Wholesome 8===D

OCTOBER 7TH I POSTED THE AUDIO ADSENSE IDEA AND TODAY (TWO MONTHS LATER) I SEE THAT GOOGLE HAS BEGUN TESTING RADIO ADS (AS REPORTED IN THE WALL STREET JOURNAL, NEWS.COM).

HOW ABOUT THAT?!?


You know... If I'm not running the world by the time I'm 40 I have only myself to blame.


Did someone from Google read my blog and start implementing an audio ad project, or is this something they've had on the drawing board there for a long time? You are part of the mystery.

Radio stations that have internet streams I listen too usually don't play ads, at least not on the stream. I know this is partially for legal reasons, and partially because it's it's a pain in the ass even for large radio stations to have specialized ads for their internet stream -- not to mention small time internet radio stations and podcasters.

I don't know why the magic of contextual advertising hasn't been applied to online radio yet, but it would probably be easier than it sounds. Though maybe harder for podcasts than for streaming radio.

Companies like Google that are all ready good at data mining could feed an audio stream through an AI that would determine a context of the audio and send specific ads to listeners during commercial breaks. Though that wouldn't be necessary. You could probably do just about as good with the name of the show, songs, and standard location information about the listener (location of IP address), etc. It picks from databases of audio ads, plays them for the user.

The difficult issue would be bandwidth and playing specific ads for users. Server software would need to be changed to communicate with the ad-embedding companies servers to get the information and then send specific ads to specific listeners. Failing that, companies interested in contextual advertising with audio would need to setup their own streaming servers for stations to use. Either of these options are in the grasp of Google-sized companies, and would produce a fairly high return on their bandwidth. AOL could easily do it with WINAMP/Shoutcast.

Podcasts would be more difficult since they're pre-recorded. But, the shows and their metadata could be ran through the AI after being produced and ads recommend for commercial breaks. Podcasters would have to create some kind of contract with the advertising company. Or the advertising company could host the podcasts and embed their ads within the podcast to guarantee to advertisers that they're being inserted and listed too. Of course, this also re-creates the TiVo problem -- people could fast forward through them. That doesn't seem to have deterred television advertisers, however.

I sent this to suggestions just now. I thought I'd share it for those who might be interested.

Short, concise description of the idea
To allow ad revenue sharing for paid users (permanent accounts holders, and early adopters), add a second version of the print_ad function that will allow paid users to pass their AdSense code to it which then replaces livejournal's code when the ads are displayed on their personal journal. Function prototype, and how it would be called:

function builtin print_ad(string type, string adSenseCode) [fixed] 
"Prints an ad if the user should see one, 
using an adsense code specified by the journal owner.";

To be called like so...

 $p->print_ad("Journal-Skyscraper", "ca-pub-5788891423790015"); 

Full description of the idea
Allowing paid users (and the like) to share ad-revene or to post ads on their accounts has been suggested by myself, and others, more others, and even on the O'Reilly Blog. From conversation with pmmarcov, I was told it wouldn't be implemented because it would be too difficult to justify the benefit. What I've suggested is a simple and parsimonious approach that would allow all paid users, using S2 who can get an AdSense account, the ability to "share" revenue from the livejournal ads feature while not violating the principles already in place (in terms of what users when logged in can see the ads, etc). There would be no extra accounting issues for LiveJournal to deal with, and it would shut-up nagging users like me who WOULD REALLY LIKE this feature. :)

An ordered list of problems/issues involved

• None that I can think of.

An organized list, or a few short paragraphs detailing suggestions for implementation

• This will involve searching for and then replacing a text string.