Want to receive my work as I publish it?

LN 039

Notes on time

Our sense of time is, like computing, based on some unchanging fundamentals, and we build on those with more concepts that help us. Our days and years are set by the earth’s rotation and orbit around the sun. Our weeks and months are imagined constructs to help us.

We experience everything within one, linear progression of time. This simplicity lets us root lots of our understanding and thinking within it. It serves as a fundamental organizing principle for our memories of the past and our plans for the future. We experience and use its various scales — seconds, minutes, hours, days, weeks, months, years, decades, centuries — all quite differently, yet each of these measures is simply embedded within the next.

When we fabricate a digital world, a freeform environment in which our digital artifacts and experiences live, we can ground that limitless realm with a deep understanding of time [1]; one formed with more than an only literal interpretation of it, but also with a more human interpretation of time, which often ebbs and flows, stretches and compresses, as our mental constructs relate to different periods of time with varying weight and associations. Rooting the digital realm on a fundamental understanding of time would allow its human operators to rely on intuitions we’ve long honed and harbored, reigning in some of the complexity inherent in our work, in our lives, and in operating the digital realm in support of these.

Aligning timelines, and considering irregular timelines

In the last lab note, we saw a demo in which a runner brought weather data into the same timeline in which he was reviewing his pace from a recent run. This let him compare his pace with temperature and humidity data shown in the same timeline. When items are associated with timestamps, users can bring them together into unified interfaces, even when items are independent of one another (of different types and/or from difference sources). Users can do this without needing a developer to build some “integration” between the two; the system is capable of integrating this related but independent data when using the items’ associated timestamps.

But there is a catch from last week’s example: the runner’s timeline was irregular. The runner doesn’t think of their workout in terms of what happened from 12:00 to 12:30, then 12:30 to 1:00, and so on; they think of the first mile, second mile, etc. What is depicted in the pace chart are time slices which aren’t exactly the same as one another; the first mile may have taken a little longer than the second, the third has its own timestamps, and so on.

This chart matches the user’s understanding of the data they would like to interpret, so it’s important that the system understands this irregular timeline as well. When the system understands even this irregular timeline, it can allow the user to take advantage of the same properties of their system that allow combining independent, related data: dragging the weather data into the mile-based timeline is not a problem; the system can overlay the changing humidity at each mile’s timestamp.

This opens up lots of opportunities to bring together independent data which is related by time.

Timeline of use

Whether it’s visualized or not, consider the timeline that forms as you use some program: each action you take moves you forward in this timeline. This “timeline of use” is one of the ways we experience the operating system’s understanding of time. Your actions are associated with the specific times at which you took them. But: are there ways to move around this timeline?

If users don’t have simple access to past program states, then progressing in the software is a destructive action; it erases options that were available before. Where this dynamic is unintentional, it undercuts the user’s success in the program. In programs where progress is a destructive action, I often find myself falling into various anti-patterns to avoid missing out on other paths forward.

Many software programs’ interfaces are a cross-section snapshot from the end of the “timeline of use,” showing you the current state or composition. Some programs show you the timeline of use, depicting your progression through them in the interface as you go. And sometimes you can pick in which of these modes you work: When you use a text editor without editing what you’ve written, only adding to the text downwards, even as you add new ways of saying or understanding something written earlier in the text, you’re implicitly using this interface in the latter way; the document exists as a visualization of your timeline of use. On the other hand, when you use a text editor by continuously modifying the text, you’re using the interface in the former way, as a snapshot of the most recent state.

Your actions move you forward in this timeline of use, but in some programs, you can conceptually “move backwards” too. When you undo a few recent changes, the program may return to a state a few frames back on the timeline. In reality, you’re still moving forward, however: your timeline of use is simply progressing forward with your new actions. You’re working with actions that have timestamps in the past, and to do so, you’re taking actions that have new timestamps in the present.

The operating system of the future should have a clear physics for moving around your timeline of use — think of “undo” as a light-weight instantiation of this thinking.

In some kinds of programs, like a DAW or NLE, what you’re composing within the software is a timeline. In these programs, the depicted timeline you’re working with is the most recent state in your timeline of use. From left to right, you see the audio or video composition. The timeline of use is perpendicularly oriented to the timeline of your composition, flowing backwards away from what you currently see on screen. If an operating system had a more complete understanding of time, with universal physics or interactions to navigate the timeline of use, it would likely want to make careful consideration for how operators interact with timeline compositions such as these.

Moving between timescales

While working on the prior lab note, not being a runner myself, I watched videos from runners who use the same app as my friend so I could become more familiar. Something that struck me: a surprising percentage of the videos had various tips about seeing different overviews of your data. One tip was to see your weekly recap, which involved a series of taps around the interface, leaving your latest race overview, going to your profile, and into a fairly plain interface summarizing the week. Another gave the tip of seeing your yearly summary, which was somewhere else, and was visually rich. Then another gave a tip that you could see a portion of the yearly summary before the year was over (it may have been a month, or some other duration larger than a week), which was retrieved in some other way.

Is there room for a universal physics for moving between different timescales? It is a common interaction, but always implemented in a different way. If there were a common set of interactions, regardless of data type, for navigating timescales, users wouldn’t need these kinds of “tips” in each and every different app; the runners could navigate to the larger summaries from the overview of their latest run. They could also navigate down to more specific detail within their latest run from that same starting point, using the system-wide interactions for adjusting timescales.

Or, instead of moving between different timescales, are there ways to explore how different timescales can be represented in singular, fluid interfaces? For example, a calendar might show a lot of detail for your events today, but that also includes increasingly less-detailed overviews of future or past days or weeks on the periphery, as a sort of fish-eye lens on the present.

Where to from here

A “universal physics” for combining independent but time-associated data, traversing the timeline of use, and moving between different timescales would allow our operating system of the future to be more supportive of the ways we think in our lives and work. With a proper understanding of time, our interfaces can become more expected and more combinatorial. They can tap into our nature; our intuitions we’ve honed over the millennia. We can have a consistent set of interactions for the self-organizing dimension in which all of our experiences are rooted.

The last two lab notes looked at making gestural view construction possible. A fundamental understanding of time is instrumental to unlocking many of these ideas with all of the time-aligned data in our lives. Time-aligning things helps make it easier to spin up interfaces quickly: instead of calling up today’s rigid, one-size-fits-all interfaces, we want to spin up what we need for each day’s work. Time-aligning and similar approaches make it possible to do this without needing a developer to have built a custom integration between two data types or services. You get implicit relationships among lots of items, without having to set up any references manually, allowing you to create extemporaneous, meaningful organizations later as needed.


[1] We can construct the digital realm out of an infinite arrangement of concepts. Concepts can be imported from the physical realm, manufactured as new abstractions, or based on the concepts we use to understand and explore the physical realm, but iterated into the digital world. [Return]

[P.S.] An additional consideration for time in the OS of the future: ambiguous times / soft edges. One of the goals of OLLOS is to explore reifying the development and decay of the importance of the things in our lives. Quite often, things don't have a hard start and end time or date; sometimes things bubble up into an important class in our lives, after starting as something small; and sometimes things slowly drift away, taking up less of our attention and time as time goes on, eventually becoming unimportant. But rigid interfaces require we make firm decisions about things — timestamps must be hard edges, set down to the second.

Something spark a thought? Email me, or come chat on Mastodon or on Twitter.

« Previous: LN 038


Want to receive my work as I publish it?