Thread by @Chronotope

The more closely tied & aware a programmatic ad is with the content it runs against the less able the site hosting that content is able to execute scale at a profit. There's an unidentified point of diminishing returns for scale and it triggers faster the more unified the content

Facebook has untied content entirely from its advertising (except in the regard that it informs user targeting) and so can scale profitably basically indefinitely.

But the flip side of that is ideologically focused news orgs which basically can't scale with programmatic at all.

There's a weird transference here of how we (& I mean consumers) understand advertising in general that--no joke--prob started w/G*m*rG*te. Which is: the first notable action it took was to call for an advertiser boycott of a site they *liked* for hosting a writer they *did not*

I'm not against advertising boycotts. I think they have become a useful tool leveraged by folks like Sleeping Giants and against hate in general, but I am wracking my brain to find a campaign that impacted not just direct but the thinking around programmatic targeting...

and it seems to me that it really started there. And with it came a really serious shift in how the advertising marketplace worked: Brand Safety.

Which is to say: before then if you saw an ad in the NYT it was generally understood to be an ad targeted at you 'the image of a NYT reader' and not in support of the particular article it was next to. This was especially true in print, but also in the early digital ad days.

But advent of the digital advertiser boycott switched something around. Advertisers had nearly 1-to-1 access to their users and that meant access to their users' interests &that meant that they could surface the fear of the users' interests not aligning with users' reading habits

The feeds on platforms like this one and Facebook mixes up content. Advertisers can target just users without concerning themselves with adjacent content because that content is not ideologically consistent within a single view and is transient.

But a single article or... say... YouTube video is very different. The signal is singular and strong, even more than it was when reading a physical newspaper where articles themselves were mixed with each other, a single view not showing a single ideology or type of content.

And there are two phenomenon that occur when single pieces of content like that go viral:

New users come in who have no targeting data, or significantly less

And uhhh... Hate readers.

Now I don't understand hate readers. Perhaps there's a good article out there explaining them. But boy is it a big thing, and viral spikes usually have a lot of them.

There's this idea that news orgs just say shitty things to harvest the hate clicks. And well... maybe dumb ones

And when I say "dumb" I don't mean the orgs all have executives faceplanting all the time or something (though hey, some do) I mean like... they don't understand what they're doing because of the paper wall...

See, Editorial Functions are separated from Business Functions in news orgs & the Social Media-posting people are usually editorial side. So sometimes they understand that numbers need to go up, but don't always get the motivation of it.

So while some might harvest 'hate clicks', it usually isn't in their best interests and this is the main reason: when you get a viral spike on hate clicks that audience doesn't align with your normal audience for targeting either direct or programmatic ads...

There ends up being an increased cost (more people hitting your site means more traffic, servers or caching spun up, etc...) w/diminishing returns, b/c you can't show them your direct ads (high value) or normal programmatic ads (often based on site patters w/individual targeting)

And despite the fact that most news sites host lots of different types of writing (just like they did when they were papers) one poison pill viral event can impact the whole site for days with echo effects on the patterns of users and inventory across the site.

I don't think it is wild to say that Facebook hosts more hateful content than many far right sites. Just because there are a lot of hateful people on there. I mean, the whole 'aunt posting racist minions on FB' stereotype exists for a reason. I'm not even saying that critically.

Like anyone who admined a forum back in the day when those were a thing can tell you that moderating constant assholes while balancing their contributions against the community is A Job and at Facebook's scale that job is huge. BUT:

Forums & news sites are both permanent pages. But Facebook is impermanent for the vast majority of its content. PS: This is why sites are so Hot for transient systems like Snapchat. Advertisers can't pull advertising from a page that exists in the universe for 5 min for 1 person

And when that content is transient the only thing you can target *is* the user. This is a big advantage for social platforms that present content stylistically like a modern newspaper front page--a mix of a wide variety of content and ideologies.

Experts talk a lot about how social platforms optimize for engagement & how it leads to hate. But this is *why* they are incentivized to continue optimizing for engagement. Hate clicks are useless to Twitter for example, but hate retweets woo! More, faster, content streams.

Now... this hasn't been discussed for a while b/c it is old news, but news org home pages are infamously low on visits. It makes more sense in the age of social media, but it has been a thing since before 2007 as far as I know. I don't know why but the result is clear:

Readers encounter news sites entirely through single article pages that show mostly whole content and increasingly are being designed to push you to read whole articles.

If you read whole articles that would theoretically have more ad views, so this, in theory, makes sense.

But the end result has been this immense shift in how advertisers understand ad placements.

And I mean... IMMENSE.

There's a reason there are so many advertisers are surprised people are upset they air betwixt racist YouTube/TV Segments... it never even occurred to them the content around the ads mattered beyond eyeballs it supplied. Or people would watch something they hated to begin with.

But this is how media consumption works now. We are consumed with consuming things we hate in order to get angry about them. And that's an all-sides thing.

