Hacker News Reader

Comment by userbinator | Hacker News Reader

Don't forget the "showing results for Y, click here to search for X" when X is what you actually wanted to search for.

It's ironic that, in trying to make search "human friendly", Google has also succeeded in giving it all the negative traits of a human --- the "human" that is more like a commissioned salesperson.

I know the argument is that humans frequently make mistakes so "we should just show them what we think they wanted", but that's just opening the door for manipulation.

Search was much better when it was closer to a "grep the Internet". IMHO machines should remain "dumb" (for lack of a better adjective) and leave the important decisions to the users, keeping the latter in control.

View on HN

Grep the internet and sort by pagerank was great. Then everyone removed links that went offsite. Then link farms. The fake blogs and influencers are the evolution of link farms. Unfortunately sellers will game the system. How do we create a way for buyers to game it?

Maybe it’s time to bring back curated web directories like the original Yahoo?

Yeah, I think so.

Curation-as-a-service could even become a viable market. It's arguably already happening to a certain extent through various platforms, creators, and aggregators; but it's not really split off as its own service yet.

Pagerank was great when it was new because it had never been used before. Once Google got popular enough to be worth gaming, it became an example of Goodhart's Law.

https://en.wikipedia.org/wiki/Goodhart%27s_law

Pagerank could be gamed too. Buy expired domain with high pagerank, establish a forwarder to your domain and your domain inherited the pagerank.

I thought that lapsed domains got their pagerank reset (or significantly reduced). Is that not the case?

I remember the days of hiding text in the same color font as the background so that grepping would make your page the top result. So I’m not sure those days were better.

The dumber search is, the easier it is for the bad actors to game it and drown out the legit sources.

Maybe page rank (mentioned in a sibling response) helped address this, but then it’s really no longer just a dumb grep.

I remember the days of hiding text in the same color font as the background so that grepping would make your page the top result.

A simple "-site:somesite.com" removes those, or just scroll past them. Keyword spam is easy to ignore because the title of the page, the domain, or the path often has nothing to do with what you're searching for, and the preview text is nonsensical.

The dumber search is, the easier it is for the bad actors to game it and drown out the legit sources.

The dumber search is, the easier it is for users to ignore the crap and find what they're looking for, even if it's not the first result. The "smart" omission of results has the effect of removing a lot of useful ones too. Google's claims that they have X number of results for a search is essentially meaningless if you can't actually see them all.

> Search was much better when ...

If you really believe that, maybe it is a nice startup opportunity for you :)

Trying to make money is precisely the reason Google is the way it is. I don't think starting another business is the answer.

I've been thinking about a distributed search where each participant in the network crawls and answers queries related to a part of the web. Make search more like bittorrent. Is there such a project?

Cool concept. Let me know if you find anything!

Nah, it wouldn't be profitable, at least not in terms of money

I have long wondered how much such a project would cost. I have no actual idea how much hard drive space would I need, nor how much processing power.

Reply on news.ycombinator.com