Edit:

Yes, the proportion of total comp that was equity hit about 50% at L7 (Twitter's Staff). L6 (Twitter's Senior) was usually around 40%.

Advertisers won't pay for ads that don't get clicked on. Political pissing matches and disinformation campaigns don't bring the kind of engagement that results in clicking on ads and buying things, or of making positive brand associations.

Engagement costs us money. It increases system load and the resources required to actually run the platform. Raising engagement for the sake of raising engagement is a net negative in the long run, because it results in too much engagement that doesn't sell products (and thus ads).

I suspect there are some here who won't believe this; a few may be tempted to throw out a certain pithy Upton Sinclair quote. I'd like to know from this camp why you think that way.

First, when a company has shitty earnings, what metrics do they typically use in their investor reports to make the company still look palatable? New user acquisitions. Increased engagement. It doesn’t matter that the engagement is not generating money because it still lets them say “well, we lost a billion dollars this quarter, but user engagement is up 80% YOY so things are looking good for the future! (Please keep buying our stock/fund our next round!)”

Second, engineers working at a social networking company can’t directly influence what advertisers spend, but they can influence the behaviour of users, and leverage that for personal gain. “Look at how valuable I am, I developed this enhanced rage generating machine that increased engagement by 20% in A/B testing! Give me the big bonus now!” It’s a much harder sell to say “hey, Boss, I deleted a feature which was causing immeasurable subjective harm to society! It probably caused 100,000 users to delete their accounts and move to Gab! Give me the big bonus now!”

> I'd like to know from this camp why you think that way.

Cool, hi.

> Engagement, in the aggregate, does not pay the bills. Advertisers paying us to sponsor tweets does.

half-truth; engagement in the aggregate does not pay the bills alone, but advertisers pay for people to see sponsored tweets, and more people see the sponsored tweets if more people are online for more time, and more people are online for more time if engagement is higher, so... engagement in the aggregate does in fact pay the bills.

> Advertisers will only do that if paying to sponsor tweets results in either improving brand image or converting sales.

false; this is a nice rosy image of advertisers as rational actors operating with perfect information that is not at all realistic.

If an advertisement performs poorly, who is to blame: twitter? the ad copywriter? a graphic designer? the product's dev team? A product owner? The marketer who made a buy on some specific targeting segment? Their manager? The guy who hired those two? The CMO? The CEO? Marketing teams get fired constantly, almost annually, but very few firms are ever going to all-out stop advertising on social media platforms on the basis of poor conversion on their ads, because they cannot know for certain that it's not their fault, and they have to keep playing the game even if it is rigged.

The thinking here is easy to comprehend: "Walmart is still doing it... so we must be doing it wrong."

> Advertisers won't pay for ads that don't get clicked on.

again, false. Advertisers do this all the time, because they're cargo-cult following buffoons and there's money sloshing around fucking everywhere.

Advertising money gets laundered, advertising money greases wheels, advertising money gets used by incompetents and people's nephews and rapacious over-confident investors who are running their own little schemes, and a billion other things, and it even occasionally gets used by savvy nerds who come to suspect it isn't actually working, but I assert: most advertisers will repeatedly pay for a lot of dumb shit that does not work at all. Always has been.

> Political pissing matches and disinformation campaigns don't bring the kind of engagement that results in clicking on ads and buying things, or of making positive brand associations.

This is an interesting academic distinction and I'm sure you know more about specific content -> behaviour readouts than I do, but the theory is simple: more time online results in more impressions and more impressions yields more conversions. Not a higher rate, mind you: just. more.

It does not really matter why or when or how we see the words "coca cola". All that matters is that we see those words more. Every day, more times per day, more places, more people. We need to be thinking about them always.

That is what they're paying for, and it doesn't even have to make sense! That might not even get them anything, but they have more money than they can possibly spend, so they may as well spend it on trying to get us to look at and think about them.

Can it be more optimal? Could you squeeze blood from this stone? Sure, probably, but whether I'm on twitter reading great video-game recommendations or reading flame wars is a pittance compared to whether I'm online or not at all.

I wouldn't be surprised to find out that the optimal situation is to make users cycle in and out of the outrage content: get me mad so I stick around a bit longer, cool me down with something funny/sweet, then show me an ad, then get me mad again. Repeat.

> Engagement costs us money. It increases system load and the resources required to actually run the platform.

this is absurd. It's absurd! It really makes me question your intent and good faith.

The platform is designed to be used, by definition! What, are you telling us twitter does not want its users online? You're saying that engagement is "expensive", is twitter running some kind of other service? Something outside of users looking at and posting tweets? This is nonsensical to the point of being upsetting.

It is like Philip Morris fretting over all the tobacco they're going to have to pay to harvest so that people can keep smoking. If only they didn't have to! You made an advertising platform and people must be present on it to be advertised to. End of story.

Now, of course, there are costs to be optimized. A dumb example: if you know you can't show me an ad between now and when I will naturally end my session, you should maybe try to get me offline right now, before I download any more video data, but that stuff is just sand on the beach for the difference it makes.

