If you weren’t creeped out by data-mining startup Rapleaf after reading about their ways in a relatively unsettling Wall Street Journal article published last October (“The San Francisco startup says it has 1 billion e-mail addresses in its database”), chances are you will be now.

For its latest ‘study’, Rapleaf has tapped its database of identifiable information to extract a sample of 6,000 Google employees (email addresses ending in @google.com) and 16,000 Microsoft employees (email addresses ending in @microsoft.com) and subsequently analyzed their grocery purchase behavior in partnership with an unnamed loyalty cards aggregator.

Rapleaf anonymized the data, so the study doesn’t reveal any names of Google or Microsoft employees, but still. The PDF featuring the study results is embedded below.

For each company, Rapleaf looked at the percentage of customers who purchased major grocery store products and compared the percentages side by side.

According to Rapleaf, more Google employees buy junk food like bacon and ice cream, but balance things out by also buying more fresh fruits and vegetables than Microsoft employees.

Microsoft employees, who are generally older than Googlers, in turn tend to buy more butter and vitamins.

Here’s how Rapleaf – creepily – puts it:

Microsoft employees are more likely to buy butter, a food rarely consumed on its own and used primarily in cooking. Microsoft employees are also more likely to buy vitamins.

Together this data paints a cozy picture of old-school Microsoft employees enjoying home cooking and soundly taking their vitamins.

Rapleaf even took a look at brand preferences, and found that Mountain View loves Mountain Dew and Doritos more than Microsofties, while Microsoft employees buy more Capri Sun and Orville Redenbacher’s. They also score about even on the Coke/Pepsi divide. Just so you know.

More study findings from the report:

Microsoft employees are more likely to buy dog and cat treats and grooming products, a tendency our expert says also correlate with a providing more for children and family.

Which brings us to demographics. Even accounting for age, more Microsoft employees than Googlers are married, and more Microsoft employees have children.

We found that the income distribution peaks for Googlers between $50,000-100,000 per year, whereas 40% of Microsoft employees have an income of over $150,000. The discrepancy can probably be explained by the higher proportion of married employees at Microsoft, since income reflects earnings per household.

Editor’s Note: The following guest post is written by a Silicon Valley CEO. Frank Dupree is a pen name

In the late 1990s, the rise of the browser was supposed to usher in an era of unprecedented opportunity for startups. A great part of that increased opportunity came as a result of the significant reduction in platform dependencies. No longer did the users’ operating system dictate their access to services or information. Even a behemoth like Microsoft was fighting hand-to-hand combat with small startups for the first time in decades. Fast forward ten years, and it’s 1985 all over again.

But even as the risks of dependencies become better understood by startups and investors, the ascent of Facebook and Twitter seem to point to an ever increasing number of startups with significant business dependencies. Recent changes to both Facebook and Twitter show that neither startups nor their investors can assume much when it comes to support for a given API in the future. Today, even the OS seems subject to dramatic shifts in record time. One only need look to Apple’s iOS to see how dramatic and unpredictable developments can change the landscape for startups, customers and investors.

Today, most startups build with significant external platform dependencies, whether it is Facebook, iOS, Google Apps or Twitter. There are a few types of dependencies. A simple distinction might be to call a dependency on a platform symbiotic or parasitic. Symbiotic dependencies are those for which both sides agree to terms of the dependency and for which both sides seem to derive a benefit. Developers on Facebook’s platform, for example. The most successful here being Zynga, which grew completely and—probably for the team and its investors—nervously within the Facebook ecosystem.

But some dependencies are parasitic. As such, the problem is considerably more sticky when a startup forms an unwelcome dependency. Consider Meebo’s initial products which violated IM network terms of service by “hacking” into the major IM networks. Meebo has had to re-invent itself as a website check-in and sharing platform and at considerable cost (the startup has just announced its most recent round of financing taking it to over $70M in total funding).

In fact, a great portion of the “aggregation” genre of startups have troubling dependencies on larger players who often consider their products either a violation of service terms or simply the next feature on the product roadmap. Take the social media aggregation tools like TweetDeck and Seesmic. It has become clear that even players like Twitter or Facebook who offer generous API access can inadvertently step on the toes of startups in its ecosystem with a new product release. For Seesmic, this appears to mean a third pivot for the startup. The company started as a video commenting platform for blogs, moved into the desktop client social aggregation business and with Twitter’s recent redesign jumped into the plugin (aka, longtail) and enterprise user space (see recent partnership with SalesForce.com and plugin announcements). So for early platforms like Twitter, players like Seesmic and TweetDeck exist in a grey space between symbiotic and parasitic for the platform.

