no_lj_ads: public text targetting

kunzite (

kunzite1) wrote in

no_lj_ads,
@ 2006-05-16 19:04:00

Entry tags:

ad implementation, ad-targeting, changelog watch

public text targetting
in s1 land, if you've posted a public entry, the entry text will be sent to the ad-targetting server.
in s2 land, if you've posted a public entry, the entry text on [Recent|Friends|Day|Entry]Page will be sent to the ad-targetting server.

mentioned here.

edit: uniquewonders shows me that it is intended to grab the first 500 words. if someone could check out those regexes and tell me what they do, that'd be awesome.

seems the SVN changelog posting script doesn't ignore whitespace changes like the CVS changelog posting script did after i complained about it.

(Post a new comment)

	tallblue 2006-05-17 11:31 am UTC (link)
	Does this mean, under S2, if I make a public post in a community, that comment will be sent to the ad-targeting server? And this is for everyone even paid accounts? (Reply to this) (Thread)

	ex_shattered767 2006-05-17 12:53 pm UTC (link)
	If you make a public post, when it shows up on the friends page of a Sponsored+ account, whether you posted it in your own journal or a community, it will be sent to the ad-targeting server. And yes, this means everyone, even Paid and Permanant accounts, will be getting their entries scanned. (Reply to this) (Parent)(Thread)

	uniquewonders 2006-05-17 01:27 pm UTC (link)
	Thanks for claryfying. And that's another public service announcement post in my lj. sigh (Reply to this) (Parent)(Thread)

	ex_shattered767 2006-05-17 01:29 pm UTC (link)
	Like I said to Mike last night, "I don't have my journal locked because of issues with stalkers, I have it locked because the LJ staff just finished reading 1984 and decided to take some business ideas from it." Though it really is locked due to stalkers. :P (Reply to this) (Parent)(Thread)

	uniquewonders 2006-05-17 01:43 pm UTC (link)
	nods I wonder why type of info they collect when you don't have gender, age, country and interests listed and no public posts either. FAQ#267 talks about the content of public pages. What is is then? Links list? Friends list? (Reply to this) (Parent)(Thread)

	foxfirefey 2006-05-17 04:31 pm UTC (link)
	We're trying to extrapolate this straight from the code, and I don't think we've seen anything like that yet. On the other hand, this means I have to edit our guide to avoiding ad targetting.... (Reply to this) (Parent)

	ex_shattered767 2006-05-17 01:32 pm UTC (link)
	Oh, btw OT - I saw your post, but I don't know how to fix it (at least not at this time of the morning). One of the other mods should have an idea, though. (Reply to this) (Parent)(Thread)

	uniquewonders 2006-05-17 01:45 pm UTC (link)
	Thanks for telling me. I have faith in you guys :) (Reply to this) (Parent)

tallblue
2006-05-17 02:55 pm UTC (link)

I thought the content we post in our journals is our intellectual property? Or is the scanning of public posts being compared to a random person reading a post I marked public?

I have to say there is a big difference between a person reading my public post and it being scanned for marketing. That one person reading it can answer, or comment and can possibly become a friend. You can't say that about a marketing ad :(

(Reply to this) (Parent)(Thread)

foxfirefey
2006-05-17 04:30 pm UTC (link)

The only reason we're kicking and screaming is because we didn't want LJ to be running ads in the first place, at least not to people who didn't explicitly sign up for Sponsored+, and so all the information that is used to market advertising to people who didn't sign up for Sponsored+ feels rather ick, even if it's publicly available.

I mean, did you ever see the study where someone analyzed the frequency of words used on LJ? Or other research thingies where people crawl public content on LJ? Did they offend you, too? No, I'm not trying to be facetious, since it's definitely possible that they do as well, but I guess I'm just saying this because I'm a scripter/coder person who's used automated means to fetch public information from LJs, so you might as well be offended at me as well, if they're all equivalent.

Information publicly available on the internet will be crawled by bots, period, and sometimes I think people in LJ-land forget that.

(Reply to this) (Parent)(Thread)

uniquewonders
2006-05-17 05:03 pm UTC (link)

If I wanted my entries to be private or flocked I would do so so I don't mind the fact that anybody - man or bot - has access to them.
What bothers me is that this info is used for ad-targetting (but I do understand why it's necessary).
What bothers me is that things like that aren't announced and therefore most people are not informed (not that a lot cares).
What bothers me is that I'll have to ask S+ users not to friend me. It may be stupid but I don't want to have to think about the fact that any public post I make is used for ad-targetting.