Maybe you think this matters because it is the work you do, or it's work you know does get done, but it does not matter in the big picture.

Side-rant: Twitter could fire 95% of its employees tomorrow and probably be more profitable than it is today in the long term, but it won't, again because of cargo cults and social wisdom.

Almost none of the feature work that has been done by twitter in the last 8+ years has materially mattered: the product is done, and you're mostly paid to arrange deck chairs on a ship that's sailing along smoothly and to look pretty. Employee headcount, profit, ad impressions: at the end of the day, all of this is just disembodied marketing metrics for investors to speculate on and to try to extract cash from before the whole thing tanks out someday.

> Raising engagement for the sake of raising engagement is a net negative in the long run, because it results in too much engagement that doesn't sell products (and thus ads).

This recap of your argument is just a rehash, but it restates the basic flawed premise in full:

"Advertisers are rational and have perfect information and will pay only for the optimal conversion strategy up to the optimal point of cost/benefit, so if we ever over or under-shoot at all, we'd lose money and we'd go right out of business!"

So, let me recap as well:

- Twitter is an advertisement platform.

- Twitter relies on dollars spent by advertisers. We agree so far, I think.

- Advertisers pay both per impression and per click.

- Advertisers have bad information, and act on social wisdom as much as or more than anything else. They cannot evaluate your performance by how well their advertisements do on your platform. This right here is the secret sauce!

- Engagement (i.e. time spent paying attention to the platform) may not raise conversion rates, (i.e. clicks per impression), but higher engagement is strongly correlated with more conversion overall. People gotta be online.

- Engagement is the core metric of your business.

To put this all very simply: if I am on twitter all day, it's a hell of a lot more time for me to run into and to think about Harry's Razors than if I never went on there at all.

That is the product you are selling to advertisers, and they will pay for it for a long time, even if it doesn't really make any sense or money.

I have no idea about Facebook. I barely even use it so all my information is second-hand.

> Why is Twitter encouraging outrage-engagement via the (somewhat filtered) Trending section?

We're not, at least not intentionally. Do you think we are? If so, why?

The Trending section is still biased towards things that are being talked about frequently. Unfortunately, that means outrage-inducing topics will tend to bubble up there. We're working on making trend identification more sophisticated in order to provide better context, but it's a really hard problem with such short snippets of context.

> Why aren't you actively trying to decrease political engagement if it's a net negative?

"Political engagement" overlaps with, but is not synonymous with, "outrage engagement". Outrage engagement is a net negative; political engagement that does not devolve into misinformation and toxicity is not. We are actively trying to decrease engagement with misinformation and outrage, but it's a hard problem. Also, frankly, it's being hampered by a lack of focus internally that is leaving the teams involved with no clear direction other than "do something, now!".

Non-toxic political engagement is a net positive for us. We don't want to throw the baby out with the bathwater by shutting it all down, assuming we even could (it's fundamentally the same problem, though a little bit easier). I see how other people might see that as the better option to shut down the toxicity and misinformation, though.

I haven't been an active Twitter user for years, but whenever I visit it these days, Trending looks like it's at least partially optimized for outrage. I understand it's not actually "what's popular" (because otherwise it would constantly be about Justin Bieber & co), it's curated, or at least topics need to be approved in some form to be allowed to trend, right?

> Non-toxic political engagement is a net positive for us.

I've seen very few non-toxic political engagement on Twitter, but maybe we have a different understanding of what is healthy engagement. I'm not sure there is a baby in that bathwater. I'm not even sure it's bathwater.

From my perspective, it's either a shouting match with a lot of groupthink and -speech or somebody talking to their followers about how evil the outgroup is. And that's not because I'm looking at controversial things, I believe. I visit some developer's Twitter feed and one or two clicks later I'm in the middle. It's like that Wikipedia game where you start on a random article and need to make it to Hitler with the least amount of clicks, but it's less challenging because almost all clicks lead to outrage.

You seem to be more focused on whether what engages the users is factually true. Is it that advertisers don't want to be seen next to obvious fake news? Do advertisers not care to have their brand associate with not-actually-honest-but-not-fake-news-either outrage-engagement by, say, Robert Reich?

Brilliant. I'm swiping it.

Can you talk more about this? On its face, this seems like an explicitly anti-leftist position. Some philosophers even claim that the role of the political Left is the organization of societal rage. For instance, if Twitter had its way in the domestication of political speech on its platform, then it might have suppressed conversations about BLM and prevented it from having a chance to become mainstream.

Unfortunately, there still aren't enough of us. We're trying to change that.

No, we really don't. In fact, we have mechanisms in place to keep ads away from certain categories of content. Do you have any evidence to support your statement?

I'd like to clarify what actually happened here (throwaway for paranoia).

The actual bug was that the account signups flow was accidentally modified to automatically opt out new signups[1]. When Legal reviewed the ads targeting and discovered the opt out setting wasn't being respected, they made ads start doing so. Two this later they discovered the sign-ups bug.

[1] I know, a bunch of people would rather it was always like that.