Now enterprise mashup platforms, such as Presto Wires from JackBe, are starting to gain traction. In January 2007, an Economist Intelligