Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Instagram ads Facebook won't show you (2021) (signal.org)
392 points by nixcraft on Dec 7, 2022 | hide | past | favorite | 89 comments


It looks like facebook disputed this story

> Joe Osborne, a Facebook spokesperson, responded on Twitter on Wednesday saying the screenshots are from early March “when the ad account was briefly disabled for a few days due to an unrelated payments issue.”

> Osborne added: “The ads themselves were never rejected as they were never set by Signal to run. The ad account has been available since early March, and the ads that don’t violate our policies could have run since then.”

This is from: https://www.cnbc.com/2021/05/06/facebook-and-signal-are-figh...

Either way, it's a great publicity stunt by Signal.


I think it'd be helpful for discussion to include Signal's dispute of Facebook's disputation of the story in your comment:

> We absolutely did try to run these. The ads were rejected, and Facebook disabled our ad account. These are real screenshots, as Facebook should know.


Are there other screenshots I'm not aware of that weren't posted on the Signal blog (as their response uses the plural)?

The only screenshot I can find shows a disabled account (not any rejected ads), and also seems to support Facebook's side of the story (you can see a banner in the background saying something about their "current balance", which would imply what Facebook said was true).


Facebook admits that they would have rejected a couple of the ads. See the grand-parents link.


Facebook claims it would have rejected ads that "assert that you have a specific medical condition or sexual orientation", which seems like a fair/reasonable policy to me, and accepted the others.


To be clear, the "reasonable policy" is "don't let your ads reveal how invasive our spying is"?


> Facebook claims it would have rejected ads that "assert that you have a specific medical condition or sexual orientation"

If I'm reading the picture correctly: Facebook collects and processes data about our medical conditions and sexual orientation. We've seen how FB allows ads targeting those metrics to be constructed - but FB also indicates it might get grumpy if those ads are posted.

Further, FB is not revealing who it makes that data available to or even explaining why it collect+processes our Med and SO data in the first place.

Does anything I've said here seem unlikely?


Your line of reasoning would make a lot of sense if we ignore the fact that Facebook (at least claims they) would allow the other ads.

It seems likely to me that rather than wanting to hide these specific categories (but be willing to expose every other type of data collection they do on you), that Facebook doesn't allow these specific two categories because of the significant negative social/legal consequences if they get them wrong.

They are very likely trying to avoid gay users getting an ad that says "You're straight, join my Straight only dating site" or an ad incorrectly diagnosing users with a disease they don't have.


The article cites Signal's response, but does not link to the tweet, that contains one additional screenshot: https://twitter.com/signalapp/status/1389732689096450048h (nothing new though)


You should record a loom and run the ads. Pretty hard to dispute hard video evidence.


Oh right, we should definitely expect to see more ads like this then... uh huh.


Interestingly, there are a lot of ads that are like this… just subtler.

If you’ve ever seen tee shirts and similar products in ads that say things like “This Kansas City redhead likes their coffee with whiskey” or some other uncanny specific junk, thats how this works.

There is a whole new category of products based around these ads. I don’t know what they’re actually called actually, but it was on HN sometime in the last year. They’re usually tee-shirts and other “made on demand” products, and they use the magic of software to generate images for the ads, and products that produce to a ton of permutations, and target ad demographics that fit all of description. The products aren’t real until someone clicks the link, and then software knows which ad you clicked and generates it for that permutation.

The whole idea that the un-canny match will be enough to attract buyers for the novelty, and they eschew the inventory risk by just printing on demand.


I've got a bit of experience in this area.

I built an ad engine for Kink.com (NSFW) which had ads placed on even higher traffic sites like PH. I built the original pixel tracker for Marin Software, and @stickfigure and I built GearLaunch, which was the backend that sold those t-shirts you're talking about.

Needless to say, you don't see as many t-shirt ads any more because FB cracked down on it... not because of the targeted ads, but because it became a copyright infringement game and it was easy to block the people creating the ads (and shirts) as they were all outside of the US (ie: Vietnam). That and thankfully FB didn't want their wall to be all ads for t-shirts, lol.


You can at the same time be against the ads and against misleading publicity stunts.


Why are you accepting Facebook's PR narrative as true?

From the article (https://www.cnbc.com/2021/05/06/facebook-and-signal-are-figh...):

> Signal countered on Twitter that it “absolutely did” try to run the ads. “The ads were rejected, and Facebook disabled our ad account. These are real screenshots, as Facebook should know.”

