[pembrook]

Except, in the real world there's more than just A) unknown snake oil and B) high quality products with perfect awareness.

Your model only includes products on the narrows of each extreme. In the middle is the wide spectrum of most products...the ones that don't meet either description.

Even if you make the exact same product as a competitor, there is no prisoner's dilemma if you're targeting a different niche market to sell that product to. Perfect competition does not exist in the real world. The only thing that comes close might be a commodity like oil or water. But even water can be targeted to different segments of the market.

[AnthonyMouse]

> Except, in the real world there's more than just A) unknown snake oil and B) high quality products with perfect awareness.

There doesn't have to not exist more than A and B. If A and B both exist then 1 and 2 are each true and the further existence of C and 3 don't change that.

Moreover, even if some additional classes exist, the two examples are still central and problematic, not least because they're more likely to represent a higher percentage of ad spending.

The first because advertising is the only way to sell crummy products, since the only way to get anyone to recommend it is to pay them to, so their incentives to use it are higher.

And the second because the existence of the prisoner's dilemma is what drives up the ad spend on both sides. If you're targeting a niche that no one else is then you buy a small amount of advertising, reach those customers, make your sales and are done. If you're locked in a prisoner's dilemma with a direct competitor, you spend a little so they spend a little so you spend a little more until you're all spending a huge amount. And the fact that you're selling Fords and they're selling Chevys and they're not completely identical products doesn't really matter when they're both still cars.

[ethbro]

Relevant: https://en.m.wikipedia.org/wiki/The_Market_for_Lemons

The presence of pure mercenary advertising increases the global noise floor, which increases the information asymmetry by decreasing the visibility of quality signals.

The irony is that Google sees itself as a company that increases access to information.

You know what that would look like in the ad space? Product testing, reviews, and endorsements. Something like "Verified by Google" (aka Wirecutter).

Instead, Google absolves itself of responsibility via algorithms, steered by marketing folks in charge of their primary profit center.

And we're surprised by the corporate decisions they make?

[6510]

What nice way of putting it. Google has indeed suppressed quality signals to the point you don't have to bother creating one. Making peoples homepage invisible was hard, some had to be hoarded into the walled gardens. Wait a few years then shut them down with very little effort.

I was naively surprised one time when I see quality writers shout down rating systems. Apparently making a popular quality product doesn't make scrutiny desirable.

[TeMPOraL]

Rating systems are universally garbage. They're thoroughly gamed and create their own shadow economy of paid reviews, spam, and bribes to take down negative reviews.

[6510]

The baby was tossed with the bath water. There are countless possibilities to create a rating system (or a system that describes the qualities of an article) I have only seen a tiny number.

The shortest description of the thought is: The meta data is more important than the data.

Back when google indexed peoples websites the organic ranking wasn't bad at all. People wrote niche articles about original topics, if you searched for one you would find those blog postings. I was often amazed by how specific the content addressed what I was looking for.

How well it works depends on the type of rating. It should probably start with things so obvious they are hard to game. Even self rating could work, something like: professionally affiliated with the topic 0-5 in the range 2-5 you get to provide an url.

I liked parts of PICS3.

I could see a system where we run our own rating service and rate things with a mix of original and unoriginal qualities. You use the bookmark list it generates or enjoy the persons work then subscribe to their ratings and add weight to it. We make collections of such subscriptions and use them the same way. When visiting a page the url (or other identifier) is passed around and a rating is returned. Similarly, people you've subscribed to crawl around the web and we arrive at a set of pages you should probably visit. If there is crap in the list a single click reduces weight on everything that endorses it.