Where will we get business mashup data? Not from Intel's MashMaker.

Business mashups promise to put the means of application production into the hands of subject-matter experts. With the right tools and some know-how, anyone can create a process-centric application that mashes-up data from several sources, all without bothering IT.

Here's a problem with that scenario: Where will we get the services? Creating and deploying web services to use in mashups requires technical skills generally found in IT. Will business mashers stand up SOA infrastructures? Will they code up some SOAP or REST-based web services to deliver the content they need to mash?

Unlikely, and this will become an increasing problem. Business mashers need a way to create these services without the help of IT. It isn't that IT doesn't want to help, but IT is simply too busy keeping the mail server running. Business requests for services will end up in the seemingly bottomless backlog of IT development requests.

It doesn't have to be that way, however. Intel, an unlikely entrant into the consumer mashup business, has announced a program called MashMaker that uses HTML screen scraping to let users mash any web page. Well, I suspect not just any old page. I expect that pages using Flash may pose some problems.

Of course, this is going to cause a fire-storm of controversy. Who owns data on a website and is it legal to pull it into mashups? You can be sure that industries relying on copyrights (music, newspapers, books...) are going to have a hissy-fit. The courts will have to figure that out.

Regardless of the copyright issues, Intel is on the right track. They want to combine del.icio.us-like tags with Amazon-like recommendations to suggest mashup ideas to users. (People who mashed this page also mashed this other page...) That's a great idea, but I'm concerned about Intel's approach. Sure you can mash-up any web-page, but you have to use Intel's mashup tools to do it. Intel seems to be going down the proprietary path, and that's a shame. We've worked long and hard in the software industry and finally have some standards available for constructing loosely-coupled applications. They may not be perfect, but they work.

Intel's MashupMaker isn't set up to work well with others. You can't deconstruct a page and access the data you've extracted in, say, a BPEL service flow or from a BPM engine. MashMaker would be much more powerful if it generated a call-able web service that could be consumed by any standards-based loosely-coupled application.

I've signed up to beta MashMaker, so I'll let you know more when I get a feel for what it can do. Meanwhile, it's a step in the right direction at the very least.