Fraud detection is just hard. And my expectation is that for every legitimate friendly fire instance there are six or ten "marginal fraudsters" trying to spin their troubles with Google via blog posts like the one in the link.
I clicked on this thinking it would be a clear case of Google doing something "evil", and had to read through very carefully before I figured out that it was just another account freeze. Meh. If it's seriously $40k, then sue them and figure out what really happened in discovery. That would be a blog post I'd want to read.
Using $40k of a service (ad space in this instance) and then refusing to return the service providers calls and emails for over 12 months may not be "evil" per se but it's certainly on the malevolent side.
If it ruined someone’s livelihood then I'd say it's getting to be in the locus of what could well be described as "evil".
For a company to do this without apology demonstrates a pretty serious ethical deficit (assuming we have the full story before us).
Look, I don't know anything about "hatchlings", I'm just saying I'm not going to believe a post like this prima facie. They look like a garden variety spammy facebook game to me. That's a market pretty well known for playing SEO games and gaming ad systems.
Are these guys guilty or not? No idea. Fight it out, get some evidence, and then come to the public for support.
But it would make it a heck of a lot easier to defend ourselves if Google had actually accused us of doing something wrong. To this day we're not even sure what we're alleged to have done to get our account banned.
I do not think it would be ethical to test it directly. But you don't have to search far to find people who have had their accounts crashed by angry exes.
Scripting that will only display the adsense advertisement once per IP visiting your site... yet guess what, Google disallows this practice!
A better solution is not to use cost-per-click in the first place, instead relying on metrics that are harder to game, like unique visitors and page views. These are arguably a better reflection of an advertising space's value even without taking click fraud into account. That's what Project Wonderful did, and it has worked out well for some of its users, but its model is also more long-tail/long-tail, which I think hurts its profitability.
Surely page views are as easy to imitate as clicks (possibly easier). Unique visitors would be reasonably easy to game as well I'd imagine.
Say I'm a fraudster with a thousand IPs and a well-anonymized browser. If I'm paid per click, I can easily turn that into a thousand apparently legitimate clicks, which, depending on how I play my cards, could net me a few hundred dollars.
By contrast, if I'm paid per unique visitor, I can turn that into a thousand apparently legitimate unique visitors, which will very optimistically pull in maybe $5 (judging by Project Wonderful).
Now, neither of those schemes are optimal, but just about any improvement you suggest for the unique visitor model will reap ten times the rewards applied to the per-click model.
Perfect. So it's your responsibility to prevent invalid clicks, even though pretty much any way you can think of to prevent them is explicitly forbidden by the TOS, but if someone does click your ads a bunch of times, let us know so we can steal all the money you've earned in the last month and shut you down.
Basically, Google's policies mean that if you don't like a website which uses AdSense for revenue, you can screw over the owner by sitting at their site and repeatedly clicking their ads. Google will see the "fraud", assume it was the site owner doing it, and shut down their account with little to no opportunity for appeal.