Whats behind Apples forthcoming Spotlight search, and how will its new interface benefit users?
When he unveiled it at last weeks Worldwide Developers Conference in San Francisco alongside Mac OS X 10.4, aka “Tiger,” Apple CEO Steve Jobs touted Spotlight as a solution to the clutter of folders and files. It will take three separate technologies to perform that trick, Ken Bereskin, senior director of Mac OS X product marketing at Apple Computer Inc., told eWEEK.com. One is already built into the Mac OS X file system, and two are newly invented.
The new twists in the Spotlight interface drew praise from Dr. Jakob Nielsen, a principal and co-founder of the Nielsen Norman Group, a user-experience consulting group.
Read more about Nielsen in this CIO Insight Q&A.
Spotlight has already drawn comparisons between the search technologies in Tiger and Microsoft Corp.s Longhorn, now slated for 2006. Both Apple and Microsoft tout their search capabilities as a boon to users seeking relief for the exponential increase in files and e-mails on networked systems.
Even though Apples own banners at WWDC taunted Microsoft with slogans such as “Redmond, we have a problem,” Bereskin demurred any direct competition between Spotlight and the Windows File System (WinFS) local machine search engine promised for Longhorn.
“The way the [Tiger] file system searches for files, for metadata, for indexed data—theyre all different technologies,” Bereskin said. In contrast with “others,” he said, he thought Apples approach was better than a “one-size-fits-all” system.
“Spotlight is much bigger” than a file-search engine, Bereskin said. He said that what Cupertino, Calif.-based Apple presents as one feature really comprises three distinct search technologies.
“It starts with a regular file search,” Bereskin said, “but it also taps into any additional info its able to get from files.” This range, he said, means that Spotlight can read the author, subject and other keywords attached to a piece of information.
Although this technique is similar to the existing keyword feature in applications such as Microsoft Word, where users can manually set keywords to associate with files, Spotlight also automatically detects copyright, author, color space of images and other information for use in searching.
Spotlight includes importers that can open and read metadata from a variety of file formats, Bereskin said. Developers also can extend Spotlight by adding their own importers. Bereskin stressed that extensibility and access were key components of Apples design of Spotlight, and why a layer of search APIs overlie and can call all three underlying search technologies.
The third search technology, Bereskin said, is the ability to access content of files, from text documents to e-mails to readable text within PDFs.
“Doing it all together is the big thing” about Spotlight, Bereskin said. The end result of a search is a list of relevant items, which can be categorized by the user to group by file type, date and other criteria.
Bereskin said developers had been asking for a system-level relational database, which will be included in Tiger. However, this database doesnt manage file system data.
“We decided a specific technology for each type of search would be more efficient,” Bereskin said. He stressed that each of the three are very low-level. The file search, he said, was based on the B-tree file system inherent in the HFS+ file system, but he declined to offer details of the other two technologies, saying both are new and based on unique data structures.
Spotlights iTunes Heritage
In how it organizes data, “the origin of Spotlight dates back to iTunes,” Bereskin said; its browser interface more closely resembles Apples music organizer than the spatial Finder metaphor.
Apples new direction is welcome news to Nielsen, who writes extensively about interface and usability issues. “It is certainly a good point that the original Mac interface … works well for a small amount of info,” he said, “but now we have a thousand, ten thousand documents, or information objects, such as e-mails.” This, he said, presents two problems.
The first is that a large collection of individual objects is not an efficient way to manage items. For example, Nielsen said, it becomes problematic even to fit a visual representation of all of the items on a computer screen. The second lies in managing and navigating a hierarchy of folders; “you forget where you put things, or how to find.” Though, Nielsen said, many people “just keep things in one big pile,” which is not an optimal solution, either.
“The old interface served us well,” he said, but “it doesnt really scale.”
Nielsen said that a search feature is, in and of itself, a “major user interface.” In a recent study, Nielsen found that Web users usually head to a search engine first, making that their entry to the content available on the Internet.
But this can cause a problem, Nielsen said, in changing the task from one of recognition–that is, seeing the command in a menu or an icon to click–to one of recall, as in remembering a word to type into a command-line interface. “If you know what to type, you can make it sing,” Nielsen said, but you do have to remember what to type. “Recognition is always easier than recall,” he said.
“What I really want is a combination of the two,” Nielsen said. He applauded Spotlight for supplying a more flexible and meaningful search context, in that it can dig for content and metadata and presents results in a structure that can provide more context, as when items are grouped by author. However, he said, “it looks like the only categorization theyre doing is [that] this is a document–not that it belongs to this project, etc.” Its not just the label thats important, he said, but the meaning of the label.
What Nielsen hopes to see eventually, he said, is a search that returns grouped files, or a search that returns synopses and uses “fuzzy words” to locate content. “You cannot rely on the average person to generate good data,” he said.
But Nielsen said he was overall favorably impressed with Spotlight, in part for its speed (Spotlight indexes files and content when it is installed and adds each object to its index as it is generated), but also as a positive first step in providing an alternative, usable interface in the era of massive data.
Searching for data is a way to use the “richness of language to express ourselves,” Nielsen said, but, he added, “If we have only search as our interface, thats also too primitive.”