Then there are the 100% parasitic players, including companies like Rapleaf. Rapleaf has a dubious reputation because it built its business on the aggressive and surreptitious collection of social network data and matched it to email addresses. Recently both Facebook and Linkedin threatened legal recourse unless the company ceased crawling and scraping their user data and expunged all existing data collected from their networks maintained on over 650 million email addresses. As a result, two weeks ago Rapleaf agreed to stop providing its customers with information from both networks. That can’t be good for business.

But it is likely even worse for the startups who had based much if not all of their products on data they were buying from Rapleaf, sort of a double dependency. Take Gist, Etacts and Rapportive. All three of these startups were dependent on Rapleaf to present social network information (photos, titles, updates, etc.) alongside emails, mostly Gmail. When their Rapleaf data stopped providing Facebook and Linkedin data, the startups were left only with what appeared to be that which they had cached prior to the cut-off this month. And so it may be no surprise that within just a couple of weeks of the Rapleaf changes Gist is reportedly in talks to sell itself to RIM and Etacts has announced its sale to SalesForce.com (rumored for $6 million). One wonders if the acquiring companies fully understood the dependencies these startups had to Rapleaf. (Prediction: Rapportive also will be forced to sell very soon).

There is an alternative to parasitic access to a platform, but it can be expensive. You can build your own data from scratch or get access to it through painstaking partnerships if it is critical to your business. You might not grow as fast as you otherwise would, but at least you can control your own destiny. Even for startups that are successful in building their business within a platform dependency, such as Zynga, one of the most important things they must do as they grow is to mitigate their dependency on the platform.

For startups and investors, however, the lure and benefits of developing on other’s platforms with or without permission may be too great to avoid. When it comes to acquiring those critical first users, it takes an Odysseian-level of cunning (or foolishness) to avoid the siren’s song of platforms like Facebook. Building on Facebook reduces both the cost of development and provides direct access to a massive user base. And what are the other options? Even if you put aside the costs of going it alone, a significant portion of startups will be faced with taking on significant dependencies for their success. How many mobile OSes can there be? Forget startups, after iOS and Android, even the future of big players like Microsoft, Nokia and Palm seems uncertain in the mobile OS space.

So in the future it will likely be that the most successful startups will be those which are best able to navigate the minefield of platform dependencies. And while the benefits of platforms like iOS, Facebook and Twitter are significant to reduce development costs and increase access to customers, one can’t help but lament the squashing of the promise we got with our first browsers: a world where startups and the behemoths of industry alike fought on more or less equal terms. But increasingly it is once again becoming a platform world. And the companies who control the platforms, control the profits.

Photo credit: Flickr/uub

Note that this traffic only accounts for members who are content with using the Twitter website, and doesn’t take into account the multitude of users who log on to third-party web services or desktop clients to access their Twitter streams. Either way you look at it, Twitter’s ongoing growth is staggering.

People information search specialist Rapleaf thought it’d be interesting to run some analysis on Twitter follower trends based on data it was monitoring closely for one of its clients, and the study gives us an interesting insight into how Twitter’s huge growth between March and June have affected following patterns of some of its most active users. We already learned most people on Twitter are sheep, but does that change over time?

Rapleaf recently helped one of the world’s largest consumer packaged goods companies identify the most influential and connected Twitter users within their customer list for a word-of-mouth marketing campaign. Part of the analysis that Rapleaf was commissioned to do involved researching how profiles of their client’s customers on Twitter changed between given periods of time, by closely analyzing the users’ following and follower count.

The company ran some numbers on their clients’ top 0.1%, top 1% and top 10% most-followed Twitter users within the company’s customer list and compared how these figures changed in nine weeks, from the beginning of March until mid-June.

Rapleaf will be releasing the numbers later today but was kind enough to give us a sneak peek.

Clearly, the catchphrase ‘the rich get rich and the poor get poorer’ is at least half true when it comes to Twitter users’ following trends. While the service’s growth understandably lifts the follower numbers of the average Twitter user along the way, there’s also an apparent popularity gap that continues to widen.

Based on the sample of 40,000 users that Raplead has analyzed – deemed active members because they have at least five followers, five friends or five updates – it seems that having lots of followers on Twitter means that you’re going to grow more popular more rapidly as the microblogging service continues to boom.

The top 0.1% of observed Twitter users climbed 275% in Twitter followers between March and June, while the top 1% increased only 146% in comparison, and the top 10% gained only 126%. Even when analyzing the median followers, the stats paint a clear picture: the top 0.1%, 1% and 10% of researched Twitter users saw their follower base grow by 78%, 65% and 59% respectively.

Could this be the Twitter Golden Ratio at work?

Looking at the difference between the popularity of the top 0.1%, the top 1%, and the top 10% during the month of June, Rapleaf’s study shows users in the top 0.1% have approximately 5 times as many followers as users in the top 1% and about 40 times as many followers as users in the top 10%. It’s unclear how many of them are spam, of course.