I'm not saying Signal's side is necessarily true, but you're making an affirmative claim that Signal is performing a "misleading publicity stunt", by taking faith on Facebook's narrative as true.

*edit: posted the wrong link


History has shown me to expect that Facebook PR is actually more likely to be untrue than true.

But that started at the top.

https://mashable.com/article/mark-zuckerberg-lying-about-fac...


Great article. Thanks :)


It isn't misleading though. It is exposing what really happens under the covers.

They didn't have to run the ads or even be truthful about being blocked. Fact is, they would have been blocked and the ads are indeed targeted. All of the details in those ads, are true.

Even if FB doesn't collect the data itself, any time you shop online, you get added to buckets and those buckets get shared when the seller uploads the target data back to FB and then the process starts all over again.


“the ads that don’t violate our policies”

They would have denied the ads anyways. The cited FB spox is news not for adding facts but rather for its laden speciousness.


Are we really at the level of hate for Facebook/Meta where people are justifying just making stuff up about them?

"The accusations are untrue, but I think they would have been true, so that's basically the same thing."

Complain about these ads being banned when it actually happens, not when Signal lies about it because you think that's what Facebook would have done.


Why do you believe Signal is lying about it, and Facebook is not?


If Facebook was lying here, you'd think Signal would speak up about it. It would be a HUGE PR win for them, I can already imagine the headlines ("Facebook denied their ads, banned their account, and then lied to cover it up"). It would be extremely easy for them to provide a screenshot of an email from Facebook or similar saying the ad was declined, so why don't they?

Secondly, while I'm no fan of Meta/Facebook (and think they're a net negative for the world), I trust them way more than Signal to get basic facts right in their statement. Facebook is a massive corporation with lawyers and PR people who know that outright lying about something like this is a horrible and possibly illegal idea.

It seems pretty obvious to me here that Signal expected Facebook not to comment on this and for them to net a juicy PR win. On the other hand, Facebook has no reason to believe Signal would stay quiet after being accused of lying if they had a shred of evidence to the contrary. Unfortunately Signal achieved their goal anyways, because we still have people like choppaface coming out of the woodwork to defend them because "Facebook would have done it anyways".


They did speak out about it, which is in the same article linked above that sources the FB disputation statement (https://www.cnbc.com/2021/05/06/facebook-and-signal-are-figh...):

> We absolutely did try to run these. The ads were rejected, and Facebook disabled our ad account. These are real screenshots, as Facebook should know.

I don't think your logic about companies lying really tracks with the historical record. Companies, when they believe they can get away with it, lie all the time.

Again, I'm not saying that one organization was certainly lying or not-lying; I'm just challenging your presupposition that Signal was lying.


Copy pasting what I sent in another response:

Signal provided no new evidence of this claim, and the existing evidence they did provide seems to support Facebook's side of the story (as I explained more here: https://news.ycombinator.com/item?id=33900582).

I just don't see the motivation for Signal to hold back this evidence if what they're saying is true. It would be so easy for them to go and screenshot an email or page saying "this ad is rejected" - why don't they? The only way it makes sense is that they don't have this evidence, because they're lying.

> I'm just challenging your presupposition that Signal was lying.

I'm not presupposing anything, I'm looking at the available evidence and the actions of the actors involved and coming to a conclusion. If anything, I hate Meta/Facebook, I think they are a net negative to the world, and I probably came into this with a mental bias against them.


You're presupposing because you're coming to a conclusion that is not justified by the evidence.

The absence of evidence is not evidence of absence.

> The only way it makes sense is

I don't think you've provided convincing evidence of that.

Are the following all not possible?

1. They are unable to access that account or page any longer

2. Facebook chose to lie about this in a way that is not easily disproved by a screenshot.

3. Facebook disabled their account because of their ad content, but chose to put up a message that was more ambiguous.

*edit: formatting & make messaging more polite


> Are the following all not possible?

Correct, none of them are! When I last used Facebook advertising at a prior employer, they sent you an email when your ad was rejected informing you of this fact. Assuming that is still true (and I don't see any reason they would change this), then Signal should have undeniable proof about the claim that Facebook rejected their ads (which Facebook claims they did not do).

Of course, I strongly believe we will never see this evidence, because I think that Signal is lying and thus it doesn't exist.