(Reply to this) (Parent)(Thread)

foxfirefey
2006-05-17 05:21 pm UTC (link)

I'm still not quite sure what the code says it's sending, although I know it's sending *something* with regards to public posts. When I have time, I'll try to look at it more. Right now I've got bioinformatics stuff that has to be done and my main work lappy broke!

Also, I might mention that if you don't want any public post you make to be used for ad targetting, you're going to want to set your syn-level to something lower, like maybe titles only.

For anyone who doesn't know how to do that, go to the console and type:

set synlevel title

I think, and that should make it so the RSS feed of your journal only has the title.

I say this because if you don't want public posts you make to be used for ad-targetting, and someone uses some kind of RSS reader that does contextual advertising and subscribes to your LJ's feed, then there you go, your posts would be used for ad targetting.

(Reply to this) (Parent)(Thread)

	uniquewonders 2006-05-17 05:26 pm UTC (link)
	I thought I'd already done that. Apparently not. Thank you. And that's another thing that isn't documented in the FAQs I think. (Reply to this) (Parent)(Thread)

	kunzite1 2006-05-17 05:34 pm UTC (link)
	the console does not provide a way to view the current value of hte user properties. it will say that it sets the value, regardless of the current value. the only way to see that you've already set it would be to view the rss feed of your journal (Reply to this) (Parent)

	foxfirefey 2006-05-17 06:01 pm UTC (link)
	Hearsay says it will never be documented. Why, I cannot say. Deal? Deal between who? Why begrugingly? (Reply to this) (Parent)

mr_z
2006-05-17 06:00 pm UTC (link)

I, for a variety of reasons, checkmarked "Exclude my journal from most search engine results." I don't mind the random person wandering in, but I do have some reservations about someone building a targeted statistical profile of *me* based on a friend's willingness to put up with ads.

It'd be as if I threw a party, and one of my friends brought over one of his creepy buddies that's just a little too nosy and nobody else really likes. I didn't invite him. Shouldn't I be able to ask him to leave?

What I want to know is: When my posts get aggregated (and I assume my comments do as well), are they stripped of the poster's context or do they get tagged with my userID and corresponding identifying information? If the data is in the form of anonymous aggregation, I have less a problem than if the data attempts to build up pictures of individuals.

I'd still like to opt out of the aggregation.

(Icon chosen for the irony.)

(Reply to this) (Parent)(Thread)

	kunzite1 2006-05-17 06:05 pm UTC (link)
	from what i understand from the code, it sends the username of the journal's friendspage that's being viewed. not the author of the entry. but i could very well be reading it incorrectly since i'm not very fluent in perl. (Reply to this) (Parent)

	tallblue 2006-05-17 06:17 pm UTC (link)
	I checked that box also, I thought that meant that nothing would scan my journal. I guess I am wrong, I should go and re-read the TOS. (Reply to this) (Parent)(Thread)

What it does

foxfirefey
2006-05-17 06:52 pm UTC (link)

It adds the following to the <head> section of your journal pages:

<meta name="robots" content="noindex, nofollow, noarchive" />
<meta name="googlebot" content="nosnippet" />

With the "noarchive" bit in the robots one explained here.

Note that following these directives is just polite and not by any means enforced. So search engines like Google&Co follow them, but you can't expect absolutely no automated harvesting will occur. There is nothing LJ can do about that, really; if you don't want it to happen, friends only is your option.

(Reply to this) (Parent)(Thread)

	Re: What it does tallblue 2006-05-18 12:58 am UTC (link)
	I see that, I made some filters and friends only is the previous public and then there is the real friends only. I guess on public posts I will put things that I don't mind ends up in other places. That's what I get for being not aware :) (Reply to this) (Parent)(Thread)

	Re: What it does foxfirefey 2006-05-18 06:04 am UTC (link)
	I think it would be interesting if LJ had a "registered users only" option for posts, just like they do for pictures, but I'm sure it's been suggested and turned down for a reason. (Reply to this) (Parent)(Thread)

	Re: What it does uniquewonders 2006-05-18 03:51 pm UTC (link)
	Paid users can use viewer_logged_in in s2 but I agree that it'd be nice if it was accessible to anybody. And I don't think it has been suggested. I can't remember anything like that. (Reply to this) (Parent)

