Tom Chi  

Gmail in Detail

April 2nd, 2004 by Tom Chi :: see related comic

So there was a lot of buzz yesterday around Gmail: the gigabyte web-based email service announced by Google on April 1st. While other Google annoucements that day were met with an appropriate amount of skepticism (the Lunar Base for example), Gmail seems to have caught many a techie’s imagination. The main question is: could it be real?Let’s quickly review what Gmail is purported to be: it is a free web-based email service offering an unprecedented 1GB of storage that is designed for quick searching. The press release also suggests that users need never delete email again.

Now let’s consider the plausibility of such a service. Clearly the searching aspect of it is well within Google’s domain, and it has been rumored for months that Google would be launching a new email service. What many people have pointed to as the “dead giveaway” is the 1GB of storage. Compared to the modest 4MB that Yahoo Mail offers, or the incredibly measly 2MB Hotmail, 1GB just sounds like a joke. That’s just way too much, right??? — especially in a world where people still pay hundreds of dollars for 1GB of web-hosting. Right?

To answer the question, we need to gauge the economic and technical feasibility of such a system. First economic. In terms of the up-front capital costs (of acquiring the HW for such a storage system), 1GB costs ~$1 USD. That’s not very much money for Google. Even if 100 million people signed up, it would still be well within their power to cover the cost of the drives. Also take into account that they need not have all the storage right away. It takes a couple years to get 1GB of email, so they could spread the acquisition cost over that time. Now for replacement costs. The mean time to failure of a 200GB drive is around 100 years. At most, it takes 50,000 drives to cover 100 million users. With 50,000 drives and and a MTTF of 100 years, we’re talking about replacing about 1 a day - not bad. If you factor in the cost of setting up the framework for periodic backups, the servers etc, we are talking maybe 2-5 times the capital cost.

Now how much would they stand to make if 100 million people sign up? Well most free email services survive off of ad revenue, and Gmail has a distinct advantage over the others because Google has already built a great infrastructure for serving targeted ads based on spidered content. So here’s a scenario:

Imagine that you are having a conversation in email about a car that you are thinking about buying. In the next mail you get, Google has placed an advertisement for that car at a great price from a dealership only 20 miles from your home. By leveraging several of their initiatives: adwords, geography-based search, and froogle, they could offer exactly that. Now that is the sort of targeted advertising which is worth a lot to companies. Most would pay considerably more than $1 to have such ads served to consumers at the right time. Afterall, businesses both big and small toss away 30 cents per person every time they send a junk snail-mail. Direct mail campaigns usually have a conversion rate of less than 3%, but are still one of the most successful ways to adverstise today.

If people come to rely on Google mail, they might be exposed to 20 such ads in a given day - thereby generating quite a bit of advertising revenue for the company. This easily outweighs the cost of the hardware, and even if the cost of maintaining that hardware was 10 times the cost of acquisition, they would still be turning a profit.

Now for technical feasibility. Wouldn’t 1GB gobble up too much bandwidth? Well, first off, the advent a new mail service with more storage does not mean users will send proportionately more mails than they do today. There is definitely a limit to the number of emails a person can send and/or read in a day, and that “human bandwidth” will not go up significantly. The only danger is if users start kicking around 100MB attachments, but this is easily solved by putting a cap of 10MB on attachment size. So given this, Gmail might not take any more bandwidth than Yahoo mail does in a given day. There may also be abuses, such as users signing up for numerous accounts, or using their email accounts only as data stores. However, this would not be too difficult to discern from real email activity, and Google could just place a line in their terms and conditions saying that accounts that are used as such will be terminated at their leisure.

So to sum up, while it sounds really quite ridiculous, Gmail is definitely within the realm of economic and technical possibility. Will this be the great tech hoax of 2004 or the start of a new era in web-based mail? I guess we’ll all find out soon enough.

16 Responses to “Gmail in Detail”
Tom Chi wrote:

Woah. I just found this AP wire article on Salon: here

I guessed correctly about everything… leveraging the ad tech, the 10MB limit on attachments, etc. Sheesh.

Tom Chi wrote:

Just realized that not everyone can read that Salon link. This Miami Herald article has some more detail.

Bob Salmon wrote:

How long before Google uses retina scans a la Minority Report to target individuals as they’re walking along the street?

Maybe NSA in America / GCHQ in the UK should enlist Google so that when you make a phone conversation you get a voice in your ear with ads relevant to your conversation (and the authorities record your every word)?

Or Google could team up with toilet makers and health care product manufacturers to analyse what you flush away for indicators of illness and then advertise the cures by printing it on the toilet paper?

The possibilities are endless…

Mike Woodhouse wrote:

I’m not sure whether I perceive this news as a Good Thing or the opposite. I do suspect that we may be seeing the dawning of a new, er… is paradigm too strong?

If Google gets it “right” - by which I mean that the ads aren’t too intrusive, the spam filters are effective and uptime has enough nines - then we may soon see a time when we’re asked for our GMail username rather than an email address.

And Google have a pretty good record of getting it “right” thus far.

I’m curious to see how, in particular, Microsoft and Yahoo! respond.

Just in case GMail becomes the ubiquitous communications medium that is clearly possible, I’ve already registered my interest…

Tom Chi wrote:

From a New York Times article:

“One internal Google study put the operational cost of maintaining electronic mail storage at less than $2 per gigabyte.”

Man, I’m not too bad at this estimating thing. Oh, for those who don’t know, you can get into NYT articles by using ‘imaguest/imaguest’ for user/pass.

As for the privacy issues, I’m sort of split on what this will mean. After all, people at Yahoo or MSN *can* read your mail if need be. So nothing is truly private. Perhaps Gmail is better because it always reminding you that your communication is not private. Of course, the one scary thing is that your Gmail account provides a unique ID which allows them to aggregate a huge amount of data about you, and the data has clearly been mined and is being looked at in order to price ads for prospective advertisers.

Maybe if they set up an auditing area where the system displays everything they know about you, and you are able to delete whatever you find undesirable. That would be very honest and full disclosure, but managing it would just be another source of stress.

J wrote:

We’re all such geeks, our hearts go all aflutter with talk of 1GB free e-mail. Sorry, I just had to take a step back for a second to gain some perspective of whats being discussed.

Mike P wrote:

Well, since e-mail accounts, pop accounts to be specific, are the main culprit for leaking userID and passwords to the network access I don’t think you can consider it “secure” per say. Anyone on a high speed cable network is basically broadcasting their userID and password to everyone on the same node every time e-mail has been checked.

So, I’m not too worried about the privacy issues but as Tom has said it, if they started compiling the information about me, that’s another story. I believe the new Canadian privacy law addresses this very issue but I don’t know how useful that would be on the net.

Regardless of all this issues, I want a Gmail account.. don’t you?

TheWorldTraveller wrote:

So I just went for a job interview with Google today and I talked to people in Mountain View about GMail: It is definitly real, not an April Fools and Google employees already have accounts. The cute girl on the front desk goes
as “Gingerly” ;-)

One of my interviewers told me that apparently the announcement around April Fools day was on purpose - its gets people talking and helps
to create even more hype. I am not sure if thats
true or just a cover-up of a little PR/marketing screwup. Having read the Tipping Point of course, it probably was on purpose and the perfect timing to get people talking …

Apparently the idea of gmail is to use search as the primary means of accessing and organizing your email - and not a folder-based structure… - it’s gonna be interesting to see what it will really look like …

Audiophile wrote:

Hi Tom, congrats again on your SXSW victory. It was great to see you and KC again. wrt MTTF, I’m not sure where you got you figure of 100 years, but MTBF calculations usually include replacement at the end of the component life cycle. For Western Digital Hard Drives, it is currently 5 years. Link to Western Digital

KC wrote:

J said:

We’re all such geeks, our hearts go all aflutter with talk of 1GB free e-mail. Sorry, I just had to take a step back for a second to gain some perspective of whats being discussed.

Agreed. 1Gb storage alone is interesting but not all that exciting for geeks with their own POP accounts. I think understanding how Google plans to sustain such a service and how the search based e-mail could benefit or hurt the e-mail experience is worth discussing, however.