The fact that Signal has not produced this evidence is strong evidence they are lying, because they have nothing to lose and everything to gain from doing so. It would also be extremely weird for Facebook to lie about this, given they would know such evidence exists and Signal could easily disprove their lies.


How is the following not possible?

>Facebook disabled their account because of their ad content, but chose to put up a message that was more ambiguous.


We can fairly reasonably assume whoever is telling the truth is doing so about both things. Signal should know whether or not their payment details were actually having issues or not.

Now, if you want, you can believe that Signal is lying about whether or not the ads were rejected, wrote an extremely misleading blog post that strongly implies the suspension was permanent when it was temporary, but has correctly guessed that the real reason for the (temporary) suspension was their (unsubmitted) ads, and that they wrote this blog post rather than trying to submit their ads for the first time after being unsuspended.

But at that point, I would say that you are "presupposing because you're coming to a conclusion that is not justified by the evidence". You are so strongly biased that Facebook must 100% prove their innocence on every single point, but are just automatically assuming Signal is telling the truth with no evidence.


Signal did speak up about it, and flatly denied Facebook's statement.


Signal provided no new evidence of this claim, and the existing evidence they did provide seems to support Facebook's side of the story (as I explained more here: https://news.ycombinator.com/item?id=33900582).

I just don't see the motivation for Signal to hold back this evidence if what they're saying is true. It would be so easy for them to go and screenshot an email or page saying "this ad is rejected" - why don't they? The only way it makes sense is that they don't have this evidence, because they're lying.


These ads follows along in the proud tradition of banned ads that were designed to be banned from the onset because saying "we got banned" is useful for positioning yourself as dangerous to the establishment.


Which doesn’t exactly invalidate them.

Sometimes manifestly unjust laws are challenged via an index violation of the law with the specific goal of getting the law thrown out.


Sure, totally. I wasn't suggesting it's invalid, just that sometimes "we got banned, oh no!" was the plan all along.


My mistake for misinterpreting. I agree!


That is untrue.

If Facebook were not shy about their algorithmic approach there would be nothing wrong with these ads.

Not so much "...ads that were designed to be banned..." but ads designed to test an hypothesis.


> ads designed to test an hypothesis.

Just so we're clear on this: Experimentation on human subjects without consent and informing them is an ethical violation. It truly doesn't matter how justified you think your particular research question is, or how small you would rate your ethical transgression.


Then you agree that Facebook shouldn’t have run psychological experiments[0] on its own users to manipulate their emotional state?

[0] https://www.wired.com/2014/06/everything-you-need-to-know-ab...


By your reasoning, only ads like this should be permitted, since they're the only ads that explicitly inform the user that they've been targeted.


It is not "Experimentation on human subjects" to inform them of the things that drive ads to them.


Meta: I'm familiar with "an historic", but I've never seen "an hypothesis" before. Is "an before h-" a broader grammatical pattern than I realized?


Yes

>In English, the pronunciation of ⟨h⟩ as /h/ can be analyzed as a voiceless vowel. That is, when the phoneme /h/ precedes a vowel, /h/ may be realized as a voiceless version of the subsequent vowel. For example the word ⟨hit⟩, /hɪt/ is realized as [ɪ̥ɪt]

For words beginning with h and where the first syllable is unstressed (historic, hypothesis, but not hitter, happy) it can be appropriate to use "an" rather than "a" as the associated indefinite article.


Isn't this an american english vs british english thing? I was puzzled by "an historic" for a long time and then I heard a british person say it and it made sense. Because personally I pronounce the h in historic, hypothesis, hitter, and happy virtually identically.


Native American English speaker here (from northeast/mid-Atlantic, but living in NorCal for 20 years now), and I generally mostly drop the "h" in "historic", such that "an" sounds reasonable. Agree that dropping the "h" from the other words you mention would feel weird, though.


I’m in the PNW, so a fairly different accent. If I say “an historic” the h gets minimized, but “a historic” sounds perfectly natural to my ear and “an” sounds wrong with all the other example words.


Mid-Atlantic here. I'd say "a historic" with the harder 'h' sound.

"an historic" sounds to me like "anhistoric" (as in "lacking history" kind of like "amoral")


The word is "ahistoric," and ask yourself why you thought it was "anhistoric"... it's related to the reason why some say "an historic." ;)

https://www.merriam-webster.com/dictionary/ahistoric


Language is so messy.

