Tumblr's Block Party

Tumblr’s Block Party

Tumblr has a spam problem. Which means I have a spam problem.

And, right now, I feel like I’m alone in fighting this problem, which drives me crazy.

Thing is, this should also be driving Tumblr crazy. Crazy enough to get on this junk. Big time. Fast. Hard.

At the risk of stating the obvious, people seem to like Tumblr, in part, because it’s super-easy and fun to see a post and say, “Hey! I like this! I’m on-board!”

So, we hit that little ♥, yes?

That this ♥ing generates a back link to the ♥er is completely sane. Everybody loves that. It’s why we hit the Dashboard like a free bottle of tequila.

But, as with troubled historical siblings like TrackBack, Technorati links, and on and on and on, that back-linking functionality gets WAY less fun and a hundred times more frustrating when it’s tacitly allowed to turn your site into a bad neighborhood.

Especially when that neighborhood is theoretically managed by a central system with access to a humongous data set—plus a community of residents who are more than happy to report no-good-niks.¹

I so understand that running a very large website is Hard To Do. But, candidly, we aging blog-types who choose to use Tumblr do so with full knowledge and acceptance that we are ceding this company a whole lot of the control that we grew accustomed to having (and needing) with our old-school, full-on blogging software. But, we did need that control and help. And, we still do, even though that control is now out of our hands.

Put differently, when I happily accepted Tumblr’s ease of posting, richness of community, and simplicity of management, I did so with the confidence that they could help me handle those things that used to require extra functionality, extra work, and extra vigilance.

And, by and large, they’ve done a great job with that. Especially, given their explosive growth.

So, while I’d never want to seem ungrateful for the wonderfully vertically integrated awesome that Tumblr provides, I can’t help feeling that—as the owners and controllers of all that wonderful, vertical integration—Tumblr is solely and uniquely capable of identifying and banishing obviously bad actors on our behalf. I have to imagine that they have ample data and means to identify this shit and terminate it with extreme prejudice.

Useful example? Consider how Google used Gmail’s scale to get really great at spam filtering. Gmail grokked enough data every day to see that when Certain Weird Things got posted to a lot of people at the same time—and especially when they shared key characteristics of other Certain Weird Things—they could guiltlessly and silently nuke both the Things and the posters of the Things in a way that didn’t harm either the app’s functionality or openness.

And, at least in my opinion, this is something Tumblr really needs to work on.

More pointedly, they need to hit this spam shit fast and hard or risk becoming the web’s next two-dollar whore.

How? I don’t exactly know how. But, I do have ideas.

As you know, I’m not a computer person. But.

If I were Tumblr, and I were looking for ways to identify possible spammers, I’d start by programmatically scanning for accounts with the same basic criteria that I have to manually apply every single morning:

New-ish Tumblr account
- Tip: I’d also grep for multiple dashes and dictionary words in the username
Uses the generic blue icon
Doles out a disproportionate number of “♥s” very quickly
- especially on posts older than a month or so
- WAY less likely to be noticed in the dashboard
Employs custom url to redirect to (SEO-targeted) non-Tumblr site
Has wacko, erratic, or non-existent posting record.

Imperfect heuristics to be sure. But it’s a start.

Does this mean everybody with a default icon is a spammer? No. Does this mean everybody who goes nuts ♥ing on strangers is a spammer? No. Does this mean every site that redirects you to a French-language erectile dysfunction site is a spammer? Well. Kinda. Maybe.

But, why not let the goddamned data do the heavy lifting for everyone?

Thing is: while my brain may be slightly more subtle than a regular expression, it’s a LOT less intelligent than a handful of cronjobby queries against 33,000,000 sites containing 12,000,000,000 posts. That’s a lot of information, and it has a lot to tell us.

But, yes, someone would have to ask. And, only Tumblr can do that asking at scale.

I’m not calling for anyone to paint their face and “Occupy Tumblr.” I’m not calling for a meme or a hashtag or a please-reblog-this or any of that other horseshit.

I’m just asking Tumblr to please try harder to begin their mornings the same way I do; by scanning the horizon for harmful shit that makes us all look like jackasses.

I’m happy to help how I can, team. I offer this with love and ♥s. Even as I type it with fingers fatigued from wearily hyper-clicking “Block.” Over, and over, and over.

Epilogue.

Since this rant is long, I wanted to add a “more” tag.

I always forget how to do that. So, I googled for it.

Just for what it’s worth, as of 10:46am PST, here’s the last 12 notes on that post from the official Tumblr Staff Blog.

Better get clicking, guys.

Me? I have zero confidence that Tumblr’s “Report this as Spam” button is actually hooked up to anything; which is crazy-making, given that I’m about 7 for 10 on guessing when a ♥ is spam—simply by looking at the icon and the username. ↩