SUBSCRIBE NOW!

Vol. 12 No. 7 — July/August 2004

Google released news of the service to the media on March 31. John Markoff covered the story for the April 1 edition of The New York Times, and many media outlets began carrying the story overnight. Your correspondent woke up in the middle of the night on the couch, saw the news story in the online edition of the Times, and posted an incredulous announcement to the Web4lib mailing list — and of course to my Weblog.

As news of Google's stunning announcement spread on April 1, many in the media and the public assumed those playful lads in Mountain View were enjoying another April Fool's Day. Internet discussion boards were abuzz with speculation. Google co-founders Larry Page and Sergey Brin wisecracked in the press release, claiming they launched Gmail to help out a single individual who'd complained that managing e-mail is chaotic. Yeah, right, out of hundreds of millions of Google users, they added e-mail to help one guy out?

Now we all know that Gmail is very, very real. This article explores what Gmail is, how well it works, some very serious flaws that Google needs to address, the embedded targeted advertising and other privacy concerns, and the behind-the-scenes global server and storage infrastructure that could let the company skyrocket past competitors.

A World-Class Webmail Client

Let's start with the good stuff. Gmail is awesome. The Webmail client is an amazing piece of engineering. After using it for just a few days, I had already grown fond of it. Now, after a few weeks, it is true love. Gmail engineers have built something that goes far beyond the limitations others thought were inherent in a Webmail client — setting a new standard for highly interactive, intuitive, Web-based clients for other applications as well.

The first thing to note is the simplicity of the interface — just what you'd expect from Google. The screen is about as uncluttered as it could be while still providing all the functions you need — well, most of the functions you need. And note the almost total absence of icons. Figure 1 on page 15 shows a Gmail screen for an account with an empty mailbox.

Gmail uses textual links where other Webmail clients would use HTML submit buttons or icons. One of its few icons is a yellow star that you use to signal you want a particular message "starred" or flagged as a priority item. However, notice that Google doesn't provide an icon for moving an item to the trash — you have to select the "More actions" drop-down and then select the "Trash" option. I've got a conspiracy theory as to why Google makes it hard to trash things; more on that later.

Once you accumulate a number of messages in your mailbox, you'll quickly appreciate the search feature. Gmail indexes your mailbox to let you search it as rapidly as Google Web searches handle the Web. The philosophy is completely different than the familiar approach of filing your mail into folders as you process new items in your in-box. Suffice it to say: The search is as fast as you'd expect from Google. Rival Webmail providers either don't have a search function or performance is painfully slow.

Other cool features of Gmail's Web client:

• You can turn on a feature that provides keyboard shortcuts. Want to type a new message? Just hit the letter C. It's kind of like using an old VT-100 mail program such as Pine or Elm from inside a Webmail environment.
• As you type in the addresses to the people you wish to e-mail, the client does very fast auto-fill-in of names or addresses it recognizes from previous mailings. When you've typed enough to uniquely identify a correspondent, just hit Enter and the whole address completes.
• Another feature grabs the first few words after the Subject line and displays them, peeking inside the message a bit, as Outlook can do.
• You know how some e-mail exchanges go back and forth like a conversation? Gmail groups those exchanges into a "conversation" and displays them as a single message. You can easily scan through the conversational history.

If you've used other Webmail clients — and I've used virtually all of 'em — the other thing that strikes you about Gmail is its speed. This sucker is fast. With most Webmail clients, a simple transaction like deleting one message can take forever. You click the "Trash" icon and you wait, wait, wait for the screen to repaint. With Gmail, the screen doesn't repaint — it sort of magically adjusts — almost as if you were using a local client like Outlook.

Similarly, the Gmail Web client quietly polls for new messages periodically and adjusts the list of messages to show the new ones without repainting the entire screen.

It appears to me that Gmail has invented a protocol that carries on background conversations between your computer and the Gmail mother ship — just as a "real" client like Eudora or Outlook would do. The local client communicates with the Gmail servers over a much more efficient channel than the relatively cumbersome process of sending up HTML form transactions and painting an entirely new screen after each. It's as if you had the advantages of a local mail client talking the IMAP protocol — but it works from any Web browser using JavaScript.

