Deprecated: Function mysql_list_tables() is deprecated in /nfs/c02/h07/mnt/28080/domains/hypergene.net/html/blog/pmSource/db/db.mysql.php on line 328

Deprecated: Function eregi() is deprecated in /nfs/c02/h07/mnt/28080/domains/hypergene.net/html/blog/pmSource/inc.lib.php on line 246

Deprecated: Function eregi() is deprecated in /nfs/c02/h07/mnt/28080/domains/hypergene.net/html/blog/pmSource/inc.lib.php on line 246

Deprecated: Function eregi() is deprecated in /nfs/c02/h07/mnt/28080/domains/hypergene.net/html/blog/pmSource/inc.lib.php on line 246
Hypergene MediaBlog » Site Review: Google News Beta

Deprecated: Function eregi() is deprecated in /nfs/c02/h07/mnt/28080/domains/hypergene.net/html/blog/pmSource/lib/weblog.fns.php on line 597

Deprecated: Function eregi() is deprecated in /nfs/c02/h07/mnt/28080/domains/hypergene.net/html/blog/pmSource/lib/weblog.fns.php on line 1027

Wednesday, 09 Oct 2002

Site Review: Google News Beta

Google NewsNever has a news product so clearly exposed the old-school thinking from the new. The prevailing banter over Google News: On one side, it’s a realized global newsstand - a stunning example of programming and understanding of the web. The other side, it’s just another portal/link aggregator, with less than average value to news consumers.

Regardless of what side you’re on, it’s indisputable that Google and Yahoo! have done more to innovate the display and interaction of online news than just about any news site could have dreamed. Sure, Google News has some hiccups, but it’s another valiant effort in the Google business ethos - “Let’s do something well, that people need, and later we’ll figure out how to make money at it.”

What is it: Google News Beta is an index of top-story links, culled from 4,000 English-language news sources around the world. The search collection is updated every 15 minutes.

Right: Online news data is almost perilously unstructured. At least, that’s what we thought until Google News came along. The triumph of Google News is that they are somehow able to accurately collect a large number of stories related to a single event - the “Bush: Saddam Tyrant” speech or a Christie’s auction of Gary Player’s 1961 US Masters green jacket - from 4,000 sources. It’s a miracle considering there’s relatively little or no structure to any of the news data on the web. It’s this collection of related stories around an event that is most valuable. In online news user studies that we’ve participated in, readers told us that what they loved most about the web was that they became the editor - vetting sources and opinions from as many media outlets as they desired. Google News provides a seemless way to do just that - be your own editor. Now, 1,390 stories about the “Bush: Saddam Tyrant” speech might be a little more than you bargained for, but Google’s algorithm attempts to give you a starting place.

In our Amazoning the News whitepaper, we defined five tools for enabling effective web experiences. Google News sizes up on all of them:

1. Live in the network. Google has made it this far because they just don’t understand the network, they live it. Some even believe that they are the network. Regardless, they exploit to the fullest.

2. Exploit time and place. These two items are fluid and subjective to web users. For Google News, the place is the world, how much more fluid can you get? And time is a tool, not just a stamp, for enacting upon news and information. Solid.

3. Be interactive. Sounds stupid, but most news sites aren’t terribly interactive. Google is only an interactive tool, so they clearly understand this. As a search tool, Google is unparalleled. Google News is assisted search. Not a blank slate, but a starting point in the mass of news.

4. Data mine and create relevant data relationships. Mine your data and create relevant relationships between them. See our previous note about a collection of related stories around an event being valuable.

5. Have a personality. For many news sites, the only thing that’s going to set you apart is personality. Is your site a Yahoo! librarian or a Drudge Report? On the surface, Google News derives its personality from one of two things: the content it indexes, or the algorithm that drives the presentation. Thus, the editors who claim Google News’ lack of appeal, because the algorithm doesn’t see the things that a seasoned editor does.

However, personality can be derived from the collective quality of the interactions on a site. For example, after using Google search successfully for a few days, you begin to associate it with terms like “reliable, consistent, fast, relevant, helpful, credible” and so on. Their search interactions work so well, a valued brand personality has developed. After they tweak the algorithm and the presentation for a few more iterations, we might say the same about Google News. It’s interesting to note that the personality keywords of the Google brand are the same that media outlets covet in the news business.

Wrong: Most of these comments concern relevance, which is the heart of the Google technology. Perhaps we know enough about Page Rank and how it works in the core search tool, but in the news business we don’t know enough.
1. Unintended collisions: News editors have had a field day with this one. Here’s the best example we could find today. This is a story about Nielsen ratings, but Morley Safer looks hysterical next the “Bush Iraq Speech Draws Big Ratings” headline. That kind of design would never hold up a real media organization. Which proves the point of some editors - “HA! You do need us.” Of course we do. Google never said they don’t.

