Hacker News Reader

Comment by hyperbovine | Hacker News Reader

The future of adblocking is an AI-based system that detects already loaded ads visually and simply masks them from you on the client side. Such an approach (sans AI) has been used in the past, but it fell out of favor due to bandwidth and CPU usage concerns. Since then, bandwidth and computing power have grown exponentially, such that the disincentive of the load-and-hide approach will continue to diminish over time.

View on HN

That approach (sans AI) hasn't fallen out of favor. Ad blockers still do it, they just also support blocking them from loading. There's no benefit to loading an ad if you're going to hide it, it's always wasteful. Growing computing power and bandwidth doesn't change that.

Of course there is a benefit—there is no way to block users who are running such an adblocker.

Yes! An AI that looks at the ads for us and just filters them out. Now it doesn't matter what they do to circumvent the blocker since the AI works by recognizing the brands and ads themselves instead of relying on page structure.

Imagine having smart glasses which seamlessly filter out product placement in movies, sporting events, even the ads in the real world. It's a dream come true. Our reality can finally be clean again.

By “orders of magnitude”

not exponentially

Say it with me so we can normalize this usage

There are plenty of people who misuse the word "exponential", but in this case, assuming much of the period in question was during Moore's Law growth, "doubling every N months for many years on end" is genuinely exponential.

At higher numbers it is very hard for any whole number exponent >1 to be reached

In any case what year was AI adblock attempted so we can check if this is one of the narrow use cases where exponential performance improvements occurred within computing

“Orders of magnitude” will be more frequently correct and also include “exponential”, whereas “exponential” will be less frequently correct

> not exponentially

We don't know that the growth was n^x and not x^n.

Wouldn't that be exponential vs polynomial?

Personally, I err on the side of only correcting someone else’s speech when I know I am right.

Reply on news.ycombinator.com