Rough and incomplete working notes, to help me think through how to use this site. Intended primarily for my own use. But I will also share this with friends, as grist for conversations about how to best design a website to support creative work. I'm a beginner at thinking about this question, so the notes are naive and very rough. Still, perhaps a few readers may be interested in peeking over my shoulder as I think through the question. Much of this won't apply to other people: it depends strongly on idiosyncracies of my values, creative goals, and personal working style.
Initial release: 10-11-2021
Last updated: 10-13-2021
“What and how do you write a personal site with the long-term in mind?… What sort of writing could you create if you worked on it (be it ever so rarely) for the next 60 years? – Gwern Branwen
Most personal websites aim at personal connection with other people, or at establishing oneself professionally. They're not usually about helping the author think and create, except incidentally.
Those things – personal connection, professional marketing – are important. But as the purpose for a website I can't get excited about them. But I can get excited about the idea of using this website to enhance my ability to think and create.
Here's a few examples I've drawn inspiration from: Gwern Branwen, Cosma Shalizi, Bret Victor, SlateStarCodex, Terry Tao, Devon Zuegel, Piero Scaruffi. In many cases I suspect those sites have greatly helped the authors think and create.
A few thoughts follow below. Let's begin with a few of the more important principles, plus a few miscellanea that don't really belong later in the essay. Some of this is repeated later in the essay.
The design is something I greatly admire about Gwern Branwen's site. I don't mean I want to emulate his specific design choices, though they are a source of stimulation. But the time and effort put into the design are clear in Branwen's site. He cares.
One conventional reason people care about such design is to make the reader's experience better.
But a better reason is that it makes the writer's experience better. In particular, improving the design can be a way of enhancing your ability to think and create. When you invest time in the design, when something looks and feels beautiful, you take it more seriously yourself.
(Aside: this happens to students of mathematics when they begin using a serious typesetting system like LaTeX. Somehow the work seems more real when it is beautifully typeset on the page. This effect is compounded when they first see their work in a journal or book that is professionally typeset. For many authors of books, it's a profound experience to have a bound paper copy in their hand for the first time. It makes the work real in way it wasn't before.)
The emotional importance of design is a challenge for me. I am not a competent website designer. And unfortunately the paragraphs above have a complement: doing a poor job actively fosters the wrong emotional relationship to the work. You get the opposite of the LaTeX effect – if you make your work look amateurish, the content will seem more amateurish to you. That's really bad.
The most obvious solution is “get a professional to do the design”. But that's not possible here, given the design goal: to create a website which supports my thinking and creative work. In some sense I must be the principal designer. But, of course, I can and should get assistance from people who know more than me. The solution – which I'm still in the early days of implementing – is straightforward:
Just keep iterating on these points. And keep in mind that the aspiration is this looks and feels superb. But also understanding that this is an ongoing process. If something feels amateurish right now, that's something to notice – and to have a plan to fix!
If I make a serious long-term project of learning the elements of design, and improving the design of the site, then it will gradually get much better. Especially if I treat this as a lifetime project. It will gradually go from “looks terrible” to “looks mediocre”, to “looks okay”, and so on, up to: “looks and feels superb”. Of course, this will take hundreds (and probably thousands) of hours of work, over decades. But the reward is a feeling of great pleasure and enjoyment of my own work. That's worth the investment of time!
As mentioned above, I greatly admire Cosma Shalizi and Piero Scaruffi's sites. But design-wise both appear stuck in 1995.
How to reconcile this with the points made above?
I'm confident that if I left my site with their design aesthetic it would strongly negatively impact how I felt about my own work.
That said, both those authors have put a tremendous amount of work into unconventional aspects of the design. In particular, they've both thought a great deal about the organization of information: about categories, hierarchies, navigation, information density, and many other elements.
It may not be beautiful, and certainly one may quibble with their choices. But there is nonetheless a great deal to be learned. While it's not what people often think of as design, it is information design, and in many respects it's extraordinarily good. So there's both a lot to be learned from these sites; and also quite a bit they leave on the table.
The sense in which I used “design” above was in the (very rough) sense people often use when they're talking about web design – choosing good typefaces and layouts, making things work effectively for the reader. But there's a very different meaning, which is of imaginative interface design. There the question is to find new fundamental objects and actions; in our case, with the object of supporting thinking and creating.
