I do not want to see an unskippable 60 second ad for a skincare product I do not care about whatsoever in the middle of a video about replacing the cambelt on a 90s French hot hatch. I especially don’t want that ad to bisect a word or sentence.

At least try to show me something that might have some passing relevance to what I’m watching, will you, please?

Edit: I'm not familiar with data on context based ads but I'm very skeptical they are significantly better in the general case. They are already used in things where it makes sense like when you're searching for something.

Your browsing history gives a more reliable base to segment you based on buyer profiles (incl age groups, location, interests), figure out your "intent" and target ads based on it. If you were to, say, read a random "Top 10 cars with highest resale value" article, on its own without historical data it won't be of any use for targeting because they don't know if you're actually a potential buyer in the market or just some teenager passing their time. Showing you those ads will waste their $$ if it were the latter.

This isn't in any way an endorsement of their intrusive advertising practices, by the way - I personally have been using ad blockers and aggressively taking every step possible to avoid all online advertisements for more than a decade. It's just to provide a perspective on why it's not so simple.

The marketing side of the business is very data driven with lots of very intelligent statisticians and scientific testing for ad placement and ad content, etc. I cant accept that the same people that are manipulating my thoughts and desires with algorithmically optimized content never once thought to run hypothesis test on performance of targeted ads based on browsing history.

I feel like you are making a bold claim, am I misunderstanding?

Also, don't try to make me feel guilty for having an "ad blocker". I don't specifically block ads, instead I have a "tracking me without my consent blocker".

They require more effort on the part of the site that produced the content, but are much more lucrative for that site.

Since most of the internet has been low-effort algorithmic slop for the last twenty years, tracking ads are more popular. They let low quality sites “steal” audience impressions from higher quality sites by displaying ads to people that happened to visit both sites.

I think personalized ads / algorithmic targeting (and even collecting the datasets that enable it) should be banned.

at any rate - not only context aware ads would be better privacy-wise but also probably would be way more lightweight…

It can be thought of a good way of branding on a cohort of customers who would be interested in your product.

Just the definition of interest is something which is skewed.

And this number is produced even with the edge case you brought up! Targeting is just that good.

Advertising is also not just product advertising, as in, "we would like to purchase this exact product". Advertising spaces are also contested, so, if one brand doesn't buy it, maybe a competitor will. Advertising also increases mindshare - you might not buy another new car of course, but people are still influenced by what they see. Brands are also bolstering their image with ads, regardless if you particularly buy or not. They are associating situations, lifestyles, emotions with their brand.

What I'm trying to get at is incentive. The incentive is huge, and measurable, from the advertiser standpoint. And so, we will never get rid of targeted ads, unless we legislate, and enforce.

So... this intuition is wrong.

Across... well, basically every category of product... the product which you are most likely to purchase next is the same (or a substitute for or a complement to) the last product you purchased. Anyone who has ever worked in retail analytics will tell you this.

Advertisers want to minimize their ad spend, so they always try to sell you first on the product which you have the highest propensity to buy. That's why they want personalized ads. The ROI is astronomically higher than for contextual ads.

(That's why the DMA's prohibition on Facebook's pay-or-consent model, and HN's general cheerleading of it, is such a joke... and that's before you even get into the adverse selection problem of people willing to pay to avoid ads)

There are plenty of greedy people in business. Rest assured: If cost of personalized grew to the point that ROI dropped below contextual ads, advertisers would switch to contextual ads in a heartbeat. They're A/B testing all the time, and it wouldn't take long at all for them to figure that out.

The level of identification and tracking possible today is scary even across devices.

In practice, that's most of them. Privacy Badger by itself is an OK ad blocker.

Also, you can easily tell Privacy Badger to block sites.

Privacy Badger warns you that blocking certain domains, such as Google Tag Manager (otherwise known as Google Backdoor Hostile Javascript Injector) will break some sites. In practice, this seems not to be a problem. I've had Google Tag Manager blocked for years.