How does Google do it? Rumor has it that Google likes to have development teams with as few as three people. Grab three really bright programmers, lock them in a basement with video games and Jolt Cola, and see what comes out in a few months. I'm betting that one of the folks in the Gmail trio is the world's best JavaScript programmer, the second is great on protocols, and the third is superb on CSS and Iframes.

Shortcomings and Omissions

In case you're starting to think Google handed me tulip bulbs for their Dutch Auction IPO, let me note some shortcomings of Gmail.

Google wants you to fill that 1-gig Gmail box just as full as you'd like. That's all well and good in theory — hey, it's their disk budget — but someday I may want to get all that stuff back. Gmail offers no POP support — Google is reportedly thinking about it.

When I do fill up that 1-gig mailbox, I'll need to clean it out. Gmail doesn't offer a way to sort your mail by size. Its rivals do. If your mailbox is full, the rational way to go about cleaning things up is to sort the largest items to the top and delete the ones you no longer need. The Gmail folks seem to expect — somewhat naively — that 1 gigabyte will suffice for all time. These days, a lot of us are working with 5 megabyte PowerPoint files or 5 megapixel digital photos. At the rate I'm filling my Gmail account, it'll be full in about a year. When that year ends, I'll demand a way to efficiently download all my stuff and archive it to a burnable DVD.

Other shortcomings:

• The spam detector is not as strong as rivals such as Netaddress.com. It lets an awful lot of spam through, and I've had some false positives that really should not have occurred.
• The on-screen layout has some quirks. For instance, a longish Subject line slides action links around in an unexpected way.
• Speaking of Subject lines, you don't see the Subject when replying to a message unless you click on Edit Subject.
• There's no way to empty the entire Trash folder or the entire Spam folder. That's basic on rival services.
• Gmail has no feature for auto-forwarding mail to other accounts. (Remember: It really wants your mail on their servers.)
• There is no automated "signature" feature — a basic shortcoming.
• There is no vacation reply feature — another basic shortcoming.
• Gmail doesn't work on Macs using Safari, the new native browser for the Mac. And when I tried it on a Mac running Internet Explorer, it still didn't work.
• The search feature, while fast, is not the Google search engine. A company spokesperson said that the IR (information retrieval) challenge presented by Gmail was trivial compared to that of the global Web search. One specific impact is the lack of spelling suggestions for near-match queries — a tremendously useful feature in Google Web searching.

Gmail Hates Folders

The Gmail tagline is, "Search, don't sort." Google says: "Gmail uses Google search technology to find messages so users don't have to create folders and file their individual e-mails." A Webmail quota of a gigabyte could let you keep years of important e-mails in one place. Combine that with the power of Google searches and you've finally got an efficient way to find that important contract negotiation from 3 years ago.

Gmail takes the "search, don't sort" mantra to an extreme. It not only says you don't need folders, you aren't even allowed to create any folders!

With Gmail, Google has gone out of its way to avoid the folder metaphor. Gmail doesn't want you to do any work as you process new mail. Read it, reply to it, react to it, archive it if you want — but don't spend time trying to figure out where to file it. Yahoo! (and conventional mail services) more or less force you to take a little effort to create a filing system and move incoming mail into folders; Gmail encourages you to archive everything in one flat All Mail folder and just search when you need old stuff.

Gmail follows the patterns of Yahoo! and Google's disparate approaches to organizing information: Yahoo! Mail is to Yahoo! as Gmail is to Google. Yahoo! was born as a hierarchically organized filing system for Web sites; Google Web search is famously pure searching of a giant, flat Webspace. The Yahoo! model requires human filing effort — it takes a team of catalogers to put millions of Web sites in the appropriate spot in the hierarchy. With Gmail or with Google Web search, robots do all the work.

Even though Gmail doesn't want you to organize your e-mail into folders, you may still want to identify categories of messages in specific ways. In lieu of folders, Gmail lets you apply metadata to a message in the form of "Labels." Gmail does have a handful of pre-defined folders — Inbox, All, Sent, Spam, Trash — but you can't create any folders of your own. (By the way, those last two folders — Spam and Trash — have no universal delete function, unlike every other known Webmail service.) It's almost as if Gmail has a religious conviction that filing into folders is evil.

But folders aren't evil. A hierarchical file system remains a powerful and useful option. For instance, I need to keep up with a couple thousand digital photos from the last several years. On my hard drive, I organize by the dates the photos were taken — \2004\03, \2004\04, etc. When I upload to a service such as Sony's Imagestation, I create topical photo albums, drawing from my chronologically organized folders. Gmail would have me simulate this with a flat system of labels.

Searching and folders are not antithetical; there's no reason why Gmail couldn't allow the user to create folders. You've already got the pre-defined Spam and Trash folders; why preclude others as the user wishes? Rival services certainly allow this. For my mailbox at usa.net, I have a folder labeled Northwest, where I file all my airline receipts. I'm not very diligent about filing into the Northwest folder as ticket receipts arrive, so periodically I use the search feature to find recent receipts, and move those messages into the Northwest folder. Gmail forces me to do this with labels, claiming this is better than folders, because the same receipt can appear labeled in more than one place (e.g., "Northwest" and "Business Travel"). Why not give me both? Let me put all of my Northwest stuff in my Northwest folder and let me label it to boot.

Folders are as important for what you don't put into them as for what you do. If you find a folder labeled "1985 Taxes" and know you are a careful filer, you feel pretty confident you can throw the entire folder away when no longer needed. I would be less confident to search an unorganized database, pull up a bunch of items on a hit list labeled "1985 Taxes," and delete all the items without inspection. I know how carefully I file things; I don't know when an unstructured search will pull up some false matches.

As readers of Searcher well understand, sometimes we put lots of effort into building good filing systems. The classic example is the Library of Congress classification and cataloging scheme used by major research libraries. This scheme works great in many ways, but the cost to catalog a book usually exceeds the cost of the book. Other online information companies put a lot of work into cleaning data and applying good metadata before content is tossed into the database.

Do individuals want to put the same kind of effort into filing their e-mail as LexisNexis or Gale would put into online professional databases? Most of us probably wouldn't. Yet some people expend a lot of energy doing just that. You may not have heard of Nathaniel Borenstein, but you've benefited from his work: He invented MIME, the mechanism used for sending and receiving e-mail attachments. Here's his strategy for coping with mail:

My incoming mail goes through several steps:

1. Heavy pre-sorting by spam filters and explicit rules sorting e-mail into about 30 different "inbox" folders.

2. Ruthless deletion of stuff I know I won't need.

3. Explicit filing of "keeper" messages in one of over 300 archival folders.

Notice that this strategy does *not* lead to the "empty inbox" so many people seem to believe in. In fact, some of my 30 inboxes can get totally out of control while I keep up with others reasonably well.

Most of us won't devote so much energy to e-mail storage. So Google tells us that in managing our e-mail, there has to be a simpler path. The question is: How much effort I'm willing to put in up front. As a packrat surrounded by eternal mess, my life is living proof that I cannot organize anything. So for me, Gmail's model is perfect.

A colleague, Mike Zakhem, concurs: "The idea of indexing your messages is a new and very Google-centric thing that Gmail has brought to the e-mail game. A lot of e-mail clients will let you search for a message, but often those searches are tedious and time-consuming (e.g., what folder should I look in? Should I look by sender address or sender name? etc.). Google has leveraged its technology to do for its e-mail what it does for its Web searches.

In general, the world is moving away from simple file systems that can only put an object (a file) on one place in a hierarchy. Many institutions use content management systems or other database back-ends to replace the simple UNIX-file-system hierarchy for hosting Web content. The file system that Microsoft is creating for Longhorn is basically an SQL database management system. These systems give us great flexibility in managing information and metadata. But great flexibility could imply a great pre-processing burden. The question is, for everyday streams of random data, such as e-mail, will folks put in enough metadata as new items arrive — or will a smart post-processing search engine suffice?

My friend Ed Vielmetti, who for many years has worked to harness the power of the Internet to help people share, describes Gmail's notion of organization this way:

In the knowledge management world, Google's Gmail seems to support a "big heap of laundry" approach to storing your memories. It does have category tags you can apply to things, but just imagining how many categories you might apply over 1G of mail makes me blanch.

Privacy Part One: Fear of Robots

As soon as the world learned that Gmail's business model called for ads to appear in your message display — ads with subject matter determined by the content of your message — the privacy community went apoplectic. They simply could not countenance the notion of a robot reading through your every message and serving up ad content based on the words of your correspondence.

After much global hand-wringing, it appears the world sorts into these categories:

• Those who will never use Gmail because they don't trust a robot to keep their secrets.
• Those who will never knowingly send to a Gmail user because they don't trust a robot to keep their secrets.
• Those who are willing to trust the robot in exchange for a highly functional Webmail client and 1 gigabyte of storage space.

I fall into the last category. The robot reading my mail matters not one whit to me. Robots read my mail every day. Every mail service provider worth its salt offers an anti-spam service. That's a robot reading your mail. Most good service providers also offer anti-virus screening as well. And that's another robot reading your mail.

What seems to stick in some people's craw is the idea that the robot reading your mail is trying to understand it at the same time. It's scanning the content and tailoring the ads to match your discussion. Of course, Google the Web search service takes the words you type and pulls ads out of its inventory based on that content — but apparently it seems much more personal when you've logged in under your own identity.

The targeted ads are unobtrusive, about like the targeted ads that appear when you do Google searches. Usually ads relate to the subject matter of the message reasonably well. Figure 2 below shows what a typical ad looks like.

I'm not sure how prominent information architecture guru Lou Rosenfeld feels about being matched up with vocabulary and juggling trainers, but you get the idea. Often the ads can be useful; if you get a message that mentions Lincoln, Nebraska, you'll get tourist and civic ad links.

Sometimes the targeting goes hilariously astray; consider this e-mail from Amazon to one of its customers. As you see in Figure 3 above, an ad for Seattle comes up!

The Gmail ad program isn't very worrisome if you think of things one message at a time. The privacy community may have a point if you consider the scope of the scanning across entire mailboxes. Gmail actually chides you if you let your mailbox become empty. It makes it hard to delete messages with no convenient Trash icon or button. There's not even a way to delete all the messages in your Spam folder.

Put this all together, and it's clear Google doesn't want you to empty out your mailbox; it wants you to leave a lot of mail lying around. The obvious conclusion: Google is data mining all your mail — not just the mail that you're reading right now or the mail that just arrived. (See Figure 4 above.)

Now perhaps that just means Google is tuning and re-tuning the ad-matching algorithms. But maybe Google's looking across customers' mailboxes constantly as part of the tuning process. Maybe some sort of network analysis is being done of the kinds of things you and your correspondents talk about — and which ad links you click on individually and collectively. The more the Google robot co-mingles data as it analyzes mail content, the more squeamish some people will become.

Privacy Part Two: A Gig of Your Life Is a Big Target

Suppose someone guessed your Hotmail password. That could lead to embarrassment, but, since you can only store so much information in a few megabytes, your exposure is limited. If you're an active e-mail user and you send and receive attachments, your exposure might be only a few weeks.

Now let's imagine you're a Gmail user and someone guesses your password. Your exposure could be years of your life. The target is bigger, as is the payoff for accessing the target.

But guessing your password is only one way someone might gain access to your Gmail. One obvious point of attack is a hacker. I asked Wayne Rosing, vice president of engineering for Google, about the threat of such a large target. Rosing said that Google has a very capable team of security experts guarding the perimeter: "We're committed to upholding the absolute highest levels of privacy. It's not really an issue of how big the mailbox is; it's a question of how good your security is." I asked if Gmail encrypts customer mail just in case someone gets past the perimeter. He said no, "encryption is computationally pricey."

The bottom line is that if a hacker did break into the perimeter and could access Gmail servers, all customer e-mail would be in clear text.