This is something gwern.net does unusually well. As just one example: at the top of each webpage is a rating of the webpage's importance; of how certain he is of the contents; of how finished the page is; what links to the page; and a bibliography of all links.
These are all unusual actions to be able to perform on a webpage. They change the experience of both the reader and (perhaps even more) of the author. They are quite deep ideas of interface design3.
More profoundly, Gwern Branwen's webpages are truly interlinked. Each page is a (often very sophisticated) narrative interface to a large chunk of the rest of the site. This is what many people thought the web was going to be early on. It's true of Branwen; it's true of Shalizi and Scaruffi too. But it's shockingly rare on the web as a whole.
In general, anything which produces a feeling of unease or embarassment can disrupt fluency. And it's worth working very hard to gain fluency. You want to avoid disrupting the flow of ideas. A reduction of 10% in any one of the above is tremendously worthwhile. I expect that many can be addressed in part through use of scoping statements at the top of a piece, to set expectations. The goal, of course, is to set expectations to help you think fluently and well. Ideally, you'll work with a kind of insatiable positivity.
Page types matter more than I think: You can see this in the inspiration sites I mentioned above. Many of the authors have thought hard about the different types of page on their site. gwern.net's metadata, in particular, gives a very fine-grained account of the purpose of a page. Much of this is about getting your own relationship to the page right: when that relationship is right, it should make it relatively easy to write. And if you're struggling, it may mean that you're trying to write the wrong kind of page.
You need to unlearn a lot from blogging, and from online writing in general.
It's true there are many tremendous blogs, blogs which support the thinking of the authors – Terry Tao, Tim Gowers, Scott Alexander, and many others. But I often wonder if they're fighting against the form.
(This is not to say that blogs aren't exceptionally good for many other purposes. And, as those examples show, they can be made to work as supports for thinking. But it does seem to be a bit of a battle.)
One trouble with blogging is that people subscribe to the blog, rather than each post finding its own audience. This encourages lowest common denominator writing. In particular it creates a context in which depth is penalized5,6. Gwern Branwen's articles would be considered strange as fodder for a blog: extremely in-depth pieces about the Dark Net, about spaced repetition memory systems, about anime, about machine learning, and so on. Making the site not-a-blog disaggregates the audience, and (I presume) makes it easier to convene the right (or at least a more apt) community for each individual piece.
A media form which does this better than blogs is academic journals. Although it is possible to subscribe to a journal, it's very uncommon to expect to read a journal entire. And the reason is that the form is designed so that each article finds its own audience. For instance: the title and abstract are designed to make it super-easy to not read a paper. They say “Oh, unless you have these interests and this background, you should ignore this”. They're sometimes described as “marketing” for papers, but really they're better thought of as anti-marketing. In particular: they set the right expectations, and help drive away the wrong kind of reader. That's invaluable not just for the reader, but perhaps even more for the author.
Another problem is that blogging is by-default ephemeral. It's not an interface to your past thoughts or your future plans. It makes such organization an afterthought, not a first-class object.
I've been dwelling hard on the negatives of blogging! Of course, there's much I admire about blogs and blogging. Still, I hope the above reflections show that it's useful to reflect on what doesn't work about blogging; what makes it uncomfortable as a medium for thinking and creating. The next section looks at a particularly pernicious form.
One of the worst ideas in online writing is the unqualified idea that you should be trying to build an audience or to build engagement. This may be good if you're trying to sell something. But it needs a lot of qualifiers if it's to support good thinking7. In fact, engagement can be a curse, actively damaging to good thinking8. I've already alluded to one form above: the way it may inhibit depth. But there are other problems too.
Suppose you're thinking deeply about something. You'll stop often. Get stuck. Backtrack, over and over – and over and over – again. You'll make mistakes, sometimes bad ones. Sometimes ones that you'll find embarrassing. You'll have half-baked and incomplete and not-quite-working ideas. Occasionally you'll have ideas that are truly dreadful. You'll pull ideas from all kinds of places, often including technical ideas only a tiny handful of people can be expected to know. And you'll want to use those ideas as rapidly as you are able, freely associating.
And yet almost all of these actions violate the conventional wisdom about how to build an audience and engagement.
A natural response to the above is: well, why not just do the thinking in private, then? Why impose it on the public at large?
