Tuesday, November 10, 2009

What is the future of search?

What is the future of search?

Search appears to be progressing into a split personality, depending on the type of data search being performed. At the moment we are using one tool (of visitor’s choice) to perform all search for data. That will continue to become more cumbersome as the data set available on the Internet grows.

1. Educational data – A significant amount of search is done in “stream of thought” for educational purpose. I apply education very broadly here as it could be a student researching Taiwan or a consumer completing research for a future purchase. Either way, this is neither life nor death, or is there an imminent purchase/decision to be made. The visitor is learning for future reference.

In this case the volume of data available on the Internet is an overwhelming burden. The searcher becomes bogged down in similarly-presented data with no guidance towards authority. The searcher is literally charged with finding data as well as discerning truth/authority.

I believe enhanced (matrix) search tools will begin to help this searcher find their way. Search tools will evolve additional relevancy based upon not only standards-driven user input but also statistical analysis of more finite pieces of data. We will see search results to data subsets versus the whole. No longer will the search be based upon an entire work so much as a piece of that work validated by user input with further statistical analysis of the search tool itself.

2. Consumable data – In this case I’m speaking of a visitor searching for data that will help them complete a timely (imminent) task such as location of a person, place, or thing. My expectation is that we will see a focus into highly-localized search (based upon known location) combined with some level of augmented reality tools.

The technology to ascertain the visitor’s location is already a simple matter. Resolving that to the data pertinent to that known location is only a matter of time. At the moment this data/location relationship is reliant on business and user data to be manually input and verified. It is only a matter of time before the processing and storage resources are applied to an intensive attack on this problem.

Once data is solidly tied to location, then some use of augmented reality will begin to pay off. A visitor will continue to search for a “keyword/thing” in a location and find a point of purchase for that item in a nearby store but computers will take it one step further.

I see visitors pointing their phone’s camera lens at an item and clicking a button. The image will be captured, combined with geolocation and proximity to data. Then the search engine will reply to the visitor with the question for refinement - “Do you want to know about that thing? Do you wish to purchase?”

Message Order Importance

Typically when building pages we see clients put together a list of links or information and put their most important information at the top, and the least important information at the bottom. This is a common practice for many sites, but Primacy and Recency Effects may cause a need to re-think this strategy.

A Primacy Effect is the chance that an item shown earlier in a message has a higher chance of being remembered by a receiver.

A Recency Effect is the change that an items at the end of a message has a higher chance of being remembered by a receiver.

Both Primacy and Recency Effects come into play, depending on a message receivers involvement with the message. Highly involved receivers (those who are actively reading your content) tend toward putting more weight on items earlier in the message. Low involvement receivers (those who are not processing the information in an active manner) will put more weight on items at the end of a message.

A quick overview of "Primacy and Recency Effects on Clicking Behavior" from the Journal of Computer-Mediated Communication:
Testing random order lists, where every item is the list is shown in each position over time, the most clicked item was the first in the list, while the second most clicked item was the last. This suggests that while some viewers are centrally processing the message, which leads to the primacy effect, there are slightly fewer people viewing peripherally, which leads to the Recency Effect.
This would suggest that when creating content, and choosing the order of the message/links/list that your most important information should be listed first, but that your least important should not be last. The last part of the message should be something that holds importance to receivers and can influence in a positive manner.

An example of this would be to perhaps not only put your main navigation across the top of your web site, but also to repeat that navigation at the bottom of the page to influence peripheral receivers to navigate the site.


Thursday, October 22, 2009

Google and Bing Fight for Content

It’s becoming clearer by the day that Google sees Bing’s presence as a threat to their service.

Because of Microsoft’s investment in improving Bing’s search algorithm and speed, the technology gap that once separated Google from its largest competitor has been narrowed significantly. Whereas years ago improvements to scalability, client-side usability, and relevancy of contextual search results helped push Google to the front of the pack, this noticeable technological disparity has slowly evaporated over time. Given Bing’s relative success when it comes to shopping and travel searches, as well as its recent deal with Twitter and Facebook to improve search results for these services (Google struck a similar deal just two hours later), it looks as if both companies are attempting to aggressively expand their feature-sets and the scope of content being queried.

The most recent development is Google’s partnership with streaming audio service Lala and the MySpace-owned iLike. The deal promises to bring streaming music to the Google page, allowing users to listen to a song once-thru for free or to pay for repeated listens (10 cents for an online-only version, and $1 for a downloadable mp3). The technical details are still a bit sketchy, but if Google manages to integrate this into their search in a substantial way (audio search, store search, improved MySpace search and caching, etc.), then they’ll have managed to set the bar higher for content search and potentially make up for any shortcomings in other areas like their previous attempts at integrating their search with online merchants.

Google clearly recognizes the level of competition that Microsoft brings to the table with Bing, and as time goes on it will be interesting to see where these two tech industry giants focus their efforts.

Tuesday, October 13, 2009

Is Google Indexing Flash Frames?

We have been running a test of Flash content inside Google for quite a while. Google started indexing text inside a Flash movie back in 2006. We never proved they were actually following action script links so we decided to put up the Flash test again and see if we pull different results.

We intentionally took the file down for 5 months to clear the cache out of Google. Now we posted it again with only a change to the date in the first file.

But are they indexing animated Flash Frames?

We have reason to believe that Google is actually indexing each frame in Flash. Our Flash developers are testing a Flash animation with 12,000 frames and a keyword at the end. It appears that Google has been banging away at the file with a vengeance but we don't see the keyword in search results pages to date.

We won't point to the test file in the blog until we can prove our hypotheses with certainty.

We should see results in the next few weeks if our last Flash in Google test has any hint to the outcome.