Elderblog Sutra: 4

This entry is part 4 of 13 in the series Elderblog Sutra

The idea of hypertext trails predates the internet. Vannevar Bush envisioned trails and trailblazing as early as 1945, in As We May Think.

I am part of a long tradition of trying and failing to build trails technology. I led a team at Xerox that built a product, called Trailmeme (2008-12, RIP), that created navigable, visual trail maps of web content that looked like this.

It was a lovely product that did exactly what I wanted. We just couldn’t find a way to sustain it.

Failures stay with you in a way successes don’t. I’m still licking my wounds from that failure, but I’m also still trying to figure trails out. So are others. Every couple of years, somebody takes a fresh tilt at the problem and fails. One part of my elderblogging experiments is a second serious stab at trails, this time from the content side rather than the technology side.

The closest we’ve ever come to trails has been the special case of chronological ordering, which eventually became the stream UX metaphor. But piggybacking chronology as a way to get to trail-based organization is not only limiting, it is a kind of cheating. Like floating down a river, but pretending to be moving under your own power.

The best generalized embodiment of trail-based organization can be found on blogs, in the form of post series. But such series rarely go beyond 2-3 parts.

But we’re close to cracking trails. The key breakthrough has been the rise of threads on Twitter (an invention arguably attributable to Marc Andreessen). Twitter threads are genuine trails, even though they’re confined to a single platform. They are not chronological sequences. The key idea: trail-like structure is created during the act of authoring, not as part of subsequent curation. The trail authoring and blazing problems are coupled.

Series Navigation<< Elderblog Sutra: 3Elderblog Sutra: 5 >>

Get Ribbonfarm in your inbox

Get new post updates by email

New post updates are sent out once a week

About Venkatesh Rao

Venkat is the founder and editor-in-chief of ribbonfarm. Follow him on Twitter


  1. I believe scuttlebutt and the patchwork client pulls this off successfully. It’s also fully decentralized to boot. Graph structures are difficult because there are always new trails that can be curated through metadata tagging and filtering.

  2. I am really enjoying this elderblog series – lots to think about. I blogged for about seven years in my site before shifting over to newsletters as the main news/thoughts vehicle. in that time i had a crack at a whole bunch of irregular thematic threads, like musings of being rejected by lit journals or why ppl write poetry. wonder if these count as trails.

    fyi some of the threads here:

    on rjections: https://theotheradamford.wordpress.com/category/rejected/

    on writing poetry: https://theotheradamford.wordpress.com/category/why-do-you-write-poetry/

  3. Dave Doolin says

    I remember Trailmeme. It was pretty cool. I tried using too, but I don’t recall the results. (I have a Pearltrees account as well but I never use it.)

    I’ve mostly retreated from such structures in the last 10 years, in favor of simply producing code which reside in various DVCS. Although, I may yet be danger: I’m making a serious effort to learn orgmode. Thankfully org files (being text) are git friendly.