The New Google Search | Hacker News Reader

The New Google Search

www2.sandbox.google.com

52 pointsbrennannovak45 comments

OK, I'll be "that guy". What exactly has changed? Perhaps I'm merely trying the wrong types of queries or have a spotty memory or am just plain unobservant, but I don't notice any difference. I don't doubt that Google has minor improvements all the time and that that therefore there is, in a matter of speaking, a "new" google search every week or so, but I'm not noticing anything newsworthy. Is this some sort of ironic joke that I've ruined by taking literally?

I see a lot more grouped results by domain and at least one very heterogeneous result from the domain. Example: http://www2.sandbox.google.com/#hl=en&q=dawdle

However, this does makes it clear that Google sees http://www.dawdle.com and http://www.dawdle.com/index.php as different pages, so apparently I have some SEO tweaking to do. I'd encourage all startups that rely on SEO to test their best keywords here to get a different feel for how Google "sees" your site.

Second result: 209.161.37.11/dictionary/dawdle That looks like a bug to me - it's http://www.merriam-webster.com, which, in addition to not being a sketchy IP address, is also a nice brand in terms of dictionaries.

Seconded. Some commentary would be appreciated.

i'm seeing more total search results.

No ads.

Maybe because it's a sandbox project, and not production yet?

"Just as Facebook announces internal search for public notes, Google counters with an effort to improve on its existing services. In today's blog post, the company unveiled its new Caffeine search infrastructure to web developers."

http://www.readwriteweb.com/archives/caffeine_google_tests_n...

The auto-suggest seems to be coming up with longer phrases. In the past, I've only seen 3 or 4 word phrases, but the new one appears to recognize longer sentences, too. Of course, this might just be me, trying to find something new when there clearly isn't. In that case, feel free to ignore me :p

Try typing "what's so great about" in the box. Mildly amusing :)

Doesn't work for me. Besides "old" and "new" google show the same things...

The main difference I can spot is that the regular site returns a web page with a search box with a 200 result code, while this version just returns a 404. So this may in fact may have been a joke that has benn pulled in the interim.

I got a 404 because I was visiting from Mobile Safari, which redirected me to the /m mobile site which apparently doesn't exist in the sandbox beta.

Past searches are on top of the auto-complete recommendations.

To me it seems that they improved a bit collapsing same domains. However they still have a lot of work to do - try searching for basically anything and you will still see domains which were on the first page repeated on 3rd or 4th. Annoying.

Another difference is the size of the indexed web:

Sandbox version: Results 1 - 10 of about 25 120 000 000 for www. (0.22 seconds)

Regular version: Results 1 - 10 of about 22 830 000 000 for www. (0.25 seconds)

Overall that's not much to write home about

The search results load via Ajax. That's the main difference that I see. Also notice that the URL changes when you type in your search result, but that the query args are sitting behind a # anchor. Normally, those values don't get sent via referrers, which would mean that you wouldn't get that information in the referer header on the server that hosts the search result when the user clicks it. However, in this case, Google is doing HTTP redirects in order to send that information along. At the very least, we can see that the format for their URLs has changed. But they may just be in a data-gathering mode right now. If they turn off that redirect, you won't be able to deduce the search query that the user typed in based on your webserver logs.

Maybe they are looking to monetize something else with this‽

I'm not seeing that at all. Normal page refreshes, no anchor in the URL.

odd, the regular google.com has done this for me for quite some time. i thought it affected everyone until now. i just loaded a different browser with no cookies on it and it does searches the old non-ajax way. even though i'm logged out of any google services, my long-lasting google cookie must have this ajax option enabled on the main google.com site.

Interesting, you guys must be on one of those A/B tests that Google does without telling anyone. I certainly don't get AJAX results in the classic or sandbox searches.

Its a completely new backend they are using so ordering of results is different.

From searchengineland: "Based on the blog post, we can guess that this new infrastructure may include ways of crawling the web more comprehensively, determining reputation and authority (possibly beyond the link graph and what’s typically thought of as PageRank), and returning more relevant results more quickly, although Google’s Matt Cutts told me that the changes are primarily in how we index.Google’s new search is only infrastructure related and includes no UI changes."

http://searchengineland.com/caffeine-googles-new-search-inde...

> Its a completely new backend they are using so ordering of results is different.

Indeed. A search for "velociraptor defense" in the old version finds 72,100 results in 0.39 seconds, while the new backend yields 565,000 results in 0.30 seconds. That's nearly eight times as much practical defensive intelligence delivered 23% more quickly!

(ok, ok... I'll shut up and stop eroding the average comment quality now.)

Sandbox version: Results 1 - 10 of about 751,000,000 for sex [definition]. (0.11 seconds)

Regular version: Results 1 - 10 of about 104,000,000 for sex [definition]. (0.13 seconds)

7 times as many results for the same exact query

edit: speed wasn't an improvement, but definitely the number of results are.

there are definitely more results. the speed, i think, is not an improvement.

yah most notably the amount of results is larger but also the actual results are different with every query I have tried. At least half of the first page is always different.

For when 104,000,000 results just aren't enough...

Perhaps the new architecture means that more pages are indexed more frequently. Could it be an architecture that supports more "realtime search" (after twitter search). I notice the news is pretty up to date.

Why do I get the feeling someone's trying to 'emperor's new clothes' me?

Still no way to permanently remove a domain from searches? I have a list of about 15 sites I absolutely never want to see in my search results. It seems like a simple feature to have.

Try SearchWiki (click the X next to each result you want to kill).

Edit: okay, yeah, that won't work. The OP is looking to kill all of a domain, and SearchWiki is only one query/result combination.

That just removes it for that one search, it doesn't remove that domain from all future searches.

Doesn't that just blacklist a specific result? I want to kill an entire domain, forever.

It seems like popular websites like Wikipedia and Flickr are ranked slightly higher in the new results compared to the old. Perhaps it's a more crowd-sourced version of pagerank that they have been training/testing. Since they have been HTTP re-directing results for the last 5 months, they could be using this clickflow data to adjust sites ranking according to users behavior on the search result page itself.

One notable observation from morning watching top searches of a client: A lot of websites which "were" high on SERP based on links are now gone.

This also connects well because of the recent change google announced in the way link juice is distributed (to avoid misuse of no-follow attribute and paid links) http://searchenginewatch.com/3633972

Spooky.

I have sites at #3 for "Travel Blog" and #2 for "Software Consulting Services" when you search from Google.com. Neither are on the first page in the new version.

I know that Google mixes thing up every few months and that my results dance around a bit. But it's a little unnerving to see something so important to so many people's businesses change so dramatically.

Did my comment get eaten , or don't I have the Karma to post? Re-post: Try searching for any term, then click on the 'Try Google Experimental' hyperlink at the bottom of page to see the ??Broken-rewrite-rule-related?? 404

I think google is still trying to squeeze the last 0.1% of gain in terms of web search. Future value will come in terms of social, recommendations, and user interfaces. not squeezing that last 0.1%

I see no real difference, I made some queries and they are (about) the same as my default goolge.

may be if you give a difference list.

Type in "Hacker New" and you'll note that the first result shows the site with it's URL, not just the name.

Is that it?

The same thing happens with old google.

It does seem to yield quite different results... but I can't quite notice patterns as to why or how they are different.

Maybe it's indexing the semantic web now :P

Looks like maybe it has more realtime results? Speculation, but still.

Searching for my first name brought up my twitter page on the first page, where it never did before.

I noticed significant improvements searching for "qt" and for "boost".

Reply on news.ycombinator.com