Also noteworthy: a user who barely makes the top 10% needs 11.4 times more followers to break into the top 1%, and nearly 55 times as many followers to enter the top 0.1%.

Wanna see how your popularity on Twitter is evolving? Check out TwitterCounter to get an idea. Not happy with what you’re seeing? Try tweeting more often.

Streakr’s social network is like Delicious for cool kids and is a less flashy take on Trig. It includes a profiles, a toolbar, and a stumble upon feature that lets you flip through links in a given category. Here’s the one for video.

The profiles look a lot like MySpace, consisting of the usual details, about me, photos, and seizure inducing layouts. Xenia is Streakrs’ Tom. However, where MySpace puts a blog and comments, Streakr puts in favorite links and your “thumbs up” rating for each. You can input the links into your profile manually, or use the Streakr toolbar to add links to your profile and vote on them. The toolbar also provides an interface to all the other functionality on the main site, and is currently only for IE, requires the .NET framework, and takes forever and a day to download and install.

There are a couple other sites with social networking meta search. Here’s the lowdown on a few:

Wink
Wink is fast and simple. It searches Friendster, MySpace, Bebo, LinkedIn, and Live Spaces. It also has advanced search features, like location, sex, status, age, and interests. It also lets you narrow your search by those fields after your first search.

ProfileLinker
ProfileLinker is the most comprehensive search engine, with 84 social networking sites including general, blog, cultural, dating, professional, student, and special interest networks. Unfortunately you have to log in to use it.

UpScoop
UpScoop comes ahead in ease of use. Unlike the others, UpScoop searches by email based on all the contacts in your address book. It searches Bebo, Classmates, Ecademy, Flickr, Friendster, Hi5, Livejournal, Multiply, MySpace, Ringo, Tickle, Tribe, Yelp, Mog, and LinkedIn. While it finds the vast majority of your friends off the bat, some drawbacks are that it can take UpScoop up to a couple hours to search for the last few and the need to hand over your email credentials.

Perhaps their flaw has been in taking the high road, and going out of their way to ensure that reputational feedback is being left by verified identities.

New startup Gorb, which I first read about on David Berlind’s blog, takes no such high road. Gorb allows, even insists on, anonymous comments and ratings about an individual. Like someone? Hate them? Tell Gorb all about it, using their handy Ajax slider to rate them from 1 – 10 in their professional and personal lives, and leave written comments as well.

Arguing that a “non-anonymous system also contains “noise,” as reciprocity creates a fake positive response,” Gorb founder Leonard Boord (trash or glorify him here) thinks that anonymous feedback is the only way to go.

They do have some checks on the wholesale slaughter of people’s reputations. Each written comment can be voted, Digg style, up or down by other users. If lots of people agree with you, your reputation is enhanced. If they disagree, your reputation suffers. The person being discussed may also respond to each comment.

I agree with Boord that services like LinkedIn are often a farce – people leave good feedback on others in the hope that the gesture will be returned. And since the user must approve feedback before it is published to the site, only gushing testaments to perfection ever see the light of day.

Rapleaf does a much better job than LinkedIn at getting balanced feedback from users. But there is still a cost to posting something negative – Rapleaf isn’t an anonymous service, and comments are at least tied to a user name.

With Gorb, there’s absolutely nothing to lose by telling someone how bad they smell, or how much you dislike the tone of their voice. Libel away.

I think Gorb goes too far, and will appeal mostly to people who have highly negative things to say about others. Without fear of being held accountable for their words, people may go a little overboard in their quest to “help” people know the truth about themselves.

But Gorb also has the smell of success about it. People will be drawn to this in the same way they slow down when they pass a highway accident. They are tapping into basic human psychology, and may make a buck or two on the back of what is sure to be the misery of others.

Upscoop is designed to help you figure out which networks your friends belong to, based on their email address. You give Upscoop your email credentials (including the password) for your AOL, Gmail, Hotmail, or Yahoo email account. Upscoop grabs your contact list, and then searches across a number of social networks and tries to find profile pages of people that you know among the 10 million profiles they’ve indexed across the major social networks. The process of searching is not instantaneous – it actually takes a few hours.

Clicking on any result will (sometimes) bring you to the profile page for that person. You can then add them as a friend or otherwise interact with them.

Asking people to give Upscoop their full email credentials to complete the search is going to be a tough sell. But this is a lot easier than searching for friends one-by-one on Wink. For people serious about connecting with friends, Upscoop may be for them.

