Month: August 2020

Back in January we discussed Twitter threads being used as mini-blog posts. Later, when I reflected on how I planned to consumer Twitter (and Reddit) going forward, I had said

… people’s use of Twitter has changed in the last year. The cumulative effect of the doubling of character limits, the ease of creating Twitter threads, the grouped display of conversations and Twitter’s own quality-filtering is that there are interesting, valuable discussions on Twitter itself…

Earlier I came across this full-fledged 29-tweet threaditcle on the oat milk producer Oatly.

1/ Oatly doesn’t think like the rest. They’ve been around for 20 years as a Swedish company fighting for attention. Last week, the oat milk company raised $200mm at a $2bn valuation. This is a lesson on creativity and how @oatly turns disadvantages into massive opportunities.

— Kevin Lee (@kevinleeme) July 24, 2020

Not only can tweets display text and linebreaks like they always have but now, just like on Medium or WordPress, tweets support multiple images, links, embedded media like video, all with rich previews. Unlike other publishing media, tweets also support calling out other people via @ replies. And although they can be viewed as a single unit, each tweet – inherently – supports its own set of breakoff conversations that can cross-reference each other [1]. Each tweet can also be retweeted, creating other conversations with the retweet as the parent. This sort of native emergent remixing is just fascinating.

Having said this, there doesn’t exist today a frictionless, natural way to view these conversations. The comments on a single tweet are still displayed primarily linearly, when in reality they’re like a tree. i’m not suprised to see this today:

I wish Twitter had a “follow conversation” option.

— Mark Johnson (@wmdmark) July 26, 2020

The service Threader today creates a single chronological view of twitter threads, but I’d like to see it extend that capability to conversations, since increasingly that is where the value lies.

[1] Medium used to have permalinks to each paragraph. I can’t find them anymore. Perhaps they are only available when logged-in as a reader, and I do not have a Medium account. Dave Winer’s blog also has such permalinks.

Continuing for a day more with the topic of Fitbit: many analytics that go beyond the basic visualization that is displayed in the app or is available for download on the web dashboard, is only available via a paid subscription, which Fitbit calls Premium. We talked about Premium a little in the context of discovery vs curation.

To give you an example, this is the sleep phase chart in the company’s app for a single night’s sleep:

From the Fitbit site, you can export your sleep data. But my data for the week only includes summaries on this sort (data hidden)

So clearly Fitbit has access to much more granular data to generate that chart than is required to generate that table. However, that access is only available via the Fitbit API. For instance here is the API documentation page for sleep data.

This endpoint supports two kinds of sleep data:

stages : Levels data is returned with 30-second granularity. ‘Sleep Stages’ levels include deep, light, rem, and wake.

classic : Levels data returned with 60-second granularity. ‘Sleep Pattern’ levels include asleep, restless, and awake.

If you write your own software, you can authenticate to and download your own data at down to thirty second granularity. For the example in the chart above, that’s 1024 data points, each stating the timestamp and which stage of sleep I was in. This is extremely valuable (and not to mention my own data recorded by a device i have paid for!) I support Fitbit’s decision to restrict analytics such as sleep score to premium users – this is Fitbit’s IP and its prerogative. However, it does need to make its users’ raw data available for download much more widely. It should not require technical expertise.

This article describes an example of a person downloading and charting his own heart rate data from Fitbit. I intend to teach myself to download and process my own data. In fact, I am going to explore if I can use iOS shortcuts and the actions that Data Jar makes available to Shortcuts to process JSON data – and Charty to plot graphs. We will revisit this some time.