How Startups are Outsourcing Distributed Spam Attacks
There are two practical reasons not to spam:
- If you’re using your own server, your ISP will quickly identify you as a pest and shut you down.
- If you’re paying to use a mailing service, they will identify you as a pest before your ISP gets around to it.
The holy grail would be to use someone else’s computer, or lots of people’s computers. Each would send just a few spam messages, so the ISPs wouldn’t have much to go on. And better yet, these computers might be able to send messages from a familiar email address, to names in the user’s addressbook.
That’s the promise of botnets. And of viral invite pages. Gaming email’s trust system is a fact of life, as is any other ad/nuisance arms race. In this case, it’s already being solved by spam filters, automatic tagging, and Gmail’s Priority Inbox.
Meanwhile, this kind of campaign tends to game the system by getting users to recommend something they’re willing to put their name on. For un-trusted users, that doesn’t mean much (if the person whose email subject lines start with “Fwd: FW: RE:…” they’re not going to get much attention). For trusted users, that means the email is going to be seen as a legitimately useful recommendation. People getting coveted invites to Spotify really do want them. And when Spotify made people take a little risk by putting their name on the invite, that ensured that Spotify invites went to the people who’d want them the most (and share them, too).
Fixing the Ad Networks’ Reputation Risk
An earlier Digital DD piece covered the risks of disaggregating ad publishers and brands. A few follow-ups:
- ComScore has purchased Adxpose, a firm that identifies the context in which ads run. That’s a useful step.
- Digiday has posted an excellent follow-up to their earlier article on ad exchanges and ads being shown in harmful contexts.
There’s still an overwhelming information problem. The shear scale of the display industry—the same scale that helps match the exact right advertiser to the exact right audience—makes it prohibitively difficult to police bad actors. Perhaps when the cost of paying someone through Mechanical Turk to spot-check on ad in a hundred approaches the typical rock-bottom CPMs, big brands will be more confident in their bargain-hunting.
(Related: why we need a Bloomberg terminal for advertising data. The easiest critique of the ad network system is that it needs to quantify the qualitative. That’s a hard problem to solve, though merely having recognized ratings would go a long way.)
Google Alumni: Startup Factories, Or…
A Quora thread on successful startups founded by ex-Google employees features an answer from perhaps the world’s bluntest investor, Keith Rabois. “I doubt any are worth more than $50 million, and am confident that none of the consumer Internet startups founded by ex-googlers are (or will be).” It’s true that Google’s employees are far better at the behind-the-scenes work that keeps a massive search engine up and running; interface and product are the tip of that particular iceberg. So while Google has plenty of employees who can do a lot of good at other massive companies (hence Facebook’s habit of hiring as many of them as it can), Google hasn’t produced as many consumer-facing startups.
Gingrich Rocks the Vote with Fake Twitter Followers
Newt Gingrich’s massive Twitter following was always a statistical aberration. Here are a few more: the vast majority have weird names, few followers, and few to no tweets. Newt, directly or indirectly, paid for fake Twitter followers. This is not exactly a crippling revelation, but it’s an amusing one. The Gingrich campaign always looked like an online think-tank that had somehow sprouted a political arm. So it made sense that Gingrich was a Twitter early adopter. His million-plus followers, however, were too good to be true.
Zuckerberg Resigns from Facebook; Redesign Imminent
Well, a Zuckerberg resigned, at least. Randi was apparently a very valuable personality for Facebook, but they won’t have too much trouble attracting a solid replacement.
Meanwhile, in more interesting news, Facebook is thinking of changing the way the “like” button works. The WSJ also reports that Facebook is creating a new, unfiltered news feed. (Facebook already has that feature, if their “Most Recent” newsfeed tab is telling the truth.) It sounds like Facebook is creating a hybrid between Facebook comments and the like button, all backed by Facebook’s taxonomy of tags.
More Google+ Demographics: Huge in India?
Search Engine Land has some background on Google+’s growth, including popular browsers, common age groups, and common genders (all exactly as expected). A surprise: much of their growth appears to be international.
USV Leads $25mm LendingClub Round
Union Square Ventures has invested in a peer-to-peer lender designed to disaggregate consumer finance markets. If their numbers are accurate, it’s an incredible business. Apparently one of their most common customers is someone with good credit and a high revolving credit balance; LendingClub basically refinances for them. This could be trouble: the “debt consolidation” keyword set is extremely competitive, and most of the other providers are taking a much bigger margin from their customer. So LendingClub will have to push for some unconventional word of mouth marketing to survive.
Google+ Spawns Low-Quality Chrome Add-Ons
Google’s vision of the app and add-on ecosystem is not like its vision of the web as a whole: Google wants complete freedom in add-ons, even when it leads to a bad user experience. A prediction: Google will start sorting add-ons and Android apps by more of the same criteria it uses to sort websites. Including the dreaded “brand” signals.
Can Query Volume Impact Rankings?
SeoMoz has some research indicating that it can. Google has alluded to this when discussing brand signals. Of course Chase Bank is a trusted authority. People Google “Chase Bank” way more often than they look for “CreditCardGuide.com.”
Zillow’s Odd Lock-up Provision
Zillow has a strange twist to their shareholder lock-up agreement: they can sell early if the stock stays high. This gives them a wonderful combination of aligned incentives (they can only dump the stock if the stock seems non-dump-worthy) and terribly misaligned ones (if they can just keep the stock higher, they can dump more shares faster!).
Bing Defines “Quality” (and not the way Google does)
Search Engine Journal reviews Bing’s quality guidelines. The most surprising: Bing likes terse content, rather than long-form content (a caveat terms of links and traffic, longer content usually performs better). Even more oddly, Bing wants to see social sharing widgets—perhaps an attempt to increase the use of the Facebook like button, thus giving Bing a bigger advantage in related searches.
Google Overhauls Spam Reporting
Google’s notorious spam reporting page now has more categories and options. This has never been the complaint against it, but it could be an indicator that Google is taking action against spam.
Recent Research
Digital Due Diligence Weekly
