Recent blog entries for dmarti

Perfectly targeted advertising would be perfectly worthless

Another angle on the targeted advertising problem.

First of all, I could target you to sell a product or service with 100% accuracy. Just walk up to your house and offer to clean your gutters.

Naturally, you probably wouldn't take me up on it, because you'd probably think I was just casing your house for a burglary, but you get the idea. People get in your face and try to sell you stuff all the time. And, because you have no context or a signal of the value they can provide, you tell them no thank you. (You are polite, aren't you?)

The more targeted that advertising gets, the less well it carries out its essential role of sending a signal, and the more it's like an unknown guy on the porch mumbling about if you want to buy something. (More detail on this subject: Ad targeting: better is worse?)

Which brings us to a must-read article by Michael Wolff, on Technology Review. "At the heart of the Internet business is one of the great business fallacies of our time: that the Web, with all its targeting abilities, can be a more efficient, and hence more profitable, advertising medium than traditional media."

Exactly. The more targeted that advertising is, the less effective that it is. Internet technology can be more efficient at targeting, but the closer it gets to perfectly tracking users, the less profitable it has to become.

The profits are in advertising that informs, entertains, or creates a spectacle—because that's what sends a signal. Targeting is a dead end. Maybe "Do Not Track" will save online advertising from itself.

Syndicated 2012-05-23 15:47:06 from Don Marti

MLP: Sunday essays

For your Sunday afternoon reading, some recent thought-provoking articles.

Zócalo Public Square: Sprawling Is For Poor People. Affluent Americans are moving to city centers, with the rest ending up in the suburbs.

Washington's Blog: Lack of Trust – Caused by Institutional Corruption – Is Killing the Economy If the occupiers don't trust the gold bugs, and nobody trusts the banks or the government, can we get any work done?

Kari Hamerschlag: Subsidy Buffet for Agribiz, Table Scraps for Good Food. Speaking of not trusting the government...

Philosopher's Beard: Recovering Adam Smith's ethical economics. "Smith’s commitment to a realistic liberalism led him to endorse commercial society over any previous socio-economic system as a social order in which the most people possible could live decent lives. But he was not the blind zealot for the market he is now portrayed as."

Timothy B. Lee: There’s No Such Thing As Intervention-Free Infrastructure Policy. Is it more intrusive for the government to require net neutrality on an existing fiber optic cable, or to use eminent domain to let a competitor dig up your lawn?

Tim Devaney and Tom Stein: "Will Crowdfunding Crowd Out Venture Capital?" If crowdfunding catches on, who needs VC? After all, if every American invested 1% of liquid net worth in crowdfunding, that would be $300 billion, ten times the size of the venture sector.

Syndicated 2012-05-20 15:38:15 from Don Marti

MLP: Coding, equations, git

(Working on a script for mindless link propagation, and posting results. Some of these have been out for a while, but all worth a look.)

Technology that Benjamin Mako Hill uses, on usesthis.com

Rob Pike's The byte order fallacy: "the computer's byte order shouldn't matter to you one bit."

Worried about access to generative technology? Charles Miller is not: Johnny and Jenny Can Code

Software Cooperative News: DLT is better than CAPTCHA. Better ways to think about comment spam protection.

Akkana Peck covers MathJAX (JavaScript meets LaTex): Displaying equations on the web

Mosh, the "mobile shell" is the new shiny. Matt Simmons: Mosh Pit, or Why Terminal Emulators Suck suihkulokki: Mosh - better remote shell

Lots of good Git workflow and infrastructure ideas out there. gun.io has an intro to GitHub. Dan Croak covers working with remote branches for code reviews: Remote Branch Kenneth Reitz explains how to organize the repository for a Python project: Repository Structure and Python Adrian Holovaty has a good migration story: Moving Django to GitHub: the postmortem. Mark Dominus uses "rebase --interactive" extensively: My Git Habits

Syndicated 2012-05-20 14:40:30 from Don Marti

FreedomHEC Taipei 2012

Just heard from Regis Huang, with the good news that FreedomHEC Taipei is scheduled for June 12-13, 2012. FreedomHEC started as a Linux driver unconference in the USA, and for the past several years it's been in Taipei. Great list of speakers, as usual.