I say "a historic" and "ahistoric" differently, I think. The first with a hard 'h' and a slight pause between words. The second with soft/no 'h' and no pause. If I heard somebody say "anhistoric" (soft 'h', no pause), I'd think they meant "ahistoric" not "an historic". Of course, context probably makes it clear regardless.

Growing up in DC with Scottish parents was always fun. Rolled Rs, different "wh" sound (Scottish English: wheels vs US English: weels) (noticeable expulsion of air on the 'wh' sound).

Mary (lady), merry (happy), marry (get hitched) all should sound different to me, but in MidAtlantic English they're often the same. Same for Kerrie vs Carrie.


Native BE speaker here, I've seen and heard "an historic" (like ahnistoric) from posher BE speakers than me, but never "an hypothesis" (ahnypothesis?) from whichever social class ... possibly I hang around with the wrong crowd?


Yes, you have been hanging out with the wrong crowd: https://youtu.be/BZLFIRkPmcE


An reason to adore that man. Great video


All the words in this thread are coming from Ancient Greek (hypnosis, history, homo, hoplite, hero etc). In Ancient Greek these words don’t start with an h (υπνοση, ιστορια, ομο, οπλίτης, ηρωας etc).

In Ancient Greek polytonic orthography, vowels can have one of two diacritic marks: οξεία and δασεία. The second one is used to denote the /h/ sound (rough breathing). Words starting with vowels with this mark in their translation to English added the H before the vowel to denote this sound because of the lack of diacritic marks/accents

https://en.wikipedia.org/wiki/Rough_breathing


Anecdata point of one, but I voice the H on hypothesis. Im a native American English speaker, but from the northwest.


"Voiceless" in this context means you dont engage your larnyx/vocal cords (you don't create a pitch), not that you don't sound it out. Compare a whispered "ahhh" - voiceless - vs the "ahh" you say at the doctor - voiced.


My guess would be people who primarily write, not speak, English. 'h' is a consonant after all, the fact that its like that here may not be apparent to people who write more than they speak English (as a second/third language)


The only example I had previously ever heard of was "an hour" because the H isn't pronounced. As far as I've ever heard it spoken the H in both hypothesis and historical are well pronounced so "an hypothesis" sounds jarringly incorrect to me.


The 'n' is really to help the flow as you are talking, so it depends on if you pronounce the 'h' or not. In british english you would nont say 'an historical' or 'an hyphothesis' since both are pronounced with the 'h'.


> but ads designed to test an hypothesis.

I guess I don't really agree. The goal of an ad is to get new customers, not test hypotheses. If it was designed to test where the edges are, they wouldn't be blogging about it. I believe they tested the edges so they could find something that crossed the line, and then get publicity for being banned. But that's a question of intent, so who knows. Maybe they really thought this would be an awesome campaign, and are making lemonaid out of it getting banned.


Or it's possible that it was a hedged bet. If the ads were run, they would have an ad campaign that showed how much Facebook knows about you, if they were blocked, the would be able to write this blog post.


I always feel like I'm the weird one out. The best thing would be no ads. But the second best thing would be hyper-targeted ads because the most annoying thing is ads that are irrelevant to me.

When I watch broadcast TV, I hate seeing ads for mattresses, kids toys, menopause medication, home appliances.

I want to see ads for camera lenses and cool sneakers.


Currently facebook has decided I want to see ads for showerheads that cost over $150 and require the use of an app. The first time I saw one I left a very smarmy remark in the comment section about how the last thing I want to buy is a showerhead that pushes content marketing at me through a newsfeed. Facebook has apparently determined I passionately engage with showerhead related ads so that's all I get now.


I don't want to have ads as, and if I have to see I don't want them to be hyper-targeted. I don't remember one time a ad had made me buy something. If I want to buy a new device I look his spec and some review.

I understand that unfortunately ads are here to stay, but I'd like to see more context-based ads (like on ddg), they seem to be a good compromise for me: they still serve ads, I can still ignore them and they don't profile me.

Also I really dislike the idea that some people are ok or even prefer being tracked across the internet because they want to see ads for some whatever thing they like.That is some dystopian level thinking.


I also always feel this way, glad to hear someone else express the same. I'm tired of seeing ads that aren't relevant to me despite most advertisers knowing who I am and where I live.


The only ads I've bought things from were ones that were hyper-targeted to my demographics, like seller of sweets from my home country in my current city.