The privacy community probably worries less about hackers breaking in than in the government using the Patriot Act to get in without knocking the doors down. Hmm, access to years of a person's life — combined with Google quality search as well. You can imagine how appealing this might be to a government pursuing suspected terrorists — or tax evaders.

Google's Gmail privacy statement is superficially reassuring, but it says Google must comply with "government" demands for access. But which government? I asked Rosing if my Gmail mailbox might be stored on a server in China, where the Patriot Act pales. He indicated that Google wouldn't store data in nations where it would be subject to government invasion and that in any event Google has no data centers in China.

Here again, I think the privacy community has a valid concern. Before Gmail, your personal mail was scattered across many places, with your most personal information probably residing on a hard drive under your control. Gmail invites millions of people to move their personal discussions to a common pool — and to leave large amounts online. I claim that Gmail will cause the largest migration of personal information in the history of the world.

If Google has a mission statement, it is its famous dictum of "do no evil." When I talk to Google officials, I sense they really believe in that creed. However, I think perhaps they are a tad naïve, not realizing that external forces may cause them to invade privacy in ways they've never imagined.

Privacy Part Three: Flaws in Gmail Are the Real Privacy Threat

Once I drove up to an ATM and was surprised to see on the screen, "Another Transaction?"

A little slow on the uptake, I took a second to realize that the car pulling away had left their card in the machine in mid-transaction. Hmm.... Tempting.... Gmail has a couple of related design flaws that ATM designers anticipated decades ago.

Scenario one: You log into a public terminal — say in a public library or a cybercafé. You forget to click the "Sign Out" button before you get up and leave.

The next person who walks up to that terminal now has your entire life at their fingertips — years of e-mail, efficiently searchable. Pretend you are a detective or a snoop, and think about some of the searches you might type in if you sat down and saw someone else's entire life history in the browser window. Your life history remains exposed as long as that computer is turned on and the Web browser remains open. The session never expires.

Scenario two: You visit your boss or a colleague and quickly check your Gmail for that report you mailed last week. You forget to sign out. When you go back into your own office, you sign back into Gmail. Everything seems perfectly fine. But the other Gmail session is still active. In fact, it's automatically updating your Inbox listing. Your boss can read your new mail, search your old mail, or even send new messages from your mailbox — indefinitely. You have no way to close his view into your life and, if he doesn't send or delete, you have no way to detect it.

Gmail's rivals figured out these exposures eons ago and implemented session timeouts and multiple login detections as remedies. While the privacy community wails over a robot serving up relevant ads, they're missing the entire point. (See Figure 5 below.)

These flaws shouldn't be hard to fix. Because Gmail can expose years of your life, Gmail should "time out" aggressively — maybe after only 15 minutes of inactivity. If you time out, Gmail should prompt for your ID and password before resuming your session. (It should not log you out and send you back to a fresh login.)

It should be easy for Gmail to detect when you log into a second session. How should the Gmail system react? At a minimum, it should log out the first session — the one you left logged in for the boss to read.

What surprises me is that Google didn't "get it" and address these issues long before the public beta. My guess is that it took such a blank slate approach that it blanked out on the public terminal issue.

The other surprise is that privacy activists jumped all over other issues without seeing the obviousness of these flaws.

The Viral Marketing Campaign

Besides the curious tactic of a confusing April 1 launch, Google also chose to introduce Gmail to the public using a new tactic. Instead of handing out accounts to reporters from the tech media and national papers, Google gave "tokens" to employees that they could use to invite friends to join. Later, beta testers might log in to see an invitation to invite their friends to join the beta.

A fascinating experiment in social networking ensued. People with Gmail accounts began sending mail to their friends, who in turn wanted in on the beta. Some people began selling their "tokens" on eBay.

In effect, Google's slow rollout became a viral marketing strategy. The media wrote about Gmail, but only a few people had accounts. Those who did bragged to friends and posted screen shots on their blogs. Welcome to the Gmail Witch Project. In terms of creating buzz, it was brilliant.

However, the strategy may not have been universally brilliant. If Google had given Gmail accounts to the Katie Hafners and Walter Mossbergs of the world up front and let them see how the targeted advertising works, the initial furor over privacy might've been muted significantly.