(I see Matthew Garrett will be there...maybe FreedomHEC will be the place to get to the bottom of the whole Secure Boot issue.)

Syndicated 2012-05-17 02:02:44 from Don Marti

Two million lawyers?

Creating jobs for the unemployed is a good thing, but hiring two million patent attorneys just to track software patents is not going to work. Timothy B. Lee and Christina Mulligan make the good point that software patents don't scale. A single program might infringe many of them, you can't automatically find which ones, and the US issues 40,000 new software patents every year. And it's not just software companies that have to worry. Anyone who hires a programmer, even to put a little JavaScript on a web page, is at risk.

To really find all the software patents you need to license, and license them, you'd need to spend more on lawyers than on the rest of your software development operation combined.

The one thing that's been bugging me about the Lee and Mulligan paper, Scaling the Patent System, is: Intellectual Ventures founder Nathan Myhrvold, who has been filing for and acquiring software patents at a tremendous rate, must know this. He's a mathematician and has been working with patent lawyers for a long time. He can't not understand the transaction costs of researching patents and negotiating licenses. Lee and Mulligan have done good research and written down some important insights on software patents for the rest of us, but none of it is going to be news to Myhrvold.

So if he's actually mathematically literate but doing this anyway, what's the real plan?

All that I can think of is that it has to be something like ASCAP for software patents. If you run a restaurant and have music, either live or recorded, you have to pay ASCAP, or the other performance rights organization, BMI, for a license. You don't have to track royalties to individual composers. ASCAP splits up its royalty stream, according to its own formula, and you're covered.

For software patents, you'd face something similar. You have five programmers? That'll be 20 grand a head, plus a 5% royalty on products and a 2.5% royalty on services. No fuss, no patent infringement lawsuits, and if you join now, we'll pay most of that back to you in generous royalties on your own patents.

There are already examples at a smaller scale. If you run Android or Samba, you can pay for a blanket licensing agreement from Microsoft, covering a large number of the company's patents. And of course you can pay the MPEG LA for a license to a pool of essential codec patents. Where this plan is different is in thinking big. All software, by anyone, under a easy-to-administer license. Easy as putting an ASCAP sticker on your door and hitting "play".

Sure, paid-up programmers would get an e-book subscription, a hat, and a messenger bag, but ultimately Myhrvold is on track to bring back, for computer programmers, the guilds or "corporations" of craftsmen that got such bad press in Adam Smith's The Wealth of Nations. But let the current patent wars get a little louder and more ridiculous, and CIOs and CTOs will be thankful for the prospect of an orderly system.

The funny thing is that most programmers above a basic skill level will probably earn more this way—if the paid-up license became a requirement of coding for a living, you'd be competing against fewer marginal programmers and wannabes.

Syndicated 2012-04-18 14:42:19 from Don Marti

QoTD: John Lanchester

"But ‘socialism for the rich’ was supposed to be a joke. The truth is that it is now genuinely the way the global economy is working."—John Lanchester

Syndicated 2012-03-31 20:48:25 from Don Marti

Dude, we're winning

Dan Gillmor calls the Jumpstart Our Business Startups Act "a bonanza to the kinds of people who see everyday Americans as sheep to be fleeced."

Apparently, crowdfunding is all well and good, but you have to watch out for shady characters who would take advantage of a crowdfunding system to peddle fraudulent "boiler room" investments to the long-suffering US middle class.

Gillmor cites a paper from law professor John C. Coffee, Jr., who writes, "every barroom in America could become a securities market, as some unregistered salesman, vaguely resembling Danny DeVito, could set up shop to market securities under the 'crowdfunding exemption.'"

And this sales guy is going to beat the socially connected, SEO-enabled network of real crowdfunders how, again?

The same way the encyclopedia salesmen beat the Wikipedia-Google Complex?

The same way that "high-touch interactions with potential customers" works so well for the software business?

