Twitter clients

I've started using SlipStream with my desktop Twitter client. It's great, but it makes me yearn for what it doesn't do.

I can:

• Mute individual users (I just unfollow, so not needed)
• Mute by hashtag or by keyword

I cannot:

• Mute by all variants on a muted hashtag/keyword without muting them individually. So when a celebrity dies and my stream is filled with RIP XXXX, unless I want to mute the first name and the last name individually (and over-mute), I have to mute a half-dozen variants. 
• Mute all the TV shows. Sure, I can mute individual hashtags associated with TV shows, but I can't mute the general idea of tweeting about whatever you're watching on TV. 
• Mute all the sports. As above, I can mute individual hashtags, but they're not always used. 
• Mute by context rather than keyword. A smart client should be able to tell that a tweet saying "Oh, they just scored" is really about the Super Bowl and should be blocked if I blocked the Super Bowl even if it doesn't say Super Bowl.

I also use Tweetdeck and Falcon Pro. Tweetdeck lets you filter out particular words, but you have to go over into settings to do it. Falcon Pro and SlipStream let you mute from the offending tweet, but Falcon doesn't let you mute by keyword, just by hashtag or user. 

Here's what I want. A smart Twitter client that has as a mute option "I just don't like this kind of thing" and makes sure I don't see much of it again. If a local celebrity dies, I want to see the first two tweets about it so I know to offer condolences to the Kiwis, but I then don't ever want to see another thing about it. I want a global setting saying "If somebody dies, I don't want to see more than 2 tweets about it unless I whitelist it or unless the Twitter volume means it really is really really important - in which case give me an "Are you sure you want to keep blocking this?" query". 

I don't want to see two dozen different tweets about some great article in Smithsonian Mag. I read it the first time, don't need to see any more tweets about it.

If the client can't parse tweets' meaning fast enough to do context rather than keyword filtering, it ought to suggest filter bundles. Like "I see you have muted "Super Bowl". Would you also like to filter the name of every major sports team that is not also a word in common usage?". That would have me miss tweets I'd want to see about silly stadium funding decisions, but having a manageable timeline with a better signal to noise ratio would be worth it.

Demand can only bring forth supply if it is revealed: I'm willing to pay $20 for a client that handles all this really well. In Arrow-Debreu worlds, if there are enough people who want this, it will exist. If someone's found a workable solution, I'm all ears. 

In Praise of Iconoclasm

The BBC, the Catholic Church, the Boy Scouts of America, the Penn State Football Club. Heck maybe Sesame Street.

Pedophiles like institutions that are too-sacred-to-critique.* Whenever we set up institutions for veneration such that the optimal within-institution response is to cover up rather than to punish in response to individual failings, we're going to have problems. Sesame Street has been great on this front; the others, really not so much.

And so iconoclasts do us a service in making sure that nothing is sacred.

Today's edition: the British Royals. I'm a fan of the Queen. But I was still pretty disappointed to see half of Twitter piling onto the couple of radio hosts whose prank call to Kate Middleton's hospital led to the suicide of the switchboard operator who connected the call. Surely there have been hundreds of thousands of prank calls in the last few decades. There's one radio station in every major city that does this kind of thing at least weekly; start adding it up. How many prank calls did the Jerky Boys run? The accident rate seems awfully low, all things considered. If you're going to pile onto these particular radio hosts, what about all of the other ones who took risks no smaller than these guys did and whose calls led only to a bit of fun?

One response in particular disappointed me though - that this prank call was different because it involved the Royal family. I'm not embedding because of coarse language. But no institution is too sacred to prank. Radio hosts prank calling venerable institutions helps to keep those institutions from being too sacred for other forms of critique. Lew, on Twitter, replied with some topics he thought too shocking and offensive for comedy; Gilbert Gottfried's performance at the Friar's Club Roast for Hugh Hefner post 9/11 is the appropriate reply.

And for the Twitteratti baying for blood - what will you do if one of the family of one of the radio hosts commits suicide over your bullying? And when the rest of Twitter decides just which tweet was most to blame and piles onto that Twit, prompting another suicide, and another round of blame, rinse and repeat. I know it's all fun to feel morally righteous by demanding punishment, but who among us has never laughed at at least one prank call on the radio at some point in our lives? You know it's listener demand that drives what's played on the radio, right? 

* If you don't believe me on the quasi-sacred status of Penn State, you've not been to Pennsylvania.

Pro tips for Twitter hackers

There's clearly a pretty strong market opportunity for criminals who aren't complete idiots to reorganize the "hack a Twitter account, send out spam tweets" industry.

