The way you’ve described it is basically how it would have to work.

Various ad-blocking detection technologies basically boil down to loading some element on the page and then querying for it during/after rendering to see if it’s still there. This could be combined with an AJAX call to load the actual content, which is how all those annoying sites work that pop a nag up in your face if you’re running uBlock or whatever. And even then you don’t get the content even if you subsequently block the nag notice.

A truly undetectable adblocker would still have to pull down and load all the ad content and render it somewhere (invisibly in the background, presumably) and then serve a second cut down version of that page with the ad elements not rendered.

Edit to add: This would be somewhat detrimental to the user, because it would by necessity not stop the types of tracking that are typically built into served ads. Current adblockers (like uBlock Origin) also by default also block various advertisers’ nonsense like cross-site tracking cookies and tracking pixels, etc.

The problem isn’t so much blocking the ads on a page, that’s a solved problem, it’s doing so without incurring side effects. The main problem usually comes in two ways.

1. Ads are now being pre-baked into the content delivery itself in which there is no easy way to rip it out without destroying the content in some way. Twitch is notorious for this on streams where the ad portion completely replaces the video feed before your browser ever sees what was originally there. You may never recover what was there, but if you try to block the ad playing you trigger problem 2.

2. There are departments dedicated to developing ever changing anti-adblock scripts and detectors that enforce ad placements and detect tampering. In some cases this results in punishing the user by refusing to deliver content until the ads load, blocking or kicking the user off the page, throttling connections or access, or in Twitch’s egregious case, more invasive ad interruptions. This has become a never ending arms race with ad blockers to keep up with minefield of invasive scrips monitoring what you do with their website.

TLDR: Ad blockers like UBlock Origin are already filtering how you’re asking for bur advertisers are attacking the plugins themselves and have their own arms race of scripts to punish those who interfere.

I wouldn’t say that it’d be strictly impossible, however if it can be done then it would come at a considerable cost to useability, versatility, etc.

One adjacent concept that comes to mind is the use of the :visited CSS tag to extract a user’s browsing habits. I remember seeing a demonstration of this where an “are you human” captcha was shown but the choice of image in each box was controlled by the :visited tag. I can’t find that post, but this medium article demonstrates a similer concept. There are mitigations to this luckily, but a fullproof solution would be to remove the tag’s functionality altogether, which would make certain websites (like the one we’re on right now!) much more inconvenient to use.

It seems trivial to me for a website to detect user behaviors that indicate the use of an adblocker. For example, if a request for a page is immediately followed by a request for a video on that page, rather than after 5-60 seconds, then they’re likey using an adblocker. If there is an ad placed between two paragaphs in an article, but two distant paragraphs are visible at the same time, it is more likely (although not guaranteed) that they are using an adblocker. If a user triggers an abnormal amount of those heuristics then they get flagged as an adblocking user.

There is no “perfect”(impossible to detect in this case) in anything. Perfection is a target that keeps changing.

What can you do when for example YouTube decides to bake in ads directly into the video source(they are actually testing this, although its feasibility is questionable)? You can’t block it, for now. In the future, you might be able to detect that with let’s say CV and AI and might be able to block it then. But then YouTube will do something about it.

If the site tries really hard, they can control serverside how many seconds of ad you watched to decide if you can access any content whatsoever. Something like this is already present on Twitch iirc. So in the endgame the only universal detection-proof solution I can imagine is AI/GPU based adblocker that will visually detect ads on your screen and overwrite them with something else without actually skipping.