The search based e-mail has implications both from an advertising/privacy perspective as well as an overall user experience/e-mail organization perspective. I don’t personally care about the 1Gb but I’ll certainly sign up for an account to see how it will differ, if at all, from hotmail/yahoo.

Che Tamahori wrote:

When calculating the cost of providing a Gb of email, assume that Google needs less than 1 Gb of disk space.

Many full-text search algorithms work by storing an exhaustive index of where every word in a document appears. This is intrinsically a compression algorithm! Keep the index, and throw away the original… then rebuild the orginal from the index when needed.

If Google uses this approach, text-only messages would be stored in a hugely compressed form.
Many people will have lots of uncompressible attachments; but on average, those millions of inboxes should compress surprisingly well…

Does the idea of a huge omni-available inbox combined with a genuinely powerful search engine remind anyone else of LifeStreams? Or am I the only one?

Neo210 wrote:

Hey guys. I noticed all the talk about the 1 gig email storage. Well I found a free email server that gives 1 gig email and 40 gig server space. I’ll get back to you after I finish backing up and get the link back. P.S. It’s in a forign language but there is a translate into english button. :D

David Heller wrote:

hi Folks,
I have a gMail beta account. LOVE IT! Wish they would do what is necessary so I can use it for everything. I’d even love some way while I’m online to sync to my desktop so that I can reduce storage requirements on there but still get the ads. yes, I want the ads. Their great! they are meaningful and I love ‘em.
But the ads and the storage is really the spin here. The details of this puppy is in the interface and the new paradigm of labeling (facets) instead of sorting into folders. Its wonderful and speaks volumes to how far we have come w/ Information Architecture and user-centered thinking when it comes to this stuff.
Other features that I like are the single character short cuts, and the conversation threading (w/ lots of expand/collapse). I have started reading almost all of my e-mail lists through it now and it makes such a big difference in how I use my e-mail.
There are still some major quirks but the overall changes in specific paradigms were some of the easiest I had to adjust to before. I’m usually really slow to paradigm changes, and this took me about 2-3 days to get used to. It probably helped that I was getting something out of it.
As for the 1gb storage. I’ve been using it for 1-2 weeks now and I alraedy have 6MB of e-mail. At this rate of growth I think I will hit my 1gb mark in 5 years or something like that. that is not forever folks. This is why I think in the next 5 years (-^) I’ll need some way to use the storage on my desktop and use the Google “search” piece as a connected web service.
I have a list of things I want in GMail, but I was told for the first few revs they just want the basics down first. For me my list are the basics:
1. alert when I get new mail. Right now I get this through my IM clients w/ other services.
2. palm/phone version of service.
3. addressbook (though the memory tooltip reminders are awesome).
4. I know there was a 4 somewhere … Oh!!! ability to connect to imap and pop sources for other e-mail. I mean if this is going to be my new repository … I need it to repose. ;)

— dave

Andrei Sedelnikov wrote:

The details of this puppy is in the interface and the new paradigm of labeling (facets) instead of sorting into folders. Its wonderful and speaks volumes to how far we have come w/ Information Architecture and user-centered thinking when it comes to this stuff

Wow! That multiple categorization instead of storage model for email was know for long years and I was constantly waiting, which e-mail software implements it first. I did not expect that it will be web-based mail! Now having even only this feature, Gmail will succeed, I’m 200%sure.

David Heller wrote:

Outlook has had “categories” for years. It was not managed yearly as well as this, but it is still applicable functionality that allows for non-heirarchical faceting.

Andrei Sedelnikov wrote:

Outlook has had “categories” for years.

Yea, but why none has used it ever? I think because the underlying model is still folder-based so one must care about where to store the message.

And there is also the famous “if the user cannot fin the feature, it does not exist”. :)


Leave a Reply


OK/Cancel is a comic strip collaboration co-written and co-illustrated by Kevin Cheng and Tom Chi. Our subject matter focuses on interfaces, good and bad and the people behind the industry of building interfaces - usability specialists, interaction designers, human-computer interaction (HCI) experts, industrial designers, etc. (Who Links Here) ?