	Re: What it does ex_shattered767 2006-05-19 04:19 am UTC (link)
	No way to be sure. It might not be documented in suggestions and they could be just rejecting posts that suggest it due to them talking about it on their own a year ago and voting no. Only real way to find out is to suggest it. (Reply to this) (Parent)

Re: What it does

burr86
2006-05-19 08:21 am UTC (link)

The reason it's been turned out is actually a technical issue -- we have 32 bits that we can set for an entry's security. One of them is "Friends-only", 30 of them are custom filters, and we're left with one; people want to be rock-solid certain that we'd be using that remaining one for the best possible purpose (because we wouldn't be able to revert the change once it's done), and I think "Registered users" got rejected as security-by-obscurity, so not worth using up the last bit for.

(This was a while back, though, so people might be more likely to do it now? dunno.)

(Reply to this) (Parent)(Thread)

	Re: What it does foxfirefey 2006-05-19 06:10 pm UTC (link)
	Aha, knew there was a reason! Although I think your counting is a bit off--there needs to be a bit for private, no, unless private is controlled by another bit somewhere else? That would leave one bit for friends only, one for private, one not taken up yet, and 29 for custom filters. (Reply to this) (Parent)(Thread)

	Re: What it does burr86 2006-05-19 06:39 pm UTC (link)
	Nope, there's a different doodad for that -- "public", "private", and then "usemask". If usemask is set, then it's displayed to the people according to the 32 bits. (Of course, why we can't just extend the "public"/"private" doodad is probably just plain technical "it makes it easier to check security this way", though I don't know for sure.) (Reply to this) (Parent)

	burr86 2006-05-19 08:17 am UTC (link)
	Not to argue with whether or not this is a good idea -- I can definitely see where you're coming from on that -- but the FAQs specifically said that targeting would use public page content. So how was this "not announced" if it's been in the FAQ from the get-go? (Reply to this) (Parent)(Thread)

	kunzite1 2006-05-19 08:19 am UTC (link)
	the part that bothers me is that it seems that if my entry appears on an s+ friendspage, it's subject to being sent to the ad server since the username of the journal owner (not the entry author) seems to be sent as the "owner" of the content. (Reply to this) (Parent)(Thread)

	burr86 2006-05-19 08:21 am UTC (link)
	I'm not arguing with that -- I'm specifically staying out of that discussion. I'm just saying, the FAQ said "public page content", so why are people complaining that this was never announced? (Reply to this) (Parent)(Thread)

	kunzite1 2006-05-19 08:22 am UTC (link)
	i agree that it was definitely announced. i'm good at taking hold of a conversation and trying to steer it in other directions. :) (Reply to this) (Parent)(Thread)

	burr86 2006-05-19 08:22 am UTC (link)
	And I'm good at not taking the bait. :P (Reply to this) (Parent)(Thread)

	kunzite1 2006-05-19 08:23 am UTC (link)
	i noticed. :P (Reply to this) (Parent)

	uniquewonders 2006-05-19 02:01 pm UTC (link)
	Well, I disagree :P I love playing the devil's advocate... Oh well you already knew that. (Reply to this) (Parent)

Once more in English

uniquewonders
2006-05-19 02:04 pm UTC (link)

1) Announced in a FAQ is not announced in a news post. I don't know if the fact that info was collected was mentioned there or not but I can't remember reading that. Of course, I may be completely wrong. :) Plus the type of info which is collected has extended and I definitely didn't see anything about that.

2) Despite what the FAQ says it didn't used to work like that, did it? Public content was fetched if no other info was available, right? Now public content is always fetched. Am I wrong?

And the FAQ says: "Specifically, LiveJournal targets ads based on the user's optional preferred ad categories and optional public information from the user's profile, such as gender, age, location and interests."

Your first public post or any other public content is not mentioned here. Is the public content that is talked about above in the FAQ "public information from the user's profile, such as gender, age, location and interests." or is it something else which isn't specified in this FAQ?
That sentence is too vague for me to know.

(Reply to this) (Parent)

	uniquewonders 2006-05-18 05:38 pm UTC (link)
	For reference: http://community.livejournal.com/changelog/3122309.html # first 500 words (Reply to this)

Username:		Create an Account Forgot your login or password? Login w/ OpenID
Password:
	Remember Me
	English • Español • Deutsch • Русский…

About

Help

Get Involved

Legal

Store

LJ Labs