There's another issue, one tough nut that Google will have to crack: How do you prevent someone from signing up for 50 Gmail accounts in order to get 50 gigabytes of free online storage? That's always been possible with Hotmail and Yahoo! mail, but it's pretty hard to stitch together a serious free data store at a pace of 4 megs at a time. At 1 gig per shot, the stakes change. So how is Google going to limit this activity? I can't think of a reasonable way for a free service to keep people from setting up multiple identities.

Google's Global Infrastructure Advantages

The economics of Gmail just seem impossible to data administrators worldwide — especially those who administer e-mail services. After the announcement proved to be true, I began asking tech types a simple question: What is the cost of a gigabyte of reliable, backed-up, broadband-connected-to-the Internet storage?

I was surprised that I couldn't find a ready answer from the folks who I thought would know. Dr. Charles Severance, my globe-trotting research professor friend at the University of Michigan, had no ready answer, but a day later came up with a figure of $1 to $2 per gig per year. Hmmm. So if Google signs up 20,000,000 Gmail users, it will need to cover $20M to $40M per year for storage alone. Or to put it another way, Google needs 1-2 per year per subscriber to cover storage costs. That's certainly easy to imagine. Not everyone who signs up will fill the quota.

One person who manages the campus e-mail service for a large Midwestern university simply could not believe the Gmail announcement. Although disk is cheap and storage costs have been plummeting for 2 decades, there are many more costs than just disk. "There's no way they can afford to back up that much data!" he exclaimed. "The costs of tapes and tape drives and personnel will kill them!"

Ahh, but that assumes that Google is backing things up to tape. Google famously uses low-cost commodity servers and the cheapest disks that money can buy. (One rumor has it that in the early days Google assembled its own Linux PC servers from commodity parts and didn't even bother to put the covers on the components. Suppose Google's thinking differently? Suppose it's backing up not to tape, but to disk? As e-mail administrators worldwide started doing back-of-the-envelope calculations, other folks began theorizing as to what Google's global infrastructure is really like.

A few years ago Google admitted that it had over 10,000 servers in production. The Web has grown since then and so has Google's dominance. How many servers does Google have online in mid-2004? Some smart folks took Google's SEC filings, assumed minimum cost servers, and calculate that Google has about 100,000 servers in production worldwide.

It's not just the raw count of servers, of course. An intriguing posting to Dave Farber's "Interesting People" mailing list suggests that Google has built a global infrastructure than no one can match. Suresh Ramasubramanian suggested what Google might have built:

Google has taken the last 10 years of systems software research out of university labs and built their own proprietary, production quality system. What is this platform that Google is building? It's a distributed computing platform that can manage Web-scale data sets on 100,000-node server clusters. It includes a petabyte, distributed, fault-tolerant file system, distributed RPC code, probably network shared memory and process migration. And a datacenter management system which lets a handful of ops engineers effectively run 100,000 servers. Any of these projects could be the sole focus of a start-up....

This computer is running the world's top search engine, a social networking service, a shopping price comparison engine, a new e-mail service, and a local search/yellow pages engine. What will they do next with the world's biggest computer and most advanced operating system?

Tech visionary George Gilder told me: "Google is exploiting the key abundances of the era: bandwidth and storage, summed up in my model as 'Storewidth,' in order to supply what is scarce: Just-in-time information. Google is the prime Storewidth company." Gilder calculates the cost of storage at about $2.33 per gigabyte per year, including depreciation and maintenance. But he thinks Google enjoys other advantages:

Since Google must sustain these costs anyway to support its search capacity, advertising model, and news services, I believe that their marginal cost for supplying e-mail is close to zero when the increasing volume of usage of all services is considered. Market share and volume are everything in these front-loaded Internet services. With more numbers and better targeted advertising, Google will make out like bandits, without the downside of encountering Wyatt Earp at the FTC corral.

How Is the Competition Reacting?

If your competition is indeed running the world's biggest computer and most advanced operating system, how do you react? How do you react when the competition offers mailboxes 100 to 200 times larger than what you brag about?

