Live data from Hacker News

Google algorithm change launched

news.ycombinator.com

181–190 of 203 posts

Re: Google algorithm change launched

#181
post #31

This is awesome! Now if only I could completely remove certain sites from my search results ala the Google Wiki stuff. I'd love to drop swik, and expertsexchange and a few other annoying sites. Its possible this algorithm change will make these sites less annoying to me though.

Bonus trick. Experts Exchange actually often has some good stuff on it. The pages are designed to make you think you have to buy a subscription to see the answers, but if you simply scroll down beyond the ads for their service the answers (which aren't scraped from elsewhere) are visible near the bottom of the page. Yes annoying, I wish they didn't work this way, but there are times that I've found experts exchange t…

> which aren't scraped from elsewhere

They used to be scraped from usenet. They may well still be, but I don't have any current evidence. Are they still doing that?

Re: Google algorithm change launched

#182

Earlier quoted context omitted.

I understand what you're saying, but doesn't Google News do the exact same thing? Google News provides snippets of content, and helps people discover the news providing direct links to the original source of news. For Google to deindex a site like celebrifi and while running a competing product (Google News) smells a bit of monopolistic behavior. It's suspect it's unintentional, but Google is going to have to walk a…

When News results show up on the search result page, they link directly to the story, they don't link to the Google News landing page or category where you'd then have to click on a link. Google News itself falls into the second category - an aggregator that stands on its own, as a destination. Much like Hacker News. I personally don't use it, but the people that do go because they find it lets them discover a bunch…

I know that you're saying that Google News is a dedicated site, and separate from the search results. But, Google links to it's own aggregation service from its results on a regular basis, and at the very top of most every results page in Google there is a link to Google News version of the search.

Google News does stand alone as an aggregator, but you have to admit that it is promoted heavily by Google search. If GOOG keeps doing stuff like that, I suspect there are going to be a lot more companies that start to take umbrage, and start challenging this behavior in court claiming that it's anti-competitive behavior.

Re: Google algorithm change launched

#183
post #60

While I applaud such an improvement, I still despair over Google's inability to handle context. For instance the search phrase 'tex decorative rules' is totally miss managed. First it overrides the search and changes tex to text. Humorously enough, when you counter override back to tex, the search is even worse. We won't even speak of what happens when you change tex to latex--- porno doth ever rise to the top I gues…

tex is a hard one. And [latex decorative rules] is even trickier--most people wouldn't expect technical documentation if they just saw that as a random query. That's more of an issue to pass on to the synonyms team and not in the scope of this change, but I'll pass this one on to the right folks

Would it be feasible to introduce a search customization setting: "lower the estimated probability that I mistyped"? I assume google uses something like two cut-offs: A A, it suggests but doesn't search for the correction, while if P(mistype)> B, it displays auto-corrected results. Simply an user option to nudge P(mistype) lower would probably suffice, or an option to increase B.

A major usability problem for me with auto-corrected results is the two lines with links: "Showing results for " and "Search instead for ". Interpreting those two lines takes me out of the flow of analyzing search results. I haven't yet been conditioned to click on the second link. I never want to click on the first link, because google is already showing me those results.

I frequently search for unusual acronyms, terms, variables, made-up words, and gibberish that shows up in logs, among other things that google likes to "correct" for my benefit. I realize those use cases are not typical, but current auto-correction behavior can be extremely aggravating in those cases.

Auto-suggest a correction all you want, but auto-correcting and suggesting the original is going too far I think, unless there are zero hits, or some equivalently strong algorithmic determination is made that I couldn't possibly want the query the way I typed it.

Re: Google algorithm change launched

#184
One potential solution to spam and personalization of results lies in doing the job on your machine, much like you can get rid of spam in your local mailbox.

I'd recommend 'Seeks', (http://www.seeks-project.info/). It requires you run it on your machine, or use a public node. Though while on your machine, it 'learns' from your navigation and re-ranks the results based on these local data. Additionally I use regexps to remove websites I don't want to hear about, like expertexchange.com.

Re: Google algorithm change launched

#185

Matt, I just went through my search history because i remembered a very specific instance of seeing this. Here's the query. http://www.google.com/search?q=nstoolbar+bottom+bar&ie=u... You'll notice that efreedom.com shows on the first page with content taken directly from stackoverflow. While stackoverflow does show in the results, the exact page that efreedom copies does not. Anyway, I'm glad you guys are taking thi…

It's strange how these results vary from person to person. When I click that link, I see an eFreedom link (with SO content) at #6 and a different SO link at #7.

Re: Google algorithm change launched

#186
If a site ranks high in a SR, isn't it because it must have good backlinks?

Shouldn't that be more important than whether it has original content or not?

Maybe an aggregator displays the content in a more useful way than the originator so it gets linked to more than the originator.

If the content creator doesn't like his content copied he can take it up with the copier. It's not Googles job to get involved in that.

Google's job is to give the searcher a list of the sites that matches his keywords in an unbiased way. They should do that mostly on what the internet thinks is the best site, not what Google thinks is the best site.

Re: Google algorithm change launched

#187

Earlier quoted context omitted.

It looks like we've got SO above efreedom for that query, but it's always nice to find a url that we didn't have that we'd like to be indexed. That lets us check whether we can improve our crawling/indexing. Thanks for the example!

If you tell me that sites who chronically scrape content elsewhere get a "pull-down weight" attached to the whole of their content, you've made my day.

That wouldn't be nice. There is a big difference between copying content from Stack Overflow, to running a Planet news aggregator on a community site for your open source project. The first is an ad scam, the second is valuable service to the members of the community.

Re: Google algorithm change launched

#189

Earlier quoted context omitted.

Second that. WTB ability to customize my search results by blacklisting certain sites. Big change I know, but would be a hugely useful service Google could provide.

why not just use the - function? Create something like: http://www.google.com/search?q=google+-google.com&ie=utf... I suspect you could hack together a quick greasemonkey script to automatically inject your negative site list when the URL is generated

Because Chrome still doesn't sync your "search engine" settings. So maintaining a massive customized search engine across the multiple computers I use would be a pain. If this was a native feature it would work so long as I logged in to Google. Of course, if they fix chrome I'll be happy too.

Re: Google algorithm change launched

#190

Earlier quoted context omitted.

Your account page doesn't say anything. I assume you work for Google?

Yep, in fact I wrote the change we're talking about in this thread. :)

Hah. Good to know. I guess I'm your 'enemy' then since if there was a label for me it'd probably be blackhatter. Though my 'spam' tends to be of much greater quality than what most of the stuff BHW sort of people produce (I have some original content written and excluding my bottom of the barrel sites, the others have their automated parts like scraping edited/checked by a hired person).
Post reply on HN