Whoever hacked Tyler Cowen's account used it to send out two spam Tweets:"An amazing new weight loss product! It worked for me and I didnt even change my diet!" and "Lose 5 lbs of Fat in a week", both sending the user to some t.co address I dare not hit. It's hard to imagine anybody who follows Tyler would have seen those and thought it more likely that he'd sent them than that his account was hacked.

I would have thought that Twitter account hackers would have run everything through a filter. Anybody with small numbers of followers or low Klout scores would get the lame spam tweets. But Tyler has a Klout score of 60 and about 20,000 followers, including some of the world's top economists and surely some top of the world's top government and central bank officials (among those on Twitter). I would have thought that a flag would go up for hackers that accounts with >10k followers or Klout >50 just might be worth a bit more individualised attention.

What sort of individualised attention? A decent proportion of Tyler's followers would have hit a link recommended by @TylerCowen to something like "This is the new best explanation of how the Euro crisis will unfold". I'm (obviously) not even trying to make it sound like Tyler. Scrape the content from some page from the Economist, FT, Scott Sumner - whatever. Put it on a malware infection site. A thousand really high value computers get directed to the site; maybe you get 250 infections depending on the strength of folks' security settings.

Just flip through the first 50 Tweets and see what's drawn a lot of clicks (hover over a bit.ly link sometime). Based on the feed, I'd have set up malware sites with fake reviews of Tyler's new book (Twitter teaser: Now this review of *An Economist Gets Lunch* is particularly unfair [link]); something on the EuroCrisis, a eulogy to Doc Watson, and something on fear of GMO foods. All of those drew lots of click-throughs. And throw in one like "A handy guide for every central banker as the Euro dissolves." Scrape dummy content into malware sites for each.

So obviously Twitter spammers aren't doing this. Or at least Tyler's hacker didn't. We can then conclude:

• The expected per user returns to malware infections are very low, even for potentially high value infections;
• I'm overestimating how easy it is to do this; cognitive limitations are more binding that I expect.
• Tyler's followers just got a lucky draw; he was hacked by somebody who installed FireSheep and isn't linked into any particularly sophisticated networks. 

The first one's potentially plausible. The second one isn't - somebody will figure it out and will pay more for hacked account login details than will other spammers. The third can be sustained in equilibrium if you've always new hackers downloading FireSheep and imperfect information on who's paying most for hacked accounts. 

Tyler Tweets

If you hacked Tyler Cowen's twitter account, with almost 20k followers, surely you'd do something more interesting than send out a generic spam note about some new diet.

An amazing new weight loss product! It worked for me and I didnt even change my diet! [link redacted]
— tylercowen (@tylercowen) May 31, 2012

What a waste. Given this opportunity, you could either go for lulz or for money.

On the money side, I have a hard time seeing a better play than shorting Spanish or Italian debt and then claiming that you have it on good authority from a central bank insider that default is imminent. There's a slight risk you cause the default by setting a run, but that's more a feature than a bug. Update: On second thought, pushing a penny stock likely gets more cash but fewer lulz.

If you're in it for the lulz, @ModeledBehavior delivers up gold at #TylerTweets - the hashtag Justin Wolfers suggested. I'd put up this:

Tucked away in a strip-mall off Columbia Pike, found a place that serves "long pig". Better than you'd think. Go! #TylerTweets
— Eric Crampton (@EricCrampton) May 31, 2012

But @ModeledBehavior's offerings are much funnier:

Its hard to imagine spoons will exist in their current form in 30 years. What does this tell us about the social discount rate? #tylertweets
— Modeled Behavior (@ModeledBehavior) May 31, 2012

Given what we know about the money illusion, should the moon be destroyed or doubled? The answer is not clear to me. #tylertweets
— Modeled Behavior (@ModeledBehavior) May 31, 2012

#tylertweets cannibalism is wrong, but not for the reasons it's critics say. We ignore the wisdom of cannibals at our peril.
— Modeled Behavior (@ModeledBehavior) May 31, 2012

#tylertweets careful viewers will note Big Momma's House 3 is biting satire of modern central banking. Most underappreciated drama of 2011?
— Modeled Behavior (@ModeledBehavior) May 31, 2012

@ModeledBehaviour, I salute you and your mad surrealist Cowenesque stylings. When you finally decide to make the end-game play with your growing Twitter and blogging audiences, it will be glorious.

Update: It gets better! While everybody who's anybody is playing #TylerTweets and often hitting @tylercowen in their updates, the spammer puts up another lame offering:

Lose 5 lbs of Fat in a week [link redacted]
— tylercowen (@tylercowen) May 31, 2012

What a waste.