I’d tend to agree. Except for the fact that to create good targeted ads it takes tracking my every click and move. And that this can be sold to parties that I’ve never interacted with nor (actively) consented to having my information.


I disagree 100% that hyper-targeted ads are second best to no ads.

For me, it's:

1) No ads

2) No ads

3) No ads

I accomplish this via a bevy of ad blockers, VPN, and custom DNS settings


so do I. no ads is best. but some ads are native to the medium and unavoidable


Note that the entity running the ads, in this case Signal, still has no way to identify _who_ was shown the ad. Only Facebook knows that.


Are campaign IDs not a thing for anyone who clicks through? Or simply have a different landing page for each campaign?


These ads intentionally identify the parameters but the point they’re making is that you can show these extremely targeted ads and put a lucrative offer that requires entering email/name/phone and voila you’d have unmasked the user.


This kind of seems like saying 'All a burglar has to do is break into the house and they can set up surveillance in your house'. Yeah sure, that's technically true, but if they're in your house, they can already do far worse.

If somebody is willing to go to some scam website and enter arbitrary PII, they've already been compromised, regardless of Facebook's involvement.


More like someone can break into the place storing your things that have already been stolen from you.

We don’t want advertisers unmasking us but Facebook shouldn’t have this data in the first place.



Thanks! Macroexpanded:

Instagram ads Facebook won't show you - https://news.ycombinator.com/item?id=27040470 - May 2021 (177 comments)


Context: from Twitter[0] - "When we made Instagram ads to show you just how you got targeted, we got our ad account disabled. But now Facebook is putting it right out there ~ https://facebook.com/ads/about/ "

[0] https://twitter.com/signalapp/status/1600559887742009345


I did something similar, years ago, for twitter ads, using autogenerated videos. They didn’t block me, although not sure it would work today:

https://thenewinquiry.com/taxonomy-of-humans-according-to-tw...


I don't know how this is new in any way.

This ad [0] was created 6 years ago showing more than the ads in the article.

[0] https://www.youtube.com/watch?v=yrjT8m0hcKU


The thing is I did see the ads, lots of times ... but it go nothing correct, and really blunted the whole impact.


Even though this is old news, I’m curious to know if any of those “attributes” were embellished. I’ve run FB ads myself and wasn’t ever able to find that granular of targeting without deeper relationships, like admin access, with other pages - looks like groups required that if I’m correct?


I've worked at a company which ran a looot of ads on fb and it was definitely this granular, it was an eye opener for me


I'm curious if it was effective and worth the trouble?

As a consumer, I don't click on ads at all, I don't watch video ads, and basically don't impulse buy anything.

And working with companies over the years, our marketing departments didn't really know how or why to do this level of targeting.


> I'm curious if it was effective

I don't think it is, it's mostly bot traffic anyways...

> and worth the trouble?

...but companies are paying $$$ for it and it's very easy to setup/run their campaigns and give them the "engagement" reports they ask for. I think everybody with half a brain understands it's a big game of moving money with very little real world/real people engagement


Yikes.


I've always had trouble with this publicity stunt. Like banning people for baseless political or personal reasons is bad, but this is directly insulting Instagram. In effect it's saying "here's the information Instagram has collected on you, now use our platform which doesn't collect this information". Like why would Meta take that reputational hit for a few thousand dollars in the short term. It's a bad deal for them and it makes sense to not do it


I don't get it, how is revealing the truth about somethning insulting all of a sudden? The displayed ads don't judge, they're shock value is generated solely by what they reveal about instagrams business practices.

It's obviously not in Metas best interest to be open about their data collection practices, but that's the entire point of this campaign, that Meta/Facebook/Instagram can only exist with intransparency.


How unethical does an action have to be before this rationalization no longer stands?


Depends; is someone promoting it to make money? Because signal isn't trying to make facebook better, they're trying to convince people to use signal instead of messenger using some ethically dubious tactics


the fact that anyone who is that terrible would logically not want you to know that they're terrible doesn't really change the terribleness just points out that they're aware of how terrible they are.


I might come off a bit naive, but I didn’t picture the data collection being _this_ precise.

Now I’m interested in my own data. Are those companies required to offer a means to download your own data?

Did anyone do that and if so, was it readable?


I wrote a generator for these if you want to make your own.

https://tacix.at/experiments/signal




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: