I actually don’t identify that situation as a problem. For one, Arlene’s measured 27% follower attrition rate could very well be standard throughout Twitter ((A widely discussed Nielsen article from a few years back pegged sitewide retention rate among new users at 40%, but there are lots of reasons to think that number could be higher today and that Arlene’s followers wouldn’t be quite so fickle.)) , and because Klout scores are basically numbered arbitrarily, “inflation” or “deflation” in this sense doesn’t matter. Put another way, Klout scores are no less useful for being absolutely lower, as long as the same factors are skewing everybody’s profile in the same direction. Of course, Klout may also identify the “bug” Arlene describes as a feature. After all, if 27% of your followers aren’t likely seeing your updates, shouldn’t that be reflected in a measure of your influence?
As I see it, though, a real problem could arise based on the way Klout weighs the different factors that they measure against each other. Enough experimentation could tell you whether removing those 27% of followers would move the Klout score up or down. If the percentage of active followers is weighed more heavily than the absolute number of followers, then running an automated process to remove inactive followers could bump the score up a bit.
Right now, little rides on a person’s Klout, so there probably aren’t yet widespread efforts to subvert their ratings through tricks like that. But if we were to start seeing jobs, sponsorships, speaking engagements, or book deals offered on the basis of Klout scores, that could change — and when it changes, Goodhart’s law starts to kick in. ((I debated for a long time whether it’s Goodhart’s or Campbell’s law that is most appropriate here. I decided Goodhart’s was better, on the presumption that it’s the presumed constants of things like follower attrition rate that break down, and not actual following behavior that changes.)) According to that law, the “statistical regularity” of the behaviors Klout measures “will tend to collapse” under corrupting forces once “pressure is placed on it for control purposes.” In other words, users begin to game the scores and the numbers begin to lose their utility.
If that situation should arise, Klout could be compelled to respond by changing their algorithm to something more arcane, in order to be less readily reverse-engineered. I’m afraid that could be a problem for them. The more abstracted and convoluted the algorithm that produces Klout scores gets, the more likely it is to produce seemingly arbitrary results. Users whose natural influence profiles resemble the artificial ones of Klout-manipulators could see their scores adjusted downward, while other users whose profiles matched an increasingly complicated pattern might experience the opposite effect. If enough of these cases show up, Klout scores cease to be a meaningful proxy for influence.
The statistician George Box is famously quoted as saying “Essentially, all models are wrong, but some are useful.” The Klout conundrum I see is not that any particular factor is incorrectly considered — after all, the algorithm can be adjusted — but that they may be forced into deciding between making their model more wrong through action, or less useful through inaction.
I put together some mosaics of common images using dice faces as tiles. So far they only exist in computerized form, but the hope is to buy some dice wholesale and actually arrange some of these on a board. I’m still experimenting with what kinds of source images make for good output, but for now I’ve stuck to simple black-and-white symbols and line drawings. I’d like to start using more grayscale and complex shapes, but the challenge is keeping the image clear at a resolution low enough that the dice are still discernible and reasonable to arrange. (Some of these contain thousands of dice, which is a bit impractical.)
I’ll make a how-to in a future post, and after that I hope to document the actual construction of one or a few of these. Also, I relied heavily on, and am very thankful for, an excellent free software tool called Metapixel.
This post is cross-posted from the Students for Free Culture blog. I’ve also submitted a shorter version to the Crimson as a letter to the editor.
“A Sensible Compromise,” an editorial published in the Harvard Crimson last week, described the actions of the MPAA in urging universities like Harvard to develop a “written plan to effectively combat the unauthorized distribution of copyright material by users” of the university network in compliance with the Higher Education Opportunity Act of 2008. The Crimson’s take, as suggested by the title, is that these actions and the law that supports them are reasonable and justified.
The evidence for the Crimson’s claim is shaky, based largely on two sweeping claims about intellectual property. The Crimson states as common sense that without an effective intellectual property regime, there will be no incentive for innovation.
But around the world there are well documented examples of innovation and creativity that function in the absence of strong copyright protection: the world’s second largest movie industry, in Nigeria, and the booming “techno brega” scene in Brazil were both documented in the documentary “Good Copy Bad Copy,” which is available for free online. And that’s to say nothing of all of the innovations that took place before the mid-1700s, the works of Mozart, Shakespeare, Michelangelo, and all the others that lived before modern copyright was developed. Lastly, enormous areas of creativity like fashion, cooking, comedy, and even magic tricks operate without copyright protection. Closer to home, the entire academic publishing system functions without authors retaining copyright for their works, instead exchanging their monopoly for the opportunity to publish. Copyright can certainly provide a motivation for entrepreneurs to create, but in light of these examples, The Crimson’s statement that the absence of IP laws would eliminate innovation seems unjustifiable.
The second overbroad claim in the editorial pertains to a concept called “moral rights.” “Intellectual property rights are important,” according to the Crimson, “because each person has a fundamental right to enjoy the fruits of his or her mental labor.” The fact is that that justification is not uncommon in parts of the world, but has no basis in American law. The Constitutional “copyright clause,” in fact, is the only right enumerated in the Constitution with an explicit purpose, and that purpose is incentivization: Congress may secure monopolies for creators in order “to promote the progress of science and the useful arts.” No less than Thomas Jefferson was uncomfortable with the “embarrassment” of monopolies, but conceded that as an incentive, they might be worthwhile. As a fundamental moral right? He never even considered it.
Finally, the editorial talks about the concept of “balance,” and then gets into a discussion of business models, debating whether the ones that exist today are convenient enough to remove the justification for piracy. This discussion is an interesting one, and has a place elsewhere, but let’s not confuse an economic argument with an ideological one. In the world’s premiere institution of higher learning—and truly, in any institution of higher learning—the balance isn’t a question of business models. Should Harvard University, at the urging of a media industry that presumes the students to be criminals, reduce the flow of information available to them?
The MPAA and similar organizations are comfortable to disregard the educational benefits that technology has brought us and to see the Harvard student body as a group of potential criminal freeloaders. One can sympathize with members of the movie industry which, in spite of consistently breaking annual box office records, purports to be having a hard time. And it’s certainly reasonable for a university to discuss what the legal and technical guidelines of its network ought to be. But it’s wrong to kowtow to the demands of a media industry at the cost of Harvard students’ technological autonomy.
Expanding on last month’s very rough Kindle screensaver experiments, I’ve now put together a real set of ten nice looking close-ups of subway maps from cities around the world and packaged them for use on the Kindle. In order to install any kind of custom screensaver, you’ll need to perform the (very simple) Kindle “jailbreak” hack. Once that’s done, you should just be able to drop the image files into the directory created by the jailbreak.
There are ten cities included in this version, and they’re all systems I’ve ridden and maps I’m familiar with. If you’d like to see another map done in the same style included in a future release, please just send me the place and, if possible, a high res version of the map to work with.
Earlier this week I read a new essay by Benjamin Mako Hill called “When Free Software Isn’t Better.” Although I found it incredibly insightful, the reaction to this essay hasn’t been universally positive. The criticism has focused on a perceived attack on the Open Source Initiative. I want to address why I think Mako has taken a stance here that’s not aligned with the OSI, and why I also think it doesn’t constitute an attack.
In fact, what he’s done is articulated how important the goals of the OSI are, and why we need to work towards those goals as a conclusion instead of just assuming them as a starting point.
There are three major benefits most often touted by free and open source software boosters:
the “four freedoms” to run, study, distribute, and improve software;
the exciting collaborative elements of participating in a free and open source project; and
the superior software that comes as a result of that collaboration and many eyes making all bugs shallow.
Mako’s premise is that the collaborative development of the kind described by open source boosters around free or open source software projects is actually very rare, and that the superiority of free software projects is not guaranteed, and he provides evidence to back up those claims. But these are not reasons to abandon freely licensed software. Rather, he takes the opportunity to stress the importance of pursuing freedom, which is ensured even where the other benefits aren’t.
Of course he agrees, and he goes out of his way to clarify, collaborative development and community projects are incredible, popular, and real. The Linux kernel, Ubuntu and Debian distributions—and in the free culture world, projects like Wikipedia—all demonstrate not just freedom but also superiority through community participation. That can and should be an end towards which groups like the Open Source Initiative strive. Mako’s consideration, then, is just that nobody take those qualities for granted.
When active community involvement and better software are taken for granted, their absence is seen as a particular shortcoming of that project. And that situation sells the developers short: the individuals that are doing great work without a lot of community behind them, and the many projects in earlier stages that are functional and useful but lack the polish of proprietary versions deserve support and users, too.
That idea resonates deeply with me as a user of free software. I grappled with a tension between a commitment to free software being better and my hands-on experience with lots of it which, to put it bluntly, isn’t. It was after seeing Mako speak about “antifeatures” at Free Culture X last year, a talk which held the seeds of this essay in it, that I realized the reason I stick with free software when it isn’t better is because it’s free.
There was a time, certainly around the 1998 founding of the Open Source Initiative and especially in the “nobody ever got fired for buying IBM” world of enterprise IT, when free software was not an option even when it was better than the proprietary alternative. Now, due in large part to the efforts of the FSF and the OSI, that’s not the case any more.
Today, it’s easy to use free software when it’s better. But there are important and non-obvious reasons to do so even when it’s not.