How A Twitter Scrap, and Covid-19, Reveal a Disruption In Process

Disruption, by Tim Lewis 2009

When is innovation just another stab at the past, and when is it revolutionary? When it becomes a bit of a Twitter storm in a teacup, is possibly when.

Here’s an interesting case study in the offing: You might need to get your head around some unfamiliar terms, like bi-directional linking, breadcrumb navigation and transclusion. Or not.

My thoughts were nudged by a post at Amplenote, an online note-taking and note-sharing app that is worth a look. Like a lot of these players, they’ve been forced into offering features by the new kid on the block, namely Roam Research, which by taking the genre by the scruff of its neck turned itself into something that knowledge workers are getting excited about.

The post was written by Bill Harding, founder of Alloy, the company behind Amplenote. They’re old style, by internet terms and their own words, not relying on VC funding, freemium models etc. You can tell by the product that it’s neat, robust and reliable.

That’s mostly what people interested in this kind of tool are looking for. I’ve been a note taker and outliner since the early days, and I’ve tried pretty much everyone that’s out there. Folks know that I feel that despite some great stuff, computer software has let us down when it comes to making software that understands us, rather than the other way around. (Why Won’t Computers Do What We Want Them To?)

But it’s taken Covid-19 to make me realise this is changing. And apps like Amplenote — or ones more familiar to you, like Evernote, are getting caught up in it. Roam Research has given us a glimpse of what note-taking — the simple act of reading, hearing, thinking or seeing something, and storing that somewhere — is ripe for disruption, and he’s going for it. It’s early days, but lockdown may be the jolt that propels this genre into the mainstream, something no other note-taking app has managed to do. A day ago Conor White-Sullivan posted to Roam’s Reddit group that 10,000 new users signed up to the app over the weekend, causing upheaval for his servers and users and forcing him to suggest to those ‘super-concerned’ use a rival app “for a week or so”.

He must know that suggestion really isn’t an option for most of his disciples. Roam is attracting a lot of interest — and beyond the usual numbers of people who dabble in this kind of thing. Why? Well, I think there are number of reasons:

  • Roam works right out of the box. It’s all online, the (initial) interface is very simple, even prepopulating the page with blank, but dated entries, prompting you to just start writing.
  • It gets as complex as you want it to get. You could just use it as a journal, if you wanted, but that would be a waste. It goes deep, with features being added with little fanfare, including sliders, nodemaps, and stuff I haven’t gotten around to figuring out.
  • But the key to all the excitement, I believe, is a key formula, which I think is the basic currency of software success: you get more out of it than you put in. Put simply, this means, for example, that if you linked one page to another page, that second page would update itself so you can see that link. This is what they mean by bi-directional links, or backlinks. It might seem to be the most trivial and useless piece of data, but if you don’t know what pages are linking to the page you’re looking at, you simply won’t know what information you have that you’ve already decided is linked to this. Your effort in linking to that page is now automatically creating extra value without you having to do the extra work.
Bi-directional linking, in simple terms

If you’ve read this far you’ll probably get it. But if not, let me just go into more detail. Computers, and the software that run them, are useless tools if they don’t allow us to do more with the stuff we tell them than we are capable of. Enter numbers into a spreadsheet because you know the software can do a load of things with that data than you can. Let your Apple Watch collect data about your body because you know it’s going to do more with it that’s useful than you could with a paper and pen. But for the most part knowledge workers have not really had anything similar for textual data. Yes, AI can help to find patterns in large bundles of it, but the same applied to our knowledge — the stuff we decide is worth keeping from our readings, talks, viewing and thinking in a computer — has not been so useful. Mostly, it’s just about being able to retrieve stuff more easily, so we don’t need to remember it, or remember where we put it.

But this is 2020. And we’re still there?

And this is where we come back to Bill Harding. And his screed. His point is a fair one: that all this talk of bidirectional linking is deja vu for many of us, when in the post-dotcom bubble burst of the early 2000s Wikipedia’s surprising success prompting interest in the underlying technology (here’s a piece I wrote in 2004 (sic) for the Journal about the disruption Wikipedia caused).

