Hacker News Reader

Comment by robbedpeter | Hacker News Reader

I've recently begun thinking that Google can't, in fact, improve their search results. They've forced the algorithmic landscape of available data onto a local optima because of seo and financial incentives. The only way they could improve results would be the cooperation of all players - website owners would have to agree to not try to game the algorithm, which will never happen. Everything is so oriented around gaming the Google algorithms that the game is deadlocked.

There's a clear opportunity right now for an awesome search engine to usurp Google's throne. Bing and Microsoft are too conservative to be a serious threat, but would happily acquire an engine better then Google. Nobody wants to be David on purpose when Goliath is a terrifyingly huge and scary as Google, but I think it's gonna happen soon.

View on HN

I don’t think it’s possible. SEO is targeted at google because that’s what the users use. If another engine became top, seo spammers would redesign around that.

It’s almost similar to how GANs work. Google tries to filter crap from content and the SEO spammers try to generate crap that is indistinguishable from content. And we settled at a point where seo spammers have won and their content looks like real valuable content but without real substance that ML can’t identify yet.

What was the tipping point where spam won the SEO arms race? Feels like SEO has been a thing for decades, cat and mouse, back and forth and the outcome for the search user has been good/decent results for most of that time, there have been big corrections against spam like Panda but generally consistent good results.. then recently bam! huge downswing in quality, spam won the war? Or Google gave up? Seems kind of unbelievable that suddenly its all gone to shit, what changed?

Let me very slightly push back: Google can't improve their search results without changing how they approach the problem. Now, it's Google, so the odds of them changing their approach is basically zero. But, it's my belief that for instance, a review team of 50 employees whose only job is to look at common searches and the results and hunt down ... let's call it low quality links... and manually blacklist them and pass aggregated results along to the machine learning teams that run the normal algorithm, could absolutely improve their quality. But, it's Google, and manual intervention is impossible (in their world view), so yes, in reality it will never be fixed.

They already do this, they have 10,000 quality raters doing this manual work.

https://www.searchenginejournal.com/google-eat/quality-rater...

https://support.google.com/websearch/answer/9281931?hl=en

Oh, I didn't know that. Okay, then it might be unsolvable.

In my opinion, it's not going to be as simple as someone outcompeting them at the existing game. It's that someone will beat them at the next iteration of the game, at a new paradigm of information discovery and distribution.

The analogue here is Microsoft losing its software dominance not due to a desktop competitor overthrowing them, but through failing to play the Web and mobile games as effectively as the new dominant players.

Reply on news.ycombinator.com