Sunday 21 June 2009

Email considered harmful

I don't like email. More precisely, I'm disappointed that we don't have something that's better than email.

I suppose it's to its creators' credit that we are still using a tool that was standardised in close to its present form in 1973. But email has so many shortcomings in its very architecture that it's high time we upgraded to something else. There are many partial solutions to the problems listed below, but to properly fix them all requires a ground-up rebuild.

No guarantee of delivery

If an SMTP server swallows your email, tough luck. An email is like a postcard hurled out into the void. If it disappears somehow then no one will ever know.

No support for high-level abstractions like conversations

People do not send emails in isolation. Often, an email will be part of a series of replies, perhaps involving multiple recipiants.

Email gives you no good way of grouping individual messages into a conversation, other than by dumping the entire previous contents of the conversation at the bottom of each message. Gmail does a valiant job of threading emails, but the process it's using doesn't help you if you're not using Gmail, is unreliable and is inherently just a hack.

The lack of any coherent high-level organising principle makes email communication chaotic when the number of messages involved is large. Sometimes this is so unmanagable that it causes individuals to take the drastic step of declaring email bankruptcy, notably including Donald Knuth (founder of literate programming) and Lawrence Lessig (of the Creative Commons and the EFF).

No canonical and independent copy

An email exists in its sender's outbox and its receiver's inbox. It may also be stored by an email server somewhere. If these copies are deleted or lost then it's gone.

If someone tampers with an email that you sent them, you may have no way of proving this to a third party. If someone tampers with your email en route then you have no way of proving this even to the receiver.

There's also no good way to introduce someone into an email conversation they have not been following (you can forward an email containing a bunch of replies, but that's hardly usable). Emails don't have a URL that you can pass around or use as a reference if, for example, the email contains an important decision that needs documenting.

No native encryption

It is possible to encrypt emails. But if you do, then both sender and receiver need to be using email clients that support encryption. The sender would also have to have access to the receiver's public key.

No way of verifying the sender's identity

The only way you know who sent an email is by looking at the 'from' field. If that field is filled out wrongly then there is no way to tell. Impersonating someone over email is technically trivial (unless you use digital signatures, which have the same disadvantages as encryption).

The future

I have high hopes that Google Wave will solve some or all of these problems. But there are there two big advantages email has over Google Wave:
  • It's proven
  • It's widely supported and understood
Email is not going to disappear overnight. After all, fax machines are still reasonably common and faxes have been almost entirely superceded - by emails.

For Google to get wide adoption of Wave they're going to have to come up with a solution that allows incremental adoption. Perhaps the Google Wave client could support email as well as Waves so that I can communicate with the vast majority of my contacts who aren't bleeding-edge adoptors.

But until then we're going to have to suffer the absurdity of disagreements and uncertainty about whether a particular email was sent, who sent it and what was in it - like this Australian political scandal.

Update: The scandalous email has turned out to be a fake.

No comments:

Post a Comment