Figure 6 at right shows how Yahoo!, weeks after Gmail was announced, is still bragging about a paltry 4 megabyte mailbox:

Any company selling a rival Webmail service must be sweating. I repeatedly asked Netaddress.com, which sells Webmail accounts with 50 megabytes of storage for about a buck per meg per year, how it reacted; Netaddress.com did not reply.

Other companies that might be affected include vendors of desktop e-mail client software. Gmail is so good that I can imagine many people abandoning their Eudora or their Outlook. (In corporate environments, however, things may differ; I suspect the network security manager for Ford Motor Company would not be too thrilled to have company discussions housed on Google's servers.)

Vendors for add-on software may also be affected. Caelo, a software vendor based in British Columbia, releases an add-on called NEO that provides a fast-indexed search of your Outlook folders. A spokesperson for the company said that Gmail may "legitimize the idea of fast searching, so that people will demand fast searching inside Outlook. Our sales may increase because of Gmail."

Here We Go Again

Whether you're a potential Gmail user, a rival Webmail vendor, a university e-mail administrator, or otherwise, you can't ignore Gmail. Clearly, a new game is on.

GROOVY GMAIL TRICKS

Many folks use their e-mail to manage their lives, keeping up with what they need to work on, deleting items as tasks are completed. Gmail could help people organize their lives in a variety of ways:

• Chuck Severance works with developers at Stanford, MIT, Indiana, Michigan, and elsewhere, building a collaboration environment called Sakai. E-mail plays a vital role in a multi-institution development project. Chuck prefers his familiar mail client software on his Mac laptop, but he finds Gmail's ability to search old mail and to group related conversations into a single thread very useful. So he reads his mail using conventional tools, but he auto-forwards a copy of all e-mail to his Gmail account. When he needs to dig up a past thread, a quick login to Gmail and a quick search bring up the relevant conversation quickly. Gmail also lets him log in quickly over the Web if he's away from his familiar client software. (Chuck's suggestion for the Gmail developers: Invent MailRank, analogous to the famous PageRank algorithm that forms the core of Google's success, e.g., "Put e-mail from the boss at the top of the hit list.")
• If you have multiple personas — day job, consulting work, family life — you might be tempted to use a separate mailbox for each persona. But Gmail's labeling, threading, and searching capabilities could help you organize it all in one mailbox.
• The threaded conversation grouping is also useful for taming a chatty mailing list. From time to time I join Sigia-L, a listserv for information architects, teeming with the output of an unruly and chatty lot. Collapsing 30 postings on a single topic into one item in the inbox is tremendously helpful.
• People such as authors, consultants, and speakers may long for the services of a rich corporate IT environment. But Gmail could offer a poor man's document management system: Leave old PowerPoint files, project plans, even contracts in your Gmail archive, using labeling and searching to fetch what you need on demand.
• Searcher's esteemed editor, Barbara Quint, had another suggestion for Gmail, but it required switching Web search engines from Google itself to Amazon's new A9 Web search engine. Don't feel bereft. Google supplies the searching in A9. Here's what bq had to say in a Newsbreak entitled "Amazon Introduces New Web Search Engine" [https://www.infotoday.com/newsbreaks/nb040426-2.shtml]:

Most of the reviewers of the new site have praised it for its appeal to end-user searchers, but pointed out that, at this point, it would have limited appeal to power searchers. On the other hand, it might have some appeal to intermediary searchers working with clients and doing extensive Web searching. Since A9 holds search histories centrally and allows toolbar users to annotate result lists, one could envision professional searchers using the service as a collaborative tool. Use A9 for searching and annotating search results. Share the registration password with clients or colleagues working in a collaborative mode. That circle of associates can search and annotate their own results, or just expand the annotations on entries provided by others. Of course, one would recommend that the user identification tying all this together not be connected to anyone's credit card, as it might if based on an Amazon.com registration. However, it should be simple to set up an e-mail identity. In fact, if one used Google's new Gmail, one could add a gigabyte of storage to the project's resources.

And all for free. God bless the Web.