In his blog post yesterday Bill writes:

To drink in the enthusiasm we’ve witnessed in some corners of Twitter, bidirectional linking will evolve what’s possible…for those who dedicate themselves to the pursuit of learning this enigmatic craft. As fate & capitalism would have it, an elite cadre has popped up to help enthusiasts learn how to benefit from bidirectional linking. By all accounts, those who successfully assimilate the ideas of these programs transform their lives for the better. It seems reasonable, and I believe these gentlemen are doing great work to which they are wholly devoted.

But how different are today’s opportunities than what came before? To technologists of a certain age, the groundswell for a better connected network of ideas hearkens back to the halcyon days of 2005, when wikis were The Next Big Thing. Back then, companies like Wetpaint raised $40m to help organize the world’s knowledge. “A lot of venture money is flowing into wiki products” said Techcrunch in 2006. Having ubiquitous, bidirectional linking with surrounding context info was creating transformative opportunities in companies where people knew how to build them.

But with a couple major exceptions, wikis fizzled out, never catching on for personal use. Having set up my share of wikis during the early 2000s, I can attest that it was “worse than WordPress”-level bad for Twiki and Mediawiki. Developers might struggle through it to better capture their personal ideas, but the benefits of bidirectional linking were largely relegated to business knowledge software.

These new startups reviving and refreshing the ideas from the wiki craze is a great outcome for productivity enthusiasts. Especially with the expert guidance of smart people like Tiago and Nat, there’s never been a better time to help the humble wiki live up to its nearly-forgotten first round of hype.

If you sniff a little snarkiness in the tone ‘elite cadre’, you might be right. After trying to lure Roam users away with an import tool and a pricing comparison, Amplenote’s twitter account said less than an hour before I wrote this, that it had been blocked by Roam’s:


It’s not really a battle of equals: Amplenote has 24 followers, Roam 16,000. And this is the thing. What we’re really seeing, I believe, is a new player come in and bring the necessary tweaks and rethink to an existing technology — in this case, personal databases — by taking the best bits from each, adding a few new ones, and stealing the show to create a following and a buzz. This is not to say Roam isn’t impressive: it is, and I have not found any other app to match it.

And that’s why Amplenote and others, Bear, TiddlyWiki, WorkFlowy, TheBrain, Dynalist etc, are all struggling to add similar features, thinking about adding them, or defending why they don’t have them. This is good, and classic Christensen disruption. It might not be Roam that ends up winning this, but they’ve shaken up the market.

But what market? Is there really one for this kind of thing? Back in the early 2000s I would have said yes, because the kinds of people interested in this kind of thing were the same kind of people who bothered to read a tech column in The Wall Street Journal. The internet was a means of connectivity, and its potential was seen in those terms — could I store my stuff in more than one place, was the common question. So it wasn’t surprising that, as Bill recounts, everyone thought that wikis (the ‘readwrite web’) were the way to go. But others saw it differently, and all the smart money ended up going to using the internet to create more passive experiences — user generated, yes, but simpler, shorter, and where possible multimedia. It was all about eyeballs, and so content as knowledge slipped into the background as Twitter (status), Facebook (sharing stuff) and Google (search) came to the fore.

I’m not saying that’s necessarily changed. But Covid-19 has helped crystallise something that was already happening, namely that smart people are exploring how to leverage their knowledge and knowledge of software to solve the unsolved problems of the past, or to reconsider tools that had been largely forgotten. Knowledge work was once an obscure term that is now on its way to describing pretty much all of us who are sat at a computer, and it’s this realisation that has made people like Conor, I imagine, realise there’s a market for tools that really address the problem of deriving more value from the cost of user input.