That's what Unlock Origin is for. I don't know if they intended it this way, but seems like they complement each other quite well.

This is incorrect. It is often relevant, and it is reflected in financial performance of personalized ads. Companies aren’t doing it for fun. They’re doing it because it is wildly profitable and that’s because people respond more to ads that are aware of that “past”. You may not have access to the “proof” but they absolutely do.

Advertisers have an idea who they want to sell to but they don't know their browsing habits.

The ad industry collects data about you to see whether you are like the person particular advertisers want to sell to and what sites you visit. With this information it can display ads relevant to you on the sites you visit.

You propose that ads shown should be conditioned just on the content of a website, but you are missing out on the fact that the content on any one website alone is a poor proxy for the type of customer that the advertiser is targeting.

It feels like the algorithm is saying "oh, they bought a mattress... they must really love mattresses and want more!", when much better ads could be suggested with the wealth of data they have on shopping habits.

That we don't see that happening at all is pretty strong evidence that personalized ads are more effective than contextual ads. I find it highly unsurprising that ads which are based on the history of a person are more effective than ones based only on their present. If someone was looking for a car last week, odds are that they are still looking for a car.

This generates a lot of false positives too - if I have bought something and I see an ad about it, I may still click at it out of curiosity but without any intention to purchase it. (And rarely has an online ad or copy induced me to purchase something again that I just bought). So Ad networks do have an incentive to keep showing me ads that I have clicked, whether it would convert into sales or not, because they make money from these clicks. I think that's why "personalisation" matter to the online ad industry - not because it increases conversion and makes more money for the advertisers, but because it does increase revenue for the advertisement platform.

It's something you do only once every 3-8 years. Targeting you, who just bought, wrongly, is better spend than targeting me, who isn't interested at all...

A good marketing team has a pretty sophisticated reporting pipeline. A bad one is doing a lot of misattribution.

> the idea that personal browsing history is a good indicator of the future

Its a good way to build a profile of a customer, and even mundane things can be connected together into interesting data conclusions... This was almost 15 years ago: https://www.forbes.com/sites/kashmirhill/2012/02/16/how-targ...

The reality is this, your consumer behaviors can be VERY predictable. No one wants to know that their ghost in the machine baahhh's like a sheep following the herd.

No we were slowly forced by Google to surrender to their algorithm that use tons of data on users.

I wish it was still easy to just put an ad for car when people search « cheap car », but the reality is that they keep removing features, making things convoluted and on the other hand push hard for their automated algorithm.

I would think that kind of ad is a lot more effective than an authentically selected ad based on your browsing history.

Some ads are masterfully made, but even they distract me, not attract. They jump into the view, they suddenly break the page flow, they strive to be clicked by mistake. Tis is especially insufferable on mobile. They clutter the screen and obscure the real content. They eat bandwidth and battery life by loading tons of content I did not ask for. They play whole videos, some are so impudent as to play sounds. They are consciously created as an impediment to reading (or sometimes watching) the content I came for! Isn't it the definition of being actively harmful?

Then, of course, they are mostly not relevant, like, 99.7% of the time. To quote: «A general trend that the advertising business is not interested in delivering ads to the people that want the product. Their real interest is in creating a stratification of product offerings that are all roughly as valuable to the advertiser as the price paid for them. They have to find ways to split up the tranches of conversion probability and sell them all separately, without revealing that this is only possible by selling ad placements that are intentionally not as good as they could be.» (https://news.ycombinator.com/item?id=42721611)

And they usually offer no way to say: "Hey, this is not interesting, show something else". Instead, Reader View does away with everything irrelevant in one click.

Interestingly, one of the places where I do not block ads is Facebook (which I use sparingly). It knows everything about my profile (obviously), it shows ads moderately, and in a way that's not infuriating, and it even actively asks my opinion about ads. As a result, I taught it to show me highly relevant ads: about electronic music, vintage computers, etc, stuff that I actually would be interested in clicking. This might look anti-privacy, but the tracking is mostly limited to the Facebook content, because I cut off third-party cookies.

