Jun 13

Twitter zombies and content theft

A few days ago, I noticed a new follower that didn’t look quite right. Check it out for yourself (@KoriWilbur).

I’ve never been a fan of people who use Twitter just to spray links – especially if they all lead back to the same site. There’s very little value in such a Twitter account. But when an account like this shows up, and all of the tweets have something in them that looks like a pattern (here the “$ ” preceding the actual content of the tweet, it looks even more strange. What’s funnier though, is Twitter on iOS provides a list of three accounts that are “Similar to …” and links to them. So when KoriWilbur started following and the account stood out, I happened to notice the similar accounts that Twitter said were similar. Man, was Twitter spot on.

Twitter recommended @CorettaBerk, @LuannJohn, @HilmaErvin, @LilliMoffett, and @MieshaTuttle, among many others. The more I refreshed, the more accounts it provided that were eerily similar. There were dozens and dozens of accounts using this same MO.

The person building out these accounts did a really exceptional job of trying to bury their “undeadness”. They always seem to have a unique profile picture and usernames that match full names (always female, as Twitter grifters like to do), some actual pictures/videos that the profile has posted, which tends to lend authenticity to a zo semi-unique bio, and the URL in their bio is always self referencing (back to their own Twitter profile). A key difference was that different accounts used a different tweet suffix than the “$ ” that my original follower did, such as “! “, “@ “, or “# ” (note that it often, but not always seemed to be a shift+numkey value – another example was “\\ “). Also, it appears that they try to often use different URL shorteners across the accounts, even when linking to the same topic.

Speaking of links – the links always go to mobilephoneadvise.com, which I had never heard of before this. Now I know why. They just plagiarize content.

Take a look at this post on mobilephoneadvise.com. Now take a look at this post on TechCrunch. They’ve blatantly lifted content off of TechCrunch (which seems to be the favorite site for mobilephoneadvice.com to steal from) and posted it on their own site. Intermingled amongst the pilfered content are links to games. Not even game reviews, mind you, just the same approach to games as content. See this game site, and then the post about the game on mobilephoneadvise.com. The entire overview paragraph is lifted from the developer’s site. At a quick glance, at least one the samples I downloaded even appeared to be malware.

In short, there’s nothing good going on at that site, and they’re another example of individuals gaming Twitter for personal gain – and in this case stealing content and possibly infecting devices to do so.

May 13

Twitter zombies? My favorite.

Within the last few weeks, a very annoying trend on Twitter began to pique my curiosity. I saw random accounts that don’t follow me marking some of my tweets as favorites. What was weird though was the tweets that were getting marked weren’t, frankly, my best work. But I started noticing more about these accounts.

First of all, as I said, the accounts that seemed odd were generally marking odd tweets as favorites. Take this tweet for instance, which has three weird accounts that have favorited it (and my friend, who was just being punchy). A few other examples that friends on Twitter noted are here, here and here. While I thought it was interesting that this one of mine was the last tweet of the day for me, it wasn’t in the case of my friends, although the tweets do tend to be marked in the evening.

The second thing I noticed about the accounts marking these was that they always had names that were nonsensical given their username. Take this one for instance. Username is Rossiengkh, name is Rosalina Harrey. The usernames of these accounts seem to consist of a first name and generally 3, 4, or 5 gibberish characters appended. The more I looked, the more often I found that the names on the account were completely unrelated to the username – or pictures that were even of the wrong gender. Pictures of men with female account names, etc.

Next, I noted that all of these accounts had few tweets (generally less than 20) and were created recently (May 3, 2013, in the case of the account above).

Upon examining each of these accounts when they marked a favorite, I found that most of them had quite a few accounts they were following, and quite a few accounts following them. The patterns I noticed with my initial favorite zombie continued through all of the accounts they were following. For example, look at all of the accounts Rossiengkh is following. When it came to their followers, the story was different. All of their earliest followers match the pattern as well (again, see those of Rossiengkh).

It’s here that I’d like to theorize why these bots don’t spam, but rather favorite tweets instead. It’ll make sense in a minute. First, imagine you’re a really new user on Twitter. Suddenly, out of the blue, some user, likely following more people than you, and more followers than you, favorites one of your tweets. You maybe poke at their account a bit, notice their followers/followees, and that they have a few tweets. So you follow them. NOOOOOO!!!!!!!!

That’s why these zombies all have nobody they are following that is legitimate, and their accounts all began with a similar stack of zombie followers, to add cred. While some of us who have been on Twitter for a while noticed the funky smell from these accounts, new users aren’t generally aware that there are people gaming Twitter.

What’s most interesting is that many of these zombies are marking tweets as favorites that existed before their account did (the account above, created in early May, has favorites dating back into at least 2012). I didn’t even know you could mark tweets that existed before your account did as favorites, though I guess on some plane it makes sense.