2. First source confusion: The logic behind Google News’ special sauce algorithm is completely hidden. This makes sense-making difficult, if not impossible. For example, take a look at this Detroit Red Wings news chunk. The NHL season is cranking up, and the preseason predictions are out. But who in the heck is Sportal.com from the UK? That’s the “top” source for this story? The second and third aren’t much better - The Billings Gazette and the Erie Otters. Never heard of either. So, either we can blindly trust the Google algorithm and click on the first one, or we can guess that maybe Billings is in Canada, where they really know their hockey. Users shouldn’t have to think this much.

We’re not suggesting that the Google News algorithm be entirely exposed, but we need something more to go on than “Page rank is one of ‘a hundred’ other factors that go into determining how a specific story will get ranked in Google News,” (Outing). If Google News cannot fix this, a magnitude of understanding (relevance), will drop. Google should create an explanation similar to Metacritic - it doesn’t reveal all of it’s algorithm, but gives a clear idea of how it works. As well, Google might consider calling the algorithm “News Rank” and devoting a small marketing effort toward explaining the way it works. They did this with Page Rank and it paid off.

3. Dupes galore: How many versions of the same wire (AP, Reuters, etc) stories do we really need to read? This is either a testament to omnipresent power of the AP and it’s ilk, or the predictability of news judgment at most media outlets. You be the judge.

We recommend to media outlets: Think outside the al-gor-ithm. Put aside your insecurities about the phrase “no human editors”. You sound defensive and whiny. This is not a Bill Gates attack on the classifieds business. Google is assisting each user with a task - finding news that’s relevant to their needs. With that in mind, media outlets should:
Get a search engine that frigging works! Forget the Google News graphic presentation for a minute, and focus only 100 pixels deep - Google News is a killer search tool for crawling through a mountain of news. If news sites had search tools that worked 1/10th as good as Google, finding what you want on any news site would be a breeze. Perhaps the first place to start is to pony up for the Google Search Appliance.

Everyone’s an editor. Remember that J-School grads are not the only people on the planet qualified to filter/edit today’s overwhelming information flow. In fact, blogs are excellent at filtering. News media get in trouble because they try to be everything to everyone. It’s all about reach. Fact is, the top headlines that Google is presenting is just window dressing on a great search engine. Since anyone can arrange their own top headlines, what’s the threat? What’s more important is that the information in and behind the headlines is accurate.

Embrace the network. Don’t bury yourself deeper in the cul-de-sac of registration. Media sites already feel disconnected from the social network of the web. Registration walls make news sites feel like a vault.

A reporter’s resource. As Outing recommends, “Google News is an incredible resource for reporters wanting to know what other journalists have written about a topic.”

Go easy on the ads. Google’s minimal navigation and lack of obtrusive - and bandwidth eating - banner ads is striking. A news site with nothing but news content - heresy! While we’re sure Google will put advertising on their site at some time, we’re equally sure that they’ll do it right.

We recommend to Google
Enable Personalization. Creating a personal sub-index of say, our top 100 news sources seems like a natural. The concept of slicing the master news index, which Google already does in B2B and International Search, is almost mind-boggling. A clipping service meets news aggregator meets search engine - except the user defines it all - including sources.

Narrowcast it. Imagine a Google News page about hockey (worldwide), or even our favorite hockey team, The Detroit Red Wings. Now there’s a start page! Seriously, this type of search service provided to likes of the NHL seems like a win. We smell a B2B competitor to Moreover here. As well, Google could turn 4,000 news sources into 4 million, which could spur serious content competition. (Phil Wolff has some good thoughts on this.)

Kill wire dupes. We know most news data has relatively little or no structure, but if you can kill/hide, the AP story duplicates, that would streamline this product and perhaps enhance relevance.

Explain relevance. When we “sort by relevance” on a search result, what does that mean?

Go more global. According to an email from Google team: “We are considering expanding this service to include other languages in the near future.” This might make this global newsstand truly global.

Provide RSS Feeds or Open the Google API to News. Anything so that we can put a top 5 on our site.

Lingering thoughts: Does anyone know how they choose story blurbs? On some stories, it’s the first couple of sentences. On others, it’s the second graf. How do they do that?

Posted on Oct 09, 2002 | 11:52 am EST
Permalink | Comment (0) | Print | Del.icio.us | Technorati