So a viral spike on anything but say... a story about an optical illusion dress... can be consumed w/hate reads who see the ads and hate the advertisers who are on them for supporting that thing. (Note the trends of smarm/positivity blogging that popped up in the last few years)

This is especially bad because the entire design of the shitty programmatic system is built on the previous understanding of how advertising works: eyeballs are about who has them, not what they're consuming. So...

Ads targeted users, & when those users angrily show up on Br**tb*rt and saw the ads that had been precisely targeted to them they were witnessing an unintended side effect. But they also totally had the right to be angry that these brands were, intentionally or not, funding hate.

And so we have this major shift in how users understood advertising and a major concern for advertisers who suddenly had to be *aware* of what their users thought about X and also not advertise against X if the user hated it.

This is a huge huge huge systemic-level shift in how advertising interacts with consumers completely apart from even the question of privacy and user tracking and it is real weird from a biz perspective and scary from a news org perspective.

B/c even before this shift started occurring, programmatic systems were pretty bad at even filling the type of inventory that occurs when massive social media-made viral hits happen. Now, the restrictions on filling that inventory are even more severe.

'Going viral' was something VC-backed news orgs did pre-implementing ads. The idea was all this traffic was just eyeballs and eyeballs sold, so the traffic meant the business would scale up the minute they turned on ads.

Spoiler alert: It did not work that way.

(Especially because a lot of these news startups were coming of age right as G*m*rG*te was happening and also seeing the impact of post-2016 significant--oft-justified--criticism of the media. And so they were even less prepared.)

There are two big reasons that ad money is flowing to Facebook and Google (specifically: Google Search) and that is first because boy do you tell them a lot about you to target.

The 2nd is b/c they are sites that host lots of content but zero context. When readers see something terrible posted to Facebook, they assign responsibility for that post to their racist relations or whatever, not Facebook (for the most part).

B/c people do not assign context to FB, advertisers will consider it safe for their brands in a way that they don't consider even the most prestigious news site, so more money flows to FB. But especially away from ideological news orgs, on any position on the political spectrum.

That's why all this concern about Brand Safety, even in the wake of Facebook's continual extremely public drubbing in press & society has had pretty much no impact on Facebook as a company. No matter how bad the general public's relationship with FB is, the ads are not in-context

And the same with Google Search.

I don't know what the end point of this is. Will advertising ever shift away from FB/G (without a major political event that impacts them structurally) at this point? Is there a way for news orgs to return to a point where ads are context-free event w/in content-based targeting?

I mean... it is worth asking if they even *should*. Like... perhaps a system that allowed some media cos to make money while publishing a mix of shitty hateful content with smart high-brow content was a bad system? It prob should be reformed!

But there's another problem in that: the system still exists, it just now supports bite-sized link posts and racist meme posts on social instead of news orgs. And the impact of that system there is far worse because it *does* scale on social in a way it *does not* for news.

This is also one of the reasons why I remain somewhat uncomfortable w/pivot towards subscriptions. Besides the fact that journalism needs to be easy to access for democracy to work. As we've seen with... uhhh recent op-eds... the subscription model supports the same bad behavior.

You should pay to support journalism because it's important (that's why I still work in it after all). But also, while readers often pay for the context they see the publication in, the actual money is assigned context-free w/no economic lever or signal against bad behavior.

So the end result leaves us not with a solution to the business problems that have gotten us to this point but instead sitting in the midst of a fire hoping that we can get enough water in a bucket to put it out.

On the flip side, I think increased privacy is a good thing. It leaves news orgs increasingly focused on *context* and it challenges advertisers to overcome their fear in regards to alignment. But it isn't a one-and-done solve. I think something more fundamental has to change.

I think in some publications it is starting, the fact that Going Viral isn't a way to build a stable media publication for most orgs anymore is starting to shift things. But I also think we need to more clearly talk to our readers about how it takes a village to make an article.

We can't just try & replicate the context-less state of the 90s or of Facebook. Consumers understanding of what advertising is & how it works online has fundamentally changed and I don't think it is going to change back. Instead, we're going to have to create different contexts.

The challenge is going to be to remember, the business side of news organizations must be driven by the same ethical imperatives as the news side. It doesn't matter if you don't think that's the way the news business works... your readers do now and they aren't going to stop.

Also tho... I think we need to talk to advertiser boycotts. There needs to be some shift there as well. For all social platforms pretend at context-free streams, they are making decisions that create context. Perhaps it is worth starting to engage w/advertisers over that issue?

My hope is we can come out of this massive upheaval in our understanding of the role of advertising in society w/some major changes in how readers understand news orgs, social platforms, and the context of creation. But also w/greater responsibility of all to readers and users.

(It's also worth noting that this whole issue started, was kicked off even, by a movement focused on hate, harassment & abuse. We also have to keep our readers honest by viewing that hate seriously and criticizing the shit out of it and not stopping or letting it set policy.)

PS: This is what I think about when I get a badly targeted ad on YouTube on a Sunday morning I guess. I am a very strange person.