This is a false dichotomy. There's a very large gap between “think and write for just oneself”, and “try to write for the largest possible audience”. And different types of creative work are best done at different places within that gap. That's really the question: where exactly in that gap do you want to be? Or, to put it another way: how best to define the audience to support your creative work?
I do keep some private notes in my local version of the site. It lets me reuse a lot of the habits and infrastructure and workflow of the site. But the challenge of private notes is that many of our standards for good work are held socially. Most people find this: you think you understand something, then try to explain it to others, and find your understanding has important holes. Or you find that when you speak to others you naturally improve your explanations. I often find that I will all of a sudden boil stuff down in really helpful ways. I'm a little sheepish about this: there's a kind of ham-ish desire to please others. But I think it's probably extremely common. Humans don't seem to be made to think alone.
There's something strangely difficult in writing just for oneself. As far as I can tell, almost no-one can do it productively9:10. We think better when the stakes are higher, and one of the best ways of raising the stakes is to make a document into something you're sharing with people whose good opinion you desire. Even just sharing with one such person makes an enormous difference to the quality of your thinking.
Thanks to Gwern Branwen, Laura Deming, Andy Matuschak, Kanjun Qiu, and Devon Zuegel for helpful conversations about the ideas behind these notes.
A metaphor I like is that this is about creating an intellectual home for myself. And that means designing and furnishing that home well, and developing habits that I'm comfortable with using in that home.↩︎
I have a sneaking suspicion that essentially anything that can increase speed is a good thing. I have my doubts about the parable of the pots (quantity leads to quality). But there's undoubtedly more than a grain of truth to it as well.↩︎
This is not to say all these ideas are original to Gwern Branwen. But they are certainly not widely used.↩︎
And lots of very good critique. I'm just talking about the low quality critique. Online criticism is strange: surprisingly many people seem determined to establish their superiority or their righteousness (or both) by criticising other people and their work. It's horrid, and can be contagious: I've occasionally spent time with such people, and then caught myself doing the same. Note that it's the intent which mattters. Critique as genuine support for doing great creative work is wonderful; critique for its own sake is not. Of course, many of the superiority- or righteousness-seeking critics tell themselves that they're just trying to make the world better. But on the receiving end it's usually pretty obvious which camp someone falls into.↩︎
Terry Tao's blog is perhaps the best exception to this rule. Many of the pieces require a great deal of background, which (I presume) most readers don't have. In some sense that's the expectation he's built for the blog: most entries will be mostly inaccessible to most readers. But it's a very unusual blog in that sense, subverting the conventional assumptions about blogging.↩︎
It's also interesting to consider Scott Alexander's writing in this sense. Unlike Tao, Alexander's entries are relatively easy to read. It makes sense to bundle them up as a blog: there are common themes, and they're relatively accessible. It also helps that Alexander is a superb writer who can draw a wide variety of people into many different subjects; he does a very good job of making the common themes explicit.↩︎
It reminds me of the enormously damaging idea that success in life is equivalent to making lots of money. One way it's similar to that idea is that when you're in a milieu (like Silicon Valley) where a lot of people believe it, it starts to infect your thinking in ways extremely hard to notice and free yourself from.↩︎
It's a sneaky idea, too. It informs all sorts of cultural norms in tech and the media, both of which default presume that engagement is good and a sign of value. If you're in contact with those cultures then it's easy to absorb some norms which make this presumption.↩︎
There are famous exceptions. I'm a fan of Simone Weil's writing, much of which seems to have been done largely for personal reasons, and only published later. But while that writing is extremely stimulating, it also has many shortcomings. I believe some of those shortcomings might have been avoided if Weil had been writing for an audience broader than herself. An example: many of her arguments are made abstractly, without reference to specific examples. When writing just for oneself this kind of abstract argumentation is tempting: you often have specific examples in mind, which you don't need to make explicit. If she'd been writing for a broader audience I believe – I am speculating – that Weil would have written the examples down. And my own experience is that when you do that detailed work you often uncover problems in your abstract argumentation, which in turn leads to improvement.↩︎
Of course, many people keep private working notes as part of projects that ultimately yield public output. I've done quite a bit of this myself, but nearly always find there to be something vaguely unsatisfactory about the notes. It may be useful to think through the why of this in more detail.↩︎