OPEN KEYNOTE: Eric Hahn, Proofpoint - "The Changing Face of Email"
After three decades of relative calm, the most important application of the internet is exploding with new requirements, new technologies and new challenges. Much-needed and long-overdue improvements in the internet’s core email services are underway. Thought leaders are developing electrifying upgrades to the otherwise stagnant end-user mailbox/message/folder metaphor. Email end points are no longer restricted to personal computers, but now include phones, PDAs, rural campgrounds and airplane seats 5 miles above the earth’s surface. Businesses are redefining email requirements to include more than the reliable delivery of information from user A to user B: they are discovering new ways to understand the semantic content of email and its flow and to relate email to vital business needs. By all accounts, email is enjoying a Renaissance of unexpected proportion. Hahn will offer a brief email history and share some of his views on what the future holds for his favorite “killer app.”
^^^Share your session notes here
^^^History -
- 1971: first email sent
- 1973: email accounts for 75% of all arpanet traffic
- 1978: average length of a message ~250 chars including header, 5 messages per user per day was top bound estimate
- 1978: first spam: DEC customer support ad
^^^Future -
1. Spam: the good guys will win
- spam is growing like a weed (64% 2004)
- but a protected user's spam load is not increasing
- spam getting through to protected user constant or dropping
- Authentication: SPF + CallerID, DomainKeys
- Machine Learning: e.g. support vector machines (SVMs) showing promise
- Novel approaches: computational currency, pennyblack/hashcash: requires sender cpu to solve 10 sec math problem
- Litigation: law suits, convictions
2. Message processing platforms emerge
- messages aren't uniform, users aren't uniform
- message content inspected/modified, routing dynamic, storage dynamic
- common messaging platform emerging to manage app steps: anti-spam, anti-virus, compliance, surveillance, archiving, content filtering
3. Email goes mobile
- 17% users have logged on using wi-fi in last month
- device evolution fast & furious
- bandwidth, latency and airtime matter
- beware of the hype
4. The UX
- Current metaphor designed when users were still receiving 5 messages per day
- Room poll: 50% receive 100+ non-spam messages per day
- New metaphors:
Notes by ross
^^Introduction and History
More changes than at any time in his career. But its important to study how we got here.
Ray Tomlinson in 1971 sent the first email message. No longer recalls what was in the message. Between two DEC PB10 computers, back when the earth was cooling. Seven years later DEC generated the first email based spam message, an ad for arpanet support for their boxes.
Over the next three decades, ups and downs:
- 1973 Email accounts for 73% of arpanet traffic. Two years after its invention it was already the killer app.
- 1978, 250 characters including the header, five messages a day was an expert guess at the max.
- MS Exchange still has X.400 deep in its code.
- 1979 first of many years of the LAN.
- In 1989, email must die because of LAN email. Loathing of the email standard back then.
- In 1973 First standardization of RFC 524/562 by Dave Crocker
- Took ten years for SMTP
- Took another ten years for MIME, to send more than ASCII characters
Why did so much time pass? The answer is it ain't broke, don't fix it.
Owe congratulations to the folks who have served us over the last 30 years.
Its broken now, its not serving us well today.
^^Predictions
^^^Spam: The good Guys Win
Spam has a nasty way of taking over everything, including keynotes. We are already winner. Spam is growing, but a protected user's spam load is not increasing! Scott Ricter is the largest marketer and in theory is all opt-in. What you don't hear enough is that users who have adopted top tier anti-spam solutions are not seeing growth in spam. This is not reported. Those people are no longer victims. Number of absolute messages in a well protected inbox is constant. But many are not protected. Spam control is not like birth control, don't need 100% effectiveness, need great best of breed solutions in most of the net.
More good news:
- Authentication -- (SPF & CallerID, DomainKeys) make a dent in crime
- Machine Learning --
- Computational Currency -- HashCash, Pennyblock systems where if i don't know who you are you have to do some math to get in. 10 seconds of CPU time that adds up to 80 CPU years for a spammer like Scott.
- Legal -- success through lowered expectations. Not a number one tool, but a number 2-3 tool. New suits and prosecutions with results to be had. Should be encouraged.
Defeatable in our lifetimes.
^^^Message Processing Platforms Emerge
Giant achitectural changes in the fabric of email in the next 10 years. Success in 2001 if you can reliably deliver message in a timely and cost-effective manner. An obsolete definition now:
- Messages are not uniform
- Users are no longer uniform. Role can all impact the disposition of a message. Regulation, chinese walls, a very different way for viewing the network and impacts design.
- Mail isn't passive anymore. Every message has to be inspected and business logic must be applied. Application integration important
A world more advanced than the prior definition.
The Sausage Factory Architecture. Passing messages through steps with different management, security, reporting, etc. requirements. An artifact of how we have built incrementally.
Now an emerging better way of building apps. Crack once, run many. Common messaging platform that applications can access. My company is working hard on architectures like this.
^^^Email goes Mobile
Gen Y at 20% penetration. Mobility is obvious. If you don't believe email is going mobile it implies that the laptop is the final form factor. History shows us smaller and more portable with each generation. Every white collar employee will carry these devices. Unique challenges:
- Rate of evolution of these devices. Don't get locked in, keep HD SW seperate.
- Technical depth required.
- Lot of hype, be careful
^^^The Tired "Message-Folder-Address" Metaphor is Upgraded
Shows 1980, 90 and 2004 Inboxes with the same interface. Some say it works, don't fix it. But the analogy is flawed, was designed when people were talking about 5 messages per day. Its an antique. Polls audience, most over 100 non spam messages per day.
Could get 2-5x improvement if we could get out of browsing one message at a time.
Messages, Addresses > Topics Participants> Search
> Hyperstore> Messaging Browser
Why is email still organized around the filing metaphor? Should be able to view life by day by person, see the social fabric. So hard to mine the data. Why is IM so seperate from email? What is the difference between a public email folder and an an RSS feed or blog or wiki?
Great opportunity to unify these metaphors.
Positive examples:
- Bloomba search based email
- Gmail search and threading
- Outlook ad-ons
^^^Closing comments
- Can't believe people talk about the end of email, this is the beginning. What economic impact would there be without email for the economy or your company. We owe much of our economic growth to enhanced productivity, not just email, but its an interesting coincidence that Ray sent his message at the tick of email growth.
Its played a powerful role in our lives. We have a lot of work to do, but so much has been done and by being here you are part of that industry and community and should be proud and should rededicate ourselves to making email better in the next 30 years.
^^^Questions
Time frame for SPF and CallerID impact?
- Immeadiately. By end of this year, rather significant coverage.
- One Phishing attack is enough to convince anybody.
- Domain keys has deeper technical robustness, would like to see it happen, next six month is a time for action.
National do not call list was the fasted product adoption in history -- 52M people signed up in four months for a product that didn't exist.
Group of people who say email is dead that say email is dead because of spam, which is why we should switch to their model, but if we do, we will have spam there.