you have to make your words more opaque to you. Or you find yourself in the weeds of a description, but you lost your way, so you make the words more transparent by attending to the function of the sentence you just wrote.

Copyediting involves editors assessing the text with their attention tuned far in the opacity direction. This allows editors to catch errors, unclear phrasing, or unnecessary verbiage that writers can’t catch for themselves because since writers know exactly what they mean when they write something, the details naturally become more transparent.

We hope you found this brief dip into attention helpful for your own work, whatever it may be. Greater attention to attention is the way we reclaim our habits, our energy, and our capacity to work deeply and meaningfully, whatever the work may be.