Bill’s error is a generational one: what was once ‘business knowledge’ is now something else entirely. Watch this video of Andy Matuschak, a software engineer who works at the Khan Academy, to see what this looks like (he’s using Bear, by the way, not Roam). It’s strangely captivating viewing, ASMR for the Knowledge Generation:

Andy Matuschak at work

The other mistake is to think of this as ‘productivity’. This is not about that. This is not just a better task manager. I believe we’ve moved on from that — or at least recognised its limits. Now the thinking is, as Tiago Forte, one of Bill’s ‘elite cadre’, has mentioned, about acquiring and processing knowledge in a way that our brain retains it. You can almost hear Andy’s brain whirring as he processes what he’s reading before he expresses it.

So where will all this go? I’m not sure. Roam is talking about charging $15 a month, which is why people like Bill still think they have a chance to grab some of this market. To me it’s a rising tide and I’m pleased to see there are boats paying attention. After 20 years of focus on either ‘Getting Things Done’ or on the cuteness/elegance of interface, we’re entering a much larger ocean, which has the potential to bring these cutters, sloops and yawls into the slipstream of the incumbent tankers. Whether they go under or catch the current is anyone’s guess. Evernote has, largely, failed to find a larger audience (for lots of reasons) but the timing might now be right, especially as knowledge workers find themselves with plenty of time, isolation, an internet connection, and an urge to learn.

Addendum: Conor points out in a tweeted response to this piece that he has “been working on this problem for the better part of a decade. Strong(ly) agree the changes in landscape are great for us, but we wandered the desert for years before this.”

Why Won’t Computers Do What We Want Them To?

Photo by Aubrey, 2014
Tottenham Court Road. Photo by Aubrey, 2014

Computers and the software that runs them have long denied us the basic right of dictating to them — not letters and grocery lists, but of what they should actually do for us – most importantly in the first step of thinking: the art of taking notes.

In the mid 80s I was studying history in London, and the first consumer PC came out: the Amstrad. I was immediately intrigued, though I’m no techie. I remember going into Dixon’s one rainy winter afternoon on Tottenham Court Road and explaining my problem to the salesman. It was simple, I thought: I am a collector of events, and I want a computer which will do exactly what I currently do, but store it so I don’t have to carry around this pile of paper. It was simple, I told him. And I explained how I took my history notes, involving two or three basic steps. He looked at me blankly and tried to change the subject. “It comes with a printer and three spare disks.” I bought it anyway. But oh, how naive was I.

Because the reality is that 35 years on — 35 years! — there is still no way to do this. No app allows you to draw lines on a page and then add pieces to it wherever you want. I should know, I’ve tried hundreds of them (and if anyone does read this, I will get responses like ‘Have you tried OneNote?’ or ‘Aeon Timeline allows you to do just that.’ Yes, and no it doesn’t. No app, in short, is smart enough to just ask you what you have in mind and just evolve into that, to help you shape the app in the way you want.

This is the fundamental failure of computers, and computer software. As a technology it’s failed to really find a place in our lives that we’re comfortable with, and that’s because it has demanded too much change in our behaviours. We are mostly compliant: back in the late 2000s executives at telcos were worried 3G was for nought, because people didn’t show any interest in using their phones for anything more than calls and SMS. It took Steve Jobs to change that, by building a consumer device we craved to hold. The rest came naturally, because of a great UI, but no one is claiming that the smartphone adapted to us; we adapted to it. That’s not to say it’s not useful, it’s just not useful in a way that we might have envisaged, if we ever sat down to think about it.

Indeed, the Apple revolution, which I would date from about 2008 cannot be detached from the broader mobile data revolution, which we’re just emerging from. This was a revolution in interfaces, but it wasn’t a revolution in terms of computing. We have become more productive, in narrow terms — we are online a lot more, we send more messages, we might even finish projects quicker — but no one is claiming that our computers mould themselves to our thinking. It’s apt that movies like Her try to explore what that might mean — that our computers learn our thinking and adapt themselves to it.

So back to me and my history problem. There of course are answers to it, but they all require us understanding the mind of the person or people who developed them. And I’m not ungrateful to these apps; they have long been welcome bedfellows. From TheBrain to Roam, MyInfo to Tinderbox, TiddlyWiki to DEVONthink, they have all rewarded the hours — days, weeks, even — I have invested in trying to understand them. But therein lies the problem. The only reward one can get is if one adapts one’s own mind to that of the creator’s vision, and, however amazing that vision is, this in itself is an admission of failure. I don’t want to have to report everything to someone else’s vision, I have one myself, but there’s no software on this earth in 35 years of looking that I can wrestle into submission to my simple vision.