So… Why all this trouble? Why build out a network of accounts following other accounts, following other accounts? Favoriting random things on Twitter? To sell followers, of course!

Follower counts generally aren’t vetted – people don’t go through and scan your followers to see if there are real people following you or not (well, not always). But buying followers, as questionable as it is, appears to be a thing to artificially add credibility to an account. I think it’s pretty sleazy and frankly devalues Twitter.

So let’s talk about one more thing almost all of the zombies have in common. Not all of them, mind you, but most of them. A short URL in their Twitter bio that at first glance appears relatively unique, and uses either bit.ly, tiny.cc, or tinyurl.com (the latter of which has now seemingly killed off the use of their service for this scam). I haven’t tabulated how many unique URLs there are (let alone how many zombies there are), but I can only assume there are quite a few. But more importantly, these URLs are not actually unique underneath.

If you click on the URLs, the final destination that you wind up with is followersdelivery.com (no link because I don’t want them to get SEO). However, they appear to have a layover along the way, at bestgod.info. Followersdelivery.com was registered through GoDaddy on February 24 of this year, with a one year registration. The registrant is an individual in Zagreb, Croatia, with – I believe – a postbox. More interestingly, bestgod.info was registered on March 24 of this year, and was last edited two days ago, on Friday, May 24. Even more interesting? That domain was registered with fake credentials through Wild West Domains, LLC. The Spurger, TX address used to register the domain doesn’t exist, and the phone number is dead.

The initial bestgod.info domain appears to do a client-side redirection to the final destination. I’ve seen this trick done before, and there’s often logic thrown in on the client side (or even before then in the server redirection) that may be defeating Twitter’s ability to detect or block this URL (assuming they’re trying to). I mentioned over a year ago the risks of trying to unwind URL shorteners when it comes to really knowing what site is at the end of the link.

But the funniest part of this exercise has to be reading the followersdelivery.com site. The site advertises (shocker) all of the following for sale in bulk:

  • Twitter followers
  • Facebook likes
  • YouTube views
  • Instagram followers

The price of Twitter followers?

  • $20 – 1,000 followers (within 24 hours)
  • $50 – 5,000 followers (within 24 hours)
  • $80 – 10,000 followers (within 48 hours)
  • $170 – 30,000 followers (within 48 hours)
  • $420 – 100,000 followers (within 3 days)

My favorite part of the followersdelivery.com site, though, has to be the following, on their FAQ page:

Can I trust Followers Delivery? Are you a reputable company?
For many people, our services seem too good to be true, so we get this question asked all the time. Followers Delivery offers very popular social media services, low pricing and excellent customers support. We believe in offering an exceptional service to all of our customers and clients. Read the reviews that our customers have left us, we are sure you’ll be impressed.

And this little gem (underline emphasis mine):

Are these real users? How do you gather the followers?
Absolutely! We guarantee that these Twitter followers are real people and that no bots or proxies will be used in the delivery of your Twitter Followers.We rely purely on proprietary marketing and promotion techniques to get the job done right. We also own and operate a few high traffic Fan pages, Twitter accounts, Youtube channels and website which we use to generate real social media users for our customers.

<SPIT TAKE/> Yes. I’m sure that those are all real people that I’ve called out above. The Twitter undead.

It gets more interesting, though, as Followersdelivery.com was explicitly called out for their role in Michelle Malkin’s Twitchy site investigating Rachel Maddow. Note that even then the site had recently been suspended by the registrar hosting it. I’m not exactly sure who paid for what in that instance either (and frankly, given the low cost of buying followers, I can imagine someone paying to have an opponent’s Twitter presence defamed by “throwing” fake followers at their account). Regardless, I hope that Twitter does something to block this service relatively soon.

Dec 12

100 Days – On Twitter and the virality of exclusive information.

Early Aug. 2012 – Short of information about how the Windows Store, the forthcoming home for Windows Store (nee Metro) applications was doing, I began exploring the store, trying to assess how many applications were actually there. I had heard rumblings of 400 or so applications. As I said late in Sept. 2012, my intentions were never malicious. I pondered whether there was any way to query the store programmatically. Here’s how it went down.

Aug. 15, 2012 – I had discussed an idea with a friend on how to query the Windows Store, and tried it. My initial results, after fiddling with my idea for a bit, resulted in a count of 534 applications available worldwide.

Aug. 28, 2012 – I registered winappupdate.com and created the Twitter account @WinAppUpdate, but let them sit dormant for a while.

I was still only polling the store on occasion. On Sept. 5, 2012, I saw the store pass 1,000 applications worldwide – on Sept. 10, 2011 I began polling it every day, and we posted a State of the Store article at Directions that sought to more broadly discuss the composition of the store, not just the count.

