Best of xkcd: Correlation
(Terms of use.)
Regular readers will recall my many posts on correlation and causation.
Improving Alexa Rank Is Embarrassingly Easy
I’m interested in many more things about the internet than I think are actually important. Two examples: (1) I like to learn about how social media services are used, though I am a very light user of them. (My Twitter and Facebook presence are dominated by my automatic blog feeds, so I’m “there” without ever really being “there.”) (2) I think Google AdWords is an ingenious idea, so I permit them on my site so I can see how they work even though they don’t actually generate much revenue. (What it does generate above costs goes to charity.)
And then there’s the world of domain ranking, of which Google’s PageRank is perhaps the best known and widely used since it informs every Google search. The Alexa rank was unknown to me until I noticed that some lists of blogs and sites are based on it. (There are many other types of ranks: Compete, mozRank, Technorati, and no doubt others).
With each type of rank there are websites and blogs that offer advice on how to improve your site’s score. In general I’m skeptical such techniques work, or was. Then, for fun, I tried some very simple approaches suggested on Dosh Dosh to boost this site’s Alexa rank, and they worked. Here’s what I did:
- On all four computers I use, I installed the SearchStatus Firefox plug-in, an Alexa toolbar for Firefox (*).
- I asked my family and a few friends to do the same, though I am only aware that two other individuals did so.
- I placed an Alexa rank meter widget on my site (scroll down and see it at the bottom of the middle column).
- I wrote this post.
That’s it. And in four month’s time this site’s Alexa rank improved by an order of magnitude. (Because it includes the same link as the Alexa rank meter widget, this post may have helped too, but I published it after the rank improvement just described had already occurred. For the same reason, item 4 can’t explain the rank improvement to date either.)
I roughly know why these techniques work. Alexa uses data sent by their toolbars and from users who click on the meter widget to estimate the proportion of all toolbar users and meter widget clickers that go to one’s site. So, by increasing toolbar users who visit this site (mostly just me and a few family members and friends) and thanks to the (likely very few) individuals clicking on the meter on my site, I am influencing Alexa’s statistics.
The fourth item in the list above also improves Alexa rank to the extent it draws other Alexa toolbar users to one’s site. The theory is that many Alexa toolbar users are hunting for ways to improve their own site’s statistics so they will visit sites with a post that screams: “How to Improve Your Alexa Rank” or “Alexa Rank Boosting.” Now, that’s not why I wrote this post, but I know that there may be Alexa-rank improving consequences, which will be fun to watch. (Like I said, sometimes I find even the useless somewhat interesting.)
What I find most interesting and surprising about all this is that the basis for Alexa ranking is so stupid. Clearly until today (with item 4) I have not changed traffic patterns to my site one bit via these techniques. Yet my site’s ranking dramatically improved. This is gaming, pure and simple, and shows what a joke the Alexa rank is. I’m not sure why anybody believes it is of value. It is a bit like fiat currency. It is of value because people think it is. That it is so easily manipulated is, frankly, embarrassing. Knowing this I mentally devalue Alexa ranks. I think they’re worthless except for the value others place on them.
Still, it seems to matter for some purposes so there is no harm in obtaining a better rank. And, clearly, it is not so hard to do just that.
(*) I’ve read that some Alexa toolbars send more than just the standard URL visitation and browser data to Alexa. Some blogs say that some toolbars send Alexa the data one types into online forms. That’s a bit frightening. But it seems the Firefox SearchStatus toolbar doesn’t do that.
Best of xkcd: G-Spot
Click to enlarge (ahem).
But solar cells really are sexy! (Terms of use.)
Austrianism, McArdle Win February
February brought the greatest number of monthly hits to this blog to date, by about a factor of two. The traffic boost was mainly due to interest in Ian’s post on Austrian economics and a series of posts reacting to Megan McArdle’s Atlantic Monthly piece on the relationship between health insurance and mortality. The common element is that both Ian’s and McArdle’s posts were provocative, as perhaps were my posts in response to hers.
Through blogging and the more focused attention on the media and politics that it demands I’ve come to understand that information and entertainment are inseparable in all but a theoretical sense. For example, you’re reading this sentence because it entertains or interests you. That it is also informative may be a reason why it interests you, but entertainment and information are nevertheless inexorably linked.
A sure way to entertain is to be provocative. Ian was provocative in one way–asking about the degree to which Austrian economics is taken seriously. McArdle was provocative in another–claiming that there is insufficient evidence to conclude that health insurance reduces mortality (despite all the evidence that it does just that). Apparently people really like to debate the value of Austrian economics and the sufficiency of the empirical literature on the insurance-mortality connection. And, thanks to multiple links from prominent bloggers and journalists, those people flocked here to read our posts on these topics.
Don’t misunderstand. Ian and I do not write blog posts for the sake of provocation. We write them principally to advance our own thinking and understanding. That they also entertain and invite debate is an additional benefit. Ian’s post on Austrianism was published after a week or so of back and forth with me and other economists about the nature of macroeconomics and the role of the Austrian perspective in it. We both concluded that we wanted some answers and his post was one way to attempt to obtain them.
My posts, and that of Michael McWilliams, in response to McArdle were motivated by a desire to correct the record and to infuse the debate over health reform with the best evidence available. In the process I learned a lot from Michael (the literature he reviewed was not one with which I was intimately familiar). Hence, though we may have entertained, we were motivated by a mission to inform.
But, as I said, the way in which we experience information cannot be separated from entertainment (perhaps broadly interpreted to include motivations that aren’t immediately rewarding, like the desire for a high grade). At the risk of being provocative, I’d go so far as to say that anyone who tells you otherwise is attempting to entertain you with clever theory that belies the practical truth. If you fall for it that only proves my point.
Best of xkcd: Natural Parenting
So long as I link to his terms of use Randall Munroe says it is OK to post his xkcd cartoons on this blog. Now that I subscribe to his cartoon feed (and his blog), I may do so more often. I’m thinking it’ll be a weekend thing, like this:
How Many Spaces Between Sentences?
This is a bit of silliness you won’t see often here. Skip it if you’re in a serious mood. Don’t skip it if what you are serious about is typesetting.
How many spaces should one put between sentences? My co-bloggers Ian and Steve both use two. See those huge gaps in their posts (Steve’s and Ian’s)? I do! (No I’m not bitter about it.)
I dropped the double-space habit years ago and am firmly in the one-space camp (no rocket ships). I’m so firmly in that camp that when I co-author papers, grant proposals, and other writing endeavors I frequently edit out all those extra spaces that the two-spacers think are required (global find/replace makes this trivial).
Maybe you weren’t even aware of this debate. Wikipedia has lengthy entry on the subject.
The use of a single space after the concluding punctuation of a sentence is the modern convention for professional typesetting, final, and published work (e.g. books, journals, and periodicals).[11][12][13][14] The use of a double space after a sentence has reversed nearly completely, primarily because of the widespread use of proportional fonts.[15][16]
What’s your opinion? Are you in the we-are-right-dammit-one-space-camp or the if-it-was-good-enough-for-grandpa-two-space-group?
Let’s Do Lunch
Below is the silly post I promised on Friday, the one that signifies that my switch to a new website host account has worked. By the way, if you notice anything unusual about the site, please let me know. It is possible not everything is configured correctly yet.
Many people have a list, if only in their heads, of famous people they’d like to meet. The difference between yours and mine is I’m going to post part of mine, right here, just for fun (really).
Below is a list of some of the people I’d like to meet for lunch. Dinner works too. Who am I kidding? Breakfast would suffice. Fine, I’ll take a phone call. A tweet? Anyway, I think I’d learn something and have a good time. And that’s enough for me. My only requirement is they pretend to enjoy my company for an hour (two if it is genuine). I’ll pay.
Here’s part of the list. I won’t bore you with the rest. Other than Stephen Colbert being my preferred lunch companion it is not in order of preference, as if that matters:
- Stephen Colbert
- Michael Pollan
- John Lithgow
- Richard Feynman (I permit dead people, so long as they don’t act that way)
- Nate Silver
But I doubt I’ll ever share a meal, let alone meet these folks. They’re far too busy (or dead) for the likes of me. Plus I’ve made the mistake of being available by e-mail (except to Feynman–his e-mails freak me out).
Also, if Stephen Colbert is wondering–and I’m certain he is not–I would accept but do not seek to be on his show. I could fill in for Obama, for example. But, really, lunch is enough. I wouldn’t make a good interview guest anyway. I’m only funny to children, animals, and the hard of hearing. Plus, I don’t have a book to discuss. I did write a PhD thesis, but if Colbert wants to discuss that he’s in the wrong business. Anyway, if we did lunch we could talk about all these things. See how rich an experience this could be for us?!?!
Geeks Heart Valentine’s Day
Remember to hug a geek today. From Wolfram Alpha:
Four (Or My Erdös Number)
Paul Erdös (1913-1996) holds the record for the most academic publications in mathematics. He was also famously itinerant, traveling continuously between the homes of colleagues all over the world. It is said that upon arrival he would announce, “My brain is open.” A few days later, after collaborating on some papers, he would move on. Despite (or because of) his eccentricities, Erdös was a beloved figure in mathematics and is still remembered fondly by many whose lives he touched. (A very nice segment of the Radiolab episode Numbers discusses Erdös.)
One way in which Erdös’ collaborative relationships and their influence is memorialized is through Erdös numbers. The concept is similar to the more popularly known Bacon numbers (as in six degrees of Kevin Bacon). Erdös himself has Erdös number zero. Individuals who coauthored a paper with him have Erdös number one. Individuals who are coauthors of his coauthors have Erdös number two, and so on.
My Erdös number is four, which I obtained by sending an e-mail to Jerry Grossman who runs the Erdös Number Project. Here’s my publication path to Erdös:
- Erdös P, Hartman S. On sequences of distances of a sequence. Colloq. Math. 17:191-193. 1976.
- Hartman S, Meyer Y. Interpolation harmonique sur les copacts. Colloq. Math. 40(2):265-276. 1978/79.
- Daoudi K, Levy J, Meyer Y. Construction of continuous functions with prescribed local regularity. Constr. Approx 14(3):349-385. 1998.
- Daoudi K, Frakt A, Willsky A. Multiscale autoregressive models and wavelets. IEEE Trans. Inform. Theory 45(3):828-845. 1999.
The implication is that anyone with whom I’ve published has an Erdös number no larger than five. Does that apply to those with whom I’ve co-authored a blog post? Not officially, but let’s just give them Blogös numbers of five for fun (Ian, Steve, Julian, that means you).
Lost and Found Cash
I walk a lot. My commute has six miles of walking round-trip. My wife and I walk everywhere we can around town and as a way to share time and thoughts. I estimate I spend about 23,000 minutes per year walking, which is about an order of magnitude more time than the average American spends doing so. Interestingly, it is nearly equal to the amount of time the average American spends in a car. (Source: Chu, Table 2; hat tip: Julian Jamison.)
Thus, all other things equal, I have a ten-fold greater chance of finding stuff on the sidewalk than does the average American. One thing I find is cash, about $20 per year. I know it isn’t much, but putting the above information together with population statistics I can estimate how much cash is lost and found on the sidewalks and streets of America per year. It’s approximately half a billion dollars.
The calculation is as follows. Based on my rate of good fortune, the average American finds $2 cash per year while walking (a factor of ten less than me, commensurate with the ratio of average walking time to mine). There are about 300 million Americans. The product of $2 and 300 million is $600 million.
However, I walk in an area of above-average wealth. There may be more cash about to be lost and found than in the average city or town (hat tip: Alison Pollock). A rough estimate is that my rate of finding cash is higher than average by a factor of 1.14, which is the ratio of the median household income in my metropolitan area ($52,800) to the U.S. median household income ($46,300). Applying this adjustment reduces the total from $600 million to $526 million (= $600M/1.14). Let’s call it a half a billion dollars in round figures.
Though it is a rough calculation, it is a plausible order of magnitude. That is, it is not obviously wrong. It passes the laugh test. Of course the calculation itself is laughably simple. One thing it neglects is the fact that I walk in urban and near-urban environments. Actually I’m not sure in which direction this affects the results. The opportunity for lost money is greater but the higher population density suggests I’d be less likely to find it. I also did not take into consideration the time of day individuals walk. Perhaps it is relevant that I do so during commute times. But again, money may be more likely lost at commute times (I find a lot near coffee shops), but there are also more people out and about to find it. Finally, it could also be that I just tend to keep my gaze downward more than most.
Having contemplated other methods of estimation and found them intractable or full of assumptions and their potential biases, I’m fairly satisfied with my approach and results. Do you believe it, at least to an order of magnitude approximation? Why or why not?
Subscribe