This is not to say the apps in question are a failure. I love them dearly and still use many of them. I have used my pulpits to promote them, and have gotten to know some of the developers behind them. These people are geniuses, without exception, and it’s not their fault their tools cannot be more than interpretations of that genius. We just lack the tools to tell our computers what to do from scratch.

Such as

‘Take an A4 sheet of paper, turn it horizontally so it’s in landscape, and then draw three perpendicular lines equidistant apart. Allow the user to write anywhere between the lines, and interpret a three-line dash as the end of each nugget. Interpret the digits at the beginning of each nugget as a date, which can be as vague as a decade and as specific as a minute. Order each nugget chronologically, whichever line it sits between, relative to each other, with gaps between according to the dates. etc etc’.

If only.

I still don’t see why I can’t have that software. I don’t see why I couldn’t have it in 1985. I probably could get a developer to whip something up, but then that’s already demonstrated the failure I’m talking about. I want the computer to do it for me, and not being able to, to have to rely on someone else’s coding skills, or even my own, means it’s not doing that.

This feeds into a broader point. Tiago Forte, a young productivity guru, wrote an interesting thread about the serial failure of hypertext, which was a precursor (and loser) to the simpler Web, and the lessons we can draw from it. In the case he describes, Roam. The simple truth: taking notes is a niche area because it’s not taken seriously at any stage of the education process (my history chronology capture was shown to me by the late and excellent Ralph B. Smith, who understood the power of note taking; I can still remember him demonstrating the technique in our first class. It has stuck with me ever since.) Note-taking is the essence of understanding, retaining, collating, connecting and propounding. And yet it’s mostly done in dull notebooks, or monochrome apps, none of which really mould themselves to what we write, take pictures of, record or otherwise store. (And no, Clippy doesn’t count.)

Tiago may well be right: the trajectory of knowledge information management apps (and there you have it; already segmented into what sounds like the most boring cocktail party ever) is that they just aren’t sexy enough to break out of a niche. Evernote was closest, but it got dragged down in part by its dependence on a vocal core of users who pushed it one way and its desperate need to justify its valuation by trying to go value. Truth is, people don’t value collecting information, in part because it’s so easy to recall: even with my 60GB DEVONthink databases, I more often than not Google something because I know I can find the document more quickly that way than in my offline library.

But this doesn’t explain the pre-Google world. Why did we let software go in the wrong direction by not demanding it submit to our will, not the other way around? Well, the truth is probably that computers were basic things, oversized calculators and typewriters for the most part. Sure it helped us write snazzier-looking letters, but heaven forbid us doodling on them, or moving the address around beyond the margins.

We’re still hidebound by our computers, so much so that we don’t realise it. I am rebuilding my life around the new tools, like Roam, and old ones like Tinderbox — a wonderful piece of exotica that is massive for those of us who like to poke around in a piece of software, but which basically means poking around in the head of its developers — and I get a lot out of them. But I am keenly aware that I would rather be just telling a blank computer screen to “take an A4 sheet of paper…”

And perhaps, one day, I will.

Extending Your Brainpower

This Software ‘Thinks’ Just Like You, But Makes Connections You Had Missed

WSJ Online June 22, 2007

By Jeremy Wagstaff

Here’s a heads-up on some organizing software that may take some getting used to. Frankly, it’s taken me nearly 10 years to appreciate its power. But now that I do, it has become something of an obsession. I even have dreams about it.

It’s a defiantly different kind of thought-mapping program called PersonalBrain, and a new version (including versions for Mac and Linux users) will be launched next month by U.S.-based TheBrain Technologies LP. Users include scientists, soldiers, inventors and others who have used it to marshal their collections of thoughts, projects and even databases on criminal syndicates. I find it so useful and absorbing, there’s nothing — be it a Web site link, a random idea, a contact, a document, a scrap of information — that I don’t add to its spider-web-like screen, knowing it will throw up links my brain had never considered or had failed to remember.