Sept. 16, 2012 – I posted the first blog post on WinAppUpdate.com, announcing that there were 1,749 applications worldwide, and linked to it from Twitter. I added myself and a few other people, but effectively I had no followers. As someone with a psychology and sociology background, the virality with which the statistics spread and followers of the Twitter account for @WinAppUpdate grew, I found it fascinating, and I was glad that people found the information and the numbers interesting and helpful.

Over the next two weeks, as Mary Jo Foley, Alex Wilhelm, and Charlie Kindel, among others, mentioned this site, my stats, and the Twitter account, it began to grow in followers significantly. I can tell you from having had it happen on getwired.com before, you do not want to get sent Mary Jo’s traffic when you are not ready. It does unpleasant things to your site.

Throughout September and October of 2012, stats that I posted on Twitter resonated pretty well, and the @WinAppUpdate account began to grow in followers. Anytime Mary Jo or Alex would mention the Twitter account or the site, I would gain several – or many – followers. What I found most fascinating was how many of these followers were Microsoft employees or partners. There was a genuine scarcity of information around Windows 8 throughout the preview process and both employees and partners, among everyone else, seemed to want to know what was going on with the Windows Store. Over the next almost 3 months, the @WinAppUpdate account grew slowly at first, but then quite dramatically.

There were, for the next two and a half months, only two sources of application counts – my site and this site, where he has diligently counted the number of apps using a different approach than me, and at first I didn’t think his methodology would work – but I think it does, and lucky for him, it can continue where mine may have hit the end of the line.

Oct. 26, 2012 – Windows 8 RTM – I noted that the store was at 9,029 applications, and as media coverage of Windows 8 increased, so did discussions of my counts, and the number of followers. Conversations with Mary Jo and Alex on Twitter, as well as mentions of the Twitter account in the news media surely helped to keep the count of followers growing, even as I both diligently tried to stop making “the count” the focus of discussion – and in particular the discussion on my @WinAppUpdate account. On the same day, I also posted on getwired.com The Turn, where I more or less bared all about how I collected my statistics.

Throughout this exercise, as I said, my intent was never malicious. I knew at some point, Microsoft could – nay, should figure out how I was doing it, and either ask me to stop or simply shut off the spigot (easier said than done, as it could affect the SEO of the Windows Store). That said, I have always felt that if Microsoft asked me, I would have stopped counting or discussing the content on the store. I was providing the numbers, and then the deeper analysis, to help people understand what was going on in the store. I did announce when the store had hit 10,000, and 20,000, applications worldwide, and when there appeared to be more apps in the China Windows Store than the US Windows Store.

After I announced how I did my counts, as I somewhat anticipated, a few people started emulating.

Nov. 30, 2012 – I was about a group of guys who have built a Web-based front-end to the Windows Store, now at http://metrostore.preweb.sk/.

Over the next week, that site got some publicity. I continued to shift discussion away from counts both on my blog and @WinAppUpdate. Though that Web-based Windows Store now included a count, I only handed out a vague “over 25K” answer when asked during the week of Dec. 3, 2012. I had spent the last month really trying to find and discuss apps that I found significant, rather than discussing counts.

Early on, the script that ran my counter did its work in the morning. I shifted it to the nighttime – at about 10PM PST, when the time that Microsoft would push out their sitemaps seemed to have reliably occurred.

Dec. 5, 2012 – The last successful polling of the Windows Store occurred from my app.

Dec. 6, 2012 – I checked my stats in the evening before going to bed. My Excel spreadsheet, which has grown from several KB to more than 7MB, was completely empty, save my Excel formulas that I inserted at build. I checked the sitemaps manually. The first part was there, and pointed to the second. The second part was there, and pointed to the actual sitemaps. The actual sitemaps… were dead. Search engines that hit the site manually are going to be confused. But if Microsoft has elected to do this on purpose, then that meant they are likely submitting the sitemaps directly to Bing, Google, and Yahoo.

Dec. 12, 2012 – So here we are. It was 100 days since from the day I first created the @WinAppUpdate account and registered  the domain name to when I lost the ability to investigate the store. The Twitter account now has 1,378 followers, after only 831 tweets. That works out to 1.65 followers per tweet, or 16.65 followers per day. When you compare that to my original @Getwired account on Twitter, which I created on May 31, 2008, it’s pretty astonishing. After an insane number of tweets (59,518) and 1,653 days on Twitter, I have only 2,513 followers. With @Getwired, I’ve found that breaking the news – finding a news source before anyone else (or saying something genuinely, uniquely witty or thoughtful) – really does help generate new followers and create tweets that resonate. Though @WinAppUpdate did not generate a massive number of tweets, nor a massive amount that resonated widely on Twitter (I think 13 or 20 RTs of one was the largest I saw, and I’ve seen better than that on @Getwired), the number of followers, and followers who recommended the account to others, speaks highly of the value people find in Twitter when the information is truly unique, and can’t be found through any other source.

