Ideally I’d have just a single inbox, but in practice I have several.
In fact, I have so many that I forget about some of them, and they fade away. Especially if we count all the different Slacks I’m part of, there’s no way I could keep up with all my inboxes. Here’s a quick list of the inboxes I’m currently trying to stay on top of: Personal email, Work email, Texts and Voicemail, a Twitter DMs, Facebook Messenger. In addition to these, there are more inboxes that I check but don’t really try to stay on top of: a shared Roam graph, some Slacks, and some Discords, and various social medias and news websites. Beyond these are the inboxes that I simply don’t think about these days: various forums, other Slacks and Discords, old game message boards. I’m blanking on the names of some of these, that’s how much they’ve faded.
Even in my ideal world with a single inbox, there are different kinds of items that could reside in the inbox. I think the major designations are: requires attention, deferred, and indexed. Basically my thinking is, some items (i.e. emails, texts, voicemails, etc… all the items that are directly specifically toward me that I want to read) go into the actual inbox part of the inbox. They form a queue of things to process. I can process them by performing the task (responding, handling, etc) or by deferring (i.e. moving into a TODO list of sorts). Other items enter the inbox differently, and let’s take messages on a giant Slack channel as an example. These don’t need to be processed individually. They don’t even need to be processed in batch, to be honest. But it is nice to know that a whole batch of messages has arrived in the last week. I think my ideal would be that across many Slack channels and across many Slack groups and even across both Slacks and Discords and Google Chats, an overall digest of all the messages missed once a week or so would be nice. Eh, honestly that’s not necessary, the messages I have in mind are safe to ignore; it would be great if they remained searchable though.
To summarize the different kinds of messages I’m thinking about, there are: (1) emails to me (each time I process the inbox, email email is an item to handle), (2) messages in large groups (these should be batched, and frozen at the time of consumption, OK to ignore lots; their presence should not indicate the need for a processing session), and (3) items I’ve saved for later (this is task tracking; tasks can appear in inbox if a system surfaces them to my attention, but can usually get out of the way).
All the inboxes I try to stay on top of should be combined into one inbox and all messages therein fall into category 1. All the inboxes I check but don’t stay on top of should get batched and form category 2 (perhaps together constituting a single item of category 1). And any of these if ignored or deferred can become a category 3 item.
RSS in some ways is an effort toward constructing a single inbox out of many disparate message locations. RSS readers, however, seem to have gone out of fashion. And even in their heyday, RSS feeds never handled emails, texts, phone calls, Twitter DMs, etc.
Then there’s Sophia’s Unigraph. It shows promise as a potential unified inbox. I’ve seen as a demo it handles incoming tweets easily and is designed to be extensible to additional data sources. It wasn’t mature enough to take on One Inbox status last time I checked in on it, but perhaps with additional development it can reach that point.
Bieber Bot does so much for me, and one of the things he does well is reducing the number of inboxes I have. Facebook Event invites, reminders about upcoming activities, and some other reminders all come through Bieber Bot, whereas without him I’d have to check Facebook and various spreadsheets and calendars as well. I already treat Facebook messenger as an inbox zero location, where all messages are processed, making Bieber Bot (who contacts me through FB Messenger) a good choice for surfacing such things. I actually treat FB Messenger more as an inbox-zero location than I do email (though to my delight I am at inbox zero for emails too these days), I guess in part because email gets a higher rate of messages making inbox zero harder to sustain.
Suppose we were to build a new inbox, designed to be the One Inbox that replaces all others. It would have to handle receiving messages in a boatload of different services and sending messages in all these services too. It would have to handle easily adding new services that the developers didn’t anticipate. Imagine using automation software like Browserflow to specify how the service should gather new items for an inbox. For example, let’s say you want to integrate Roam or Twitter into your inbox, but the One Inbox developers haven’t heard of Roam or Twitter. So you make a Browserflow flow that collects the inbox items from Roam or Twitter, and One Inbox runs it periodically on your behalf to gather new messages. You make a second Browserflow flow that shows One Inbox what it means to send messages through the new service. Better even than that experience would be something like what the latest Adept AI demo is hinting at. In that direction, you don’t need to construct a program to show the One Inbox software how to receive and send messages; you just teach it with a combination of demonstration and natural language, and it can handle the rest from there.
As we build the hypothetical One Inbox software, there are latency considerations. How long after a message is received on a service is it before the message appears in our One Inbox? Does this matter? Yes. The case for yes: you don’t want to miss items. The case for no: if we’re following an infrequent processing workflow, several minutes of latency doesn’t matter most of the time. Clearly the yeses have it.
The software I use that comes closest to a One Inbox (besides Bieber Bot) right now is Ferdi. It doesn’t do anything intelligent to normalize incoming messages into a shared format across different messaging systems. That would be quite difficult (I think texts.com does this; I would like to try it, but haven’t yet). Instead, Ferdi just acts as a browser, loading all your different inboxes alongside one another. When described like this, it sounds silly and pointless – “why not just use tabs in your browser?”. Well, the experience in Ferdi is different than just using tabs in your browser because it’s a separate application from your browser, it handles loading and unloading differently, and services are pinned with little icons for easy access. These are mostly superficial things, but they add up to an improved experience compared with keeping lots of tabs for lots of inboxes open in Chrome.