So what is it, and what does it do? Well, if you’ve ever created a so-called mind-map — a brainstorming technique that creates a burst of thoughts from one central idea or topic — you’ll notice the similarities. Ideas branch out from the center, organizing your thoughts in hierarchies. PersonalBrain, however, is less interested in building hierarchies, and more interested in mimicking the way the brain works. You nominate whatever is uppermost in your mind, and it rearranges things to illustrate the connections that thought has with other ideas in your head. Think less about branches, more about a freeform spiderweb.

PersonalBrain’s screen appearance hasn’t changed much since 1998, and it still looks contemporary. You’d be hard pressed to say that about any other software program. First impressions are positive. It looks good when you launch it, with its navy blue background and spinning central wheel and its spiderweb of links. Once you’ve got the hang of dragging little circles around a word, you get the idea of adding more threads to the web. Click on a word and it jumps to the center, the web of other words and links rearranging themselves around it.

It’s about now that new users tend to flounder a little. Certainly I did: I couldn’t immediately grasp the idea that PersonalBrain is less about getting a bird’s eye view of a subject — there are plenty of tools for that, like MindManager, or TopicScape 3D — than about helping you build and find connections between things you’re interested in. Italian consultant to the Italian police Roberto Capodieci (www.excomputer.net) uses it for tracking the connections among members of a criminal network, while British science historian James Burke uses it to track the links between history’s great inventors.

Now there’s nothing particularly magical in this. It’s not as if PersonalBrain is doing the linking for you. You have to build the links yourself. But remembering all the connections is something else. That’s where PersonalBrain comes in. Bali-based Mr. Capodieci, for example, adds a few basic terms (what the software calls thoughts) as categories — suspects, locations, criminal activities, phone records, etc. For each suspect, he adds a thought. Under locations, he adds places he is surveying — bars, restaurants, clubs — and then under criminal activities adds prostitution, drug dealing, robberies, etc. The next step is to start linking the suspects to the locations and to the activities. Pretty soon it is clear that two suspects in the same bar engaged in the same kind of activity are likely to know each other. Those frequenting more than one bar might be the links between two groups of suspects. Then he adds the suspects’ phone-call records, further linking them together and building a picture of the gangs he is dealing with.

Now your work or interests may not stretch to Soprano-like family trees. (Mr. Capodieci says he began using PersonalBrain when he found it installed on a hacker’s computer: the hacker was using it to store information about employees at a company to improve his efforts at engineering a scam.) But whatever your interest, however smart you are and however good a memory you have, you’re unlikely to be able to make and remember all these kinds of connections — especially over years. One longtime user, U.S.-based technology consultant Jerry Michalksi, has more than 60,000 so-called thoughts in his PersonalBrain, covering everything he has collected in 10 years.

While the new version dovetails better with Microsoft Outlook and has several important new features, it doesn’t feel quite as attractive as its predecessor. Nonetheless, after a few unsuccessful attempts I finally got going with it a few months ago, and now I can’t leave it alone. Pretty much every idea I have (admittedly not many), every Web site I like, every contact I’ve made ends up in my PersonalBrain, linked together by topic (such as “Web sites promoting good shaving practice”), or place, or friends in common, or temporary categories (such as “What I need to work on next”). While other tools would balk at trying to relate one item to more than a handful of others, PersonalBrain positively cries out for it.

Even after three months I’m only scratching the surface with about 4,000 thoughts. I’m also discovering connections that wouldn’t have occurred to me — and finding, lurking there in my PersonalBrain things I’d already forgotten I knew. As its creator, Harlan Hugh, said recently: “It’s like anything that’s truly new; you’re not going to be an instant expert, but we think it’s easy enough to see the benefits.”

I can’t guarantee PersonalBrain will help you sleep better. But if you persevere, I feel sure it will grab you as much as it has grabbed me.