You're claiming they don't have any real proof, but you yourself are not providing proof of that. On the contrary, I assume there's tons of proof (data), whole oceans of the stuff, because personalized advertising goes far beyond just checking if you searched for a car, then showing you car ads.

Instead its: if you searched for marriage related stuff, a year later they'll start showing you baby stuff. If you searched for "why is my husband so..." they'll start showing you ads for divorce lawyers. If you search for "why are there no jobs", they'll start showing you extremist political ads about immigrants stealing all the jobs, and on and on.

This stuff, personalized, designed to manipulate you and hit you at the times when you're emotionally vulnerable, does work. Of course it works. Humans are easy to manipulate if you know their private wants, needs, and emotional state.

https://www.eff.org/deeplinks/2023/09/new-privacy-badger-pre...

UBO is a request-level filtering system. It blocks certain requests based on a set of patterns. It's incredibly simple, incredibly fast, and surprisingly effective, since most adds and trackers are served by 3rd party sources that can be recognized. This doesn't catch everything, though, and trackers can be sent alongside the core website content. PB provides content-level filtering that can catch some things that slip by UBO.

For more see https://privacybadger.org/#How-is-Privacy-Badger-different-f...

That wiki page is a bunch of nonsense. For example:

>Redundant with Total Cookie Protection (dFPI)

https://privacybadger.org/#Is-Privacy-Badger-compatible-with...

Those trackers, such as Facebook and Google, aren't loaded at all, so they are unaware of the request that was not tracked.

What you are advocating is loading those libraries, etc., anyway and allowing them to have their way with your browser session. This will always be less private than not doing it. Even Tor Browser has all sorts of protections from these types of things in place, which you would need far less of if you just blocked these tracking libraries to begin with.

Yes, theoretically, my blog or The New York Times could start profiling the missing requests and send them over to Facebook through the back-end, which is what is referred to as 'server-side tracking' in the industry, as far as I know. However, the chances of most websites doing this are slim, as it requires at least some effort on the server side. The way these websites usually do this is by passing along the account information they have on you, such as e-mail addresses, phone numbers, etc. Even if you signed in on some site with Tor, they'd still send those things along if they had gone through this trouble.

Ironically, even Tor relies on clearing cookies, disabling JavaScript, and blocking specific requests to protect your identity, not just the origin obfuscation. So, the thing you are claiming makes it easier to track you, and suggesting that Tor is the solution is somewhat at odds.

…OK, looks like LocalCDN could do this (e.g. with a Firefox extension), anyone tried it?

LocalCDN - https://localcdn.org

I forget which level of blocking it was applying; some cookies it just keeps from being cross-site, it isolates them. Others it blocks entirely. You can easily adjust which it is doing for any given cookie.

I think it's true that if you have uBlock Origin you probably don't need this though, that seems likely. I don't run uBlock Origin.

For more, see https://privacybadger.org/#How-is-Privacy-Badger-different-f...

- https://privacybadger.org/#Is-Privacy-Badger-compatible-with...

- https://privacybadger.org/#How-is-Privacy-Badger-different-f...

For more, see https://privacybadger.org/#How-is-Privacy-Badger-different-f...

Just FYI, you can always disable Privacy Badger on a particular site by using the "Disable for this site" button in PB's popup.

You can also help make PB better by using the "Report broken site" button in PB's popup.

For more, see https://privacybadger.org/#How-is-Privacy-Badger-different-f...

Probably a good benchmark if you are developing standard cookie consent dialogs is whether they work with this.

Just FYI, you can always disable Privacy Badger on a particular site by using the "Disable for this site" button in PB's popup.

You can also help make PB better by using the "Report broken site" button in PB's popup.