If we get a better loophole for crowdfunding in the USA, it's likely to cause fewer opportunities for fraud, since the potential investors would already have sunk their money into bona fide companies—companies that they ran into while doing regular things on the dang ol'Internet, as customers, followers, employers of friends, whatever. Crowdfunding is more likely to displace the shady investment opportunities that people already have than to create more.

Fraud is an evil form of sales, and sales doesn't scale. When fraud tries to scale, it turns into spam, and the Internet has an immune system for that. (Hasn't everyone who would have sent money to bogus crowdfunders already sent their money to a 419 scammer, anyway?)

In conclusion, how about a nice hot cup of good old Internet optimism from JP Rangaswami? You're welcome, and enjoy your future crowdfunded investments just like you enjoy dealing with great small companies.

Syndicated 2012-03-25 19:46:38 from Don Marti

Too much buffering is bad, mmmkay?

POS display at Fry's: Looks like Linksys is down with the Bufferbloat project.

(Sorry about the crappy photo.)

Syndicated 2012-03-25 19:41:02 from Don Marti

Cato Institute: can anyone save the good parts?

Blog frenzy this weekend over the Ed Crane and the Cato Institute vs. the Kochtopus! story.

A little background: most of the libertarian-sounding noises inside the Beltway come from rent-seekers. Corporate-welfare-receiving companies pay for the think tanks, and you, the taxpayer, pay for the corporate welfare. (Hey, wait a minute. That means, in a way, Solveig Singleton works for me! We need to have a talk.)

One exception, at least a lot of the time, has been the relatively independent Cato Institute. They actually published "Archer Daniels Midland: A Case Study In Corporate Welfare." Nobody else in the think tank business wants to publish an article that will scratch a potential donor off the fundraising list, but Cato pulled it off.

But the end of an era is here. Cato co-founder Charles Koch, who along with his brother David is a prominent corporate welfare queen, is tightening up on the leash. Yawn. Looks like Cato is going to turn into yet another rent-seeking group.

With Cato either melting down or joining the noisy herd of fake libertarians with both front feet in the trough, what's a would-be reader of corporate welfare research to do? How about looking outside the think tank system entirely? After all, Kickstarter's funding for art projects is approaching the size of the NEA.

Your typical libertarian-sounding think tank is essentially a government program anyway. But it's an inefficient top-down channel for funneling corporate welfare dollars into content creation, and loses a lot to transaction costs. Let's think outside of the box here, and put the market to work.

Anybody want to write a think-tank-style report on a corporate welfare problem, but can't get it funded? Why not put it on Kickstarter?

Syndicated 2012-03-05 04:37:24 from Don Marti

Google gets DNT

Good news from the big G: Google commits to respecting "Do Not Track" in their advertisements.

SFX: the exploding head of Jeff Jarvis, head cheerleader for creepy database marketing everywhere. DNT is the "wholesale devaluing of advertising in a medium." People don't realize how much great online stuff owes its existence to hard-to-understand tracking practices.

Really?

There has to be something more going on here. If tracking is worth so much, why would the undisputed champion of online advertising just throw away so much potential revenue? Have the Googlers made themselves stupid?

I don't think so. Let's dig up the latest version of a classic chart. Compare the time people spend on different media to the percentage of advertising budgets spent there.

Internet and mobile advertising, which have the best potential for tracking, consistently pull in less ad revenue than their time share would justify.

If tracked ads are so much more valuable, then extra money should be chasing those highly-tracked Internet users. Thirty seconds of exposure to Jane Doe, expectant mom in ZIP code 90210, should be worth way more than 30 seconds in front of Joe Random Super Bowl Fan.

But they aren't. Manifest Density blog: "Is it really a coincidence that the advertising medium with the best instrumentation also appears to be the least effective?"

It's not a coincidence. It's basic economics. Advertising's main role is signaling. Most of why you spend money on advertising is to say "I spent money on advertising."

Tracking reduces the effectiveness of the signal. The recipient can't tell if you're really supporting a product with advertising, or just targeting him. In a DNT world, Google can get the best of both worlds: match ads to relevant content, but get the extra value of signaling to non-targeted users.

Background and links here: Ad targeting: better is worse?

Syndicated 2012-02-24 15:47:58 from Don Marti