IDF, Intel's annual developer forum, starts tomorrow in San Francisco and the hardware company will make one of its most interesting moves into the consumer software world yet.
Included in the event will be the unveiling of two new projects: a developer focused beta testing site called Whatif and a mashup tool called MashMaker. Mashmaker looks like a lot of fun, but I do have some concerns about it as well. Whatif was opened up late last night and currently showcases a debugger, a compiler and and a performance optimizer that developers can download and test out.
MashMaker will open to the public at the end of this week's conference and is the project of greatest interest to me. The service is a cross-platform Firefox toolbar that lets users create their own mashups on the fly and recommends mashups already created for any site you visit based on the MashMaker mashups you've used before. A mashup, for the uninitiated, is a service that adds value to one website by integrating appropriate data brought in dynamically from a different website. They are a lot of fun and have endless potential to be useful.
MashMaker appears to be a relatively accessible screen-scraper backed by a major industry name and implemented with some smart community features. The polite word for screen scraping is "data extraction by HTML" and subsequent users are able to edit the path by field that MashMaker uses to extract that data. Intel says that the identification of various fields on a page, putting them in context and later editing by other users if needed, will help "draw on the wisdom of the community to understand the structure and semantics of information on the web." That sounds like a realistic and exciting claim to me. You might remember that the semantic web was named the #1 future web trend in a Read/WriteWeb post earlier this month. This does, however, raise concerns about proprietary data formats - more on this and other concerns below. All mashups created using the tool are owned by Intel .
You can watch the embedded preview from several months ago to get some idea of how it may look when it's available.
I love these sorts of tools and use 3rd party services to combine websites frequently. This offering is particularly interesting for a number of reasons.
Intel's market footprint, if that word can even describe the size of the Intel community, takes projects like this to a level unattainable by small, independent players. Small startups have to wrestle with the fact that many of their projects are great in theory but don't have tangible value until they can leverage the network effect of economies of scale. In this case, that shouldn't be an issue. When Dell launched a Digg cloned community site for product and feature requests in February called IdeaStorm, for example, that would likely languish in obscurity elsewhere. As a Dell project the site sees items receive thousands of votes.
The world probably doesn't need many more browser toolbars, but this could be an exception to that rule. Passive discovery of new, recommended mashups will be facilitated by the presence of that toolbar. If the toolbar recommended not only existing mashups but also recognized when there were mashup-friendly fields like locations or dates on a page and suggested sites that could be pulled in - that would be a big boost to user engagement.
Ease of use appears to be a big focus of this project. Intel's Brian McCarthy says his mother isn't using or making mashups today and MashMaker aims to change that. Even if mashup creation is only opened up to a subset of at-least-psuedo-technical power-users ala Yahoo! Pipes or Dapper, the toolbar and recommendation functions could certainly make use of those mashups far more widespread. If mashup creation really is made substantially easier than is the case with existing services then MashMaker could be a game changer.
Concerns definitely exist around property rights. Say "screen scraping" and many people rightly ask how outside site owners will feel about services like MashMaker. Dapper, for example, launched with the aim of asking site owners to claim the data feeds extracted from their sites - ultimately creating a market of opt-in participants on top of the catalog of feeds extracted without permission. Services like Dapper are often hard to use, though, and don't have the big name backing of Intel. Unfortunately, most companies online are more likely to accept data extraction from an established behemoth, perhaps with the hope that they will be able to access some Intel money somehow. Site owners who do not want their data extracted can currently opt-out of MashMaker by emailing Intel. This isn't like scraping blog posts for a splog - this is about an exciting future of data portability and web services. I applaud both Intel and the smaller players for taking an opt-out approach, though I hope that smaller players will be able to continue that practice.
It's also a concern that all mashups created are property of Intel. (Update: See comments below, this is no longer entirely true.) This doesn't jive terribly well with the goal of broadening the total mashup ecosystem. Likewise, though utility is a powerful motivator - potential monetization is another one. Perhaps these projects should be open sourced automatically and frozen from editing in their current iterations after a period of time (the Dapper model).
Likewise, if the semantic analysis that comes out of MashMaker is not reusable elsewhere, that will be a major opportunity lost to the web community at large. I hope that Intel will not chose such a path but will work with existing microformat standards, for example. MashMaker is being made available as a research project technology preview - everything about it looks great, except for the risk of this great tool being used to compile semantic web information in a closed and proprietary data format. Perhaps someone involved in the project will comment here about that concern.
Nonetheless, I love this kind of stuff. I don't think that it's aimed at what Teqlo funder Peter Rip called in an interesting post admitting defeat of a similar project, "the mashup rathole" - a solution looking for a problem. The problems faced by other companies are in part problems of scale (in this case, they are too small) and of design - MashMaker appears to be particularly well designed with community features like recommendation. I hope that other small players can find their own solutions to these problems so that the innovation ecosystem can continue to be broadened beyond the umbrella of existing industry giants. Dapper, for example, has enabled users to create a Facebook application (speaking of ratholes) out of any site on the web - that's probably smart. In efforts like this by small startups, Intel's MashMaker could provide some powerful inspiration or it could subsume the bulk of user interest in this sort of tool. We'll see.
Some of the earliest conversation around MashMaker can be found at and via trackbacks to ProgrammableWeb. To read more Read/WriteWeb coverage of this class of tools, see our write ups on Yahoo! Pipes, Google Mashup Editor and Dapper. See also a related, pre-launch company in Santa Clara called DataMash (via).
You can sign up for a MashMaker account now at the project's site. The service should go live at the end of the IDF conference on Thursday. I'm looking forward to it.