In April last year we began work on a new product to sell .blog domains. We just open sourced this project. Check out the announcement from my friend and team-mate, Stéphane Thomas.

Screen Shot 2017-02-10 at 10.52.02.png

This was a really fun project to work on. We worked very closely with a team of designers to trial some new product concepts (like passwordless signup), and were able to experiment with some interesting new technologies (like CSS modules). I am really proud of the things we achieved together. I believe that this is an great demonstration of how a team is greater than the sum of its parts.

If you want your own .blog domain, then take a look at


For thousands of years humans have been communicating in two main ways, spoken word and written word. Spoken word is generally real time and responsive – one person speaks, another responds; a conversation. Written word is usually delayed, and long form – someone will write a letter or an email, and the response will come hours or days later. Over time many conventions have been established in these communication channels – we read body language, tone, expression, and many other subtle cues in the message. This is harder to do in written communication, but the cues are still present.

In the last few decades these two communication channels have merged in the form of text based chat. Text chat is written communication, but its short form and often real time. It also has subtle cues that can be inferred, which I have dubbed “subtext”. Many of these cues are lost when the conversation is read back. Some examples include:

  • The “typing” indicator; often the typing indicator will show for a long time, even though the messages being sent are very short. This is a sure sign that the person you chatting to is having trouble putting their thoughts into words. I
  • The timestamp of the message; sometimes replies to message get sent after another message has been sent, so the replies are out of order. This is often clear at the time that the conversation is happening, but when it is read back it could be misleading. The timestamp can often help clear up this confusion.
  • Typos/messages sent early; these mistakes may indicate some level of emotion on the behalf of the sender – they might be excited, or upset or tired etc.
  • Read indicator; Sometimes replies to messages are slow. This may indicate that the person you are talking to isn’t engaged in the chat, or that they have other priorities. This can be understood later from the time of the message,  but more information can be inferred from the “read” indicator on messages. If a message has been read but has not been replied to this might be an indicator that a reply is difficult, whereas if the message has not been read the delay is more likely to be because the chatter is busy elsewhere. Some messaging apps also distinguish between sent and delivered messages, which provides additional subtext – if a message isn’t even delivered then their phone is probably out of power or they are out of signal.
  • Last seen time; The time the chatter was last seen could be a telling sign. If the person you are talking to was seen more recently than your last message then they might be ignoring you, or engaged on another chat.
  • Length of messages; Short messages might be another indicator that the conversation isn’t a priority for the chatter right now, whereas long ones a sign that they are engaged with the chat.

Like all implicit communication it is important to treat these cues with caution. There are many other plausible explanations for the behaviours listed above. Just as with spoken and written communication clarity is important; if the “subtext” of your chat is conflicting with the content of the chat then it is probably best to make your concerns explicit, so that the implicit cues are not being read incorrectly and causing confusion.

What other kinds of “subtext” can you think of?