How can they tell it’s a discussion forum? Does the scraped search spam that copies stack overflow content look enough like a discussion it fools their heuristi

How can they tell it’s a discussion forum? Does the scraped search spam that copies stack overflow content look enough like a discussion it fools their heuristics? Is it a manual process (in which case you can bet it “doesn’t scale” and won’t be built) this is the problem they face. Literally nothing is simple given the size of their dataset, the scope of their user base, and the adversarial nature of the very world around them generating new data they must work with in order to do the job. There’s definitely an element of “we got our profit so fuck it” with respect to the search engine advertising business and Google’s incentives to make search quality better, but that doesn’t change the difficulty of the underlying problem. If I wanted to pay Google $5 per search for super high quality results, even $50, they can’t just make this product better to get my money, they are fighting an ongoing war against adversarial SEO which prevents this from ever being better than stalemate at best or more likely due to economics, the slow slide into declining quality we see due to the SEO side having more money with which to pay for engineering brainpower.