As for me, for now (for obvious reasons) I’m standing down on counting apps or mining the store. I can’t be certain whether I was the reason, or someone else was the reason, as to why the sitemaps went offline – or if they went offline by accident. @WinAppUpdate has been an amazing ride, and I’ve thoroughly enjoyed the conversations I’ve had around the Windows Store over the last several months. I’m leaving the app counting to those who have other means to count than my methodology, and leaving analysis of the contents and quality of Windows Store apps to those who have the time to do such an exercise justice – given the need to dig through the store to find them now, I don’t have the time to carry on with it.

I plan to retire the @WinAppUpdate Twitter account shortly. The blog URL will most likely redirect to this site, or perhaps a partner site if I can find a relevant one. All of the blog contents have been moved to Getwired.com for posterity, under the account  at WinAppUpdate. To my followers on Twitter – I thank you all for joining me on this wild and crazy ride, and hope you’ll join me at @getwired as well, where I’m really just getting started with Windows 8.

Oct 12

Windows Store: The Turn

The Prestige is one of my favorite movies. Every time I watch it, I notice a nuanced plot twist that I missed before. Since I started counting, the second most frequent question I get after, “how many apps are there?” is, “how are you doing this?”

“Now you’re looking for the secret. But you won’t find it because of course, you’re not really looking. You don’t really want to work it out. You want to be fooled.”

Back in August, I had an idea. I pondered… “What if there was a way I could count the number of apps in the store?” I had my reasons, as I’ve discussed before.

I bandied the idea around a bit, ran it by a few friends, and we were all stumped about how to come up with the best way to tally the number of apps in the store in any automated, robust way. While we could fire up the Windows Store app on Windows 8 and do a simple * query and get a dynamic result back of how many apps are there, it:

  1. Can’t be automated
  2. Isn’t categorized
  3. Includes only one locale at a a time
  4. It includes desktop applications that can’t be bought in the store and more importantly, don’t make Windows RT viable – that’s up to Windows Store (WinRT/Metro/Modern) apps.

So… what to do…

“Are you watching closely?”

I often say that everything in my career happened for a reason – and without a doubt, skills I’ve picked up in each job snowballed to be useful in the next one… As I tried to figure out how to peruse the store, I thought to myself – I know how to do this!

“Many of you may be familiar with this technique, but for those of you who aren’t, do not be alarmed. What you’re about to see is considered safe. “

I started poking around on Microsoft’s public facing Windows Store site (apps.microsoft.com). In a previous job, we focused on creating product videos – an innately unsearchable binary format. But you see – there are standard tools to make video indexed. The big giant heads of search all agreed several years ago on the concept of sitemaps. A very simple XML format, sitemaps indicate the content of a Website as the Webmaster wants it indexed. While sitemaps can be manually submitted by a site to each search engine if the Website wants to avoid random crawls, most sites simply post their sitemap publicly (see the NY Times’ sitemap as an example). I found that searching Bing and Google for Windows Store apps most definitely let me find very freshly added apps – indicating that there must be a sitemap in place on the Windows Store. Indeed, there was.

“Exact science, Mr Angier, is not an exact science.”

Through a process of <ahem> programmatic duct tape, I was able to create a system that iterated through the sitemap and indexed the content in the store. Through some futzing with XML, a few rough Regular Expressions, my existing experience with sitemaps, and some Excel automation (yes – really), I built a roughed-in prototype that worked – and took off. Bear in mind this is also why I only get results once per day. While Microsoft may be pushing new apps out to the Windows Store, the sitemap is not updated that dynamically. Someday it likely will. But my process of daily queries carried through the RTM and GA just fine.

“A real magician tries to invent something new, that other magicians are gonna scratch their heads over.”

So that’s it. It’s not really magic. I haven’t shared the source, nor do I intend to (you wouldn’t want it). I have been talking with a friend about possibly creating an app that helps provide better insight into what’s going on in the store – to help users find apps that they’ll love. Regardless, what I do intend to do from this point on is seriously constrain posts on statistics (at least sheer number statistics) related to the store. From this day on, regardless of what this site posts, the Windows Store will win or lose its role in the world not through the sheer count of apps… But instead through the caliber of apps that are there. The apps that users clamor for and say, “I need a Windows 8 or Windows RT device so I can run <insert title>”.

I may have been/still be a bit bearish on Windows RT and Windows 8 – and the WinRT platform as a whole. I don’t think that it as easy to ignite a platform -> app -> developer virtuous cycle as Microsoft thinks. But I also may be underestimating how important the typical consumer finds Windows (even the more constrained world of Windows RT) or Office to their life. We’ll see in time.