In what I see as a good sign for Rapleaf, eBay appears to be selectively removing listings from sellers who point to their Rapleaf reputation profiles:

From: auended@ebay.com
Date: May 16, 2006 5:53 AM
Subject: eBay Listing Removed: Inappropriate Links (=LS &12362 JM10590049)
To: [redacted]@gmail.com

Date: Mon, 15 May 2006 14:54:23 PDT

Dear [redacted],

We appreciate that you chose eBay to list the following listing(s):

[redacted]
[redacted]
[redacted]

However, your listing was in violation of eBay’s Inappropriate Links policy and has been removed from eBay. We have credited all associated fees to your account and notified eBay users associated with the transaction that it has been cancelled.

We would like to take this opportunity to let you know what part of your listing is not permitted.

Your listing(s) contains the following information:

http://www.rapleaf.com/profile/view/[redacted]

http://www.rapleaf.com/profile/view/[redacted]

http://www.rapleaf.com/profile/view/[redacted]

eBay does have complicated rules for what type of outside links may be included within a post (see policies here). However, these rules are rarely enforced and there are countless examples of policy breaches that do not result in a listing being terminated. For example, see this listing for the sale of search engine DigForIt, which contains numerous third party links and which also clearly violates eBay’s linking policy.

In this case, it seems clear that eBay does not want Rapleaf’s new reputation system encroaching on their territory. And by banning links to Rapleaf, they may have just given them the marketing push that the new startup needed.

CEO Auren Hoffman promises the APIs for the core Rapleaf functions sometime this week, so expect to see Rapleaf mashups very soon. Users can pull their “reputation box” via a code snippet and post it on a website. Mine is below.

My TechCrunch reputation score on Rapleaf is here. As I said before, I believe Rapleaf is going to be a winner if they can close the right business development deals.

Last year I asked for an open version of eBay’s feedback system to be created (see no. 3 here). I also suggested that iKarma, who’s in this space, make changes to their product to address this larger market. But until now, no one came at this problem head-on.

Enter Rapleaf, a new San Francisco-based service created by Auren Hoffman and Manish Shah that is a fully open version of eBay’s feedback system. It’s in private beta, but will be launching to the public on May 7, 2006.

Rapleaf will allow anyone to leave feedback for anyone they’ve transacted with. Others can use this feedback to help them determine if they are doing business with someone who’d likely to engage in fraud. Rapleaf is eBay feedback for the rest of the web, and the offline world.

There are three important things to understand about Rapleaf – the interface and basic feature set, fraud prevention and detection, and their API set and related policies.

Interface and Features

Rapleaf allows any user to leave feedback for anyone (whether they are a user or not), based on a unique email address or phone number. Type that identifier into the search bar. If no results occur, you can be the first to leave feedback for that person. This can be someone you’ve bought or sold with, or just a friend that you want to endorse. Like eBay, a free-text area is included for comments in addition to a positive/neutral/negative rating.

A person’s feedback rating is a raw score, with a point added for a positive review and a point subtracted for a negative review (just like eBay). Non-transaction endorsements are calculated and shown separately. Various stats are calculated and displayed based on feedback ratings by others (see screen shot below).

Fraud Prevention and Detection

eBay knows when two people have transacted. That removes a lot of fraud issues (although a big issues on eBay has been users selling thousands of low priced items to jack up their feedback and then engaging in fraud with high ticket items). Rapleaf won’t have this luxury, and so they are focused on preventing and detecting fraud in their system.

First, any feedback left of a user can be challenged. At first this will be handled by Rapleaf employees. Later, they will probably use a third party to handle disputes.

Rapleaf will also employ human and machine based analysis of all feedback to look for patterns that suggest fraud, particularly in boosting a feedback rating. A user can probably get a an extra point or two every few months without being detected. But co-founder Auren Hoffman tells me he’s confident that they will be able to detect any large scale fraud attempts and shut the offending account(s) down.

He won’t disclose his trade secrets, of course, and so I have no way of telling if he’s right. But if they’ve nailed this issue, the biggest one facing them, then they’ll have gone a long way to achieving success.

Open, Open, Open APIs

Rapleaf will (on launch) deploy an API to allow third parties to access the key parts of the service. Account creation, review creation and feedback scores will all be accessible, and free, to third parties who choose to integrate. For non-eBay shopping sites, Rapleaf will be a competitive leveler. And if enough sites start to integrate with Rapleaf over time, the data will become even more relevant than eBay’s.

Final Thoughts

If Rapleaf succeeds in their plans, it will make the Internet a better place to do business. And Rapleaf won’t be limited to online transactions – there is no reason that transactions occuring offline can’t be measured by Rapleaf as well. The fact that an identifier can also be a phone number tells me that Rapleaf is already thinking this way.

Business development will be an important area for Rapleaf to nail. Their APIs will spur countless mashups without a direct business deal. But I also expect to see some big retailers, classified sites and shopping engines integrate with Rapleaf over time. That will require a good business development team to search out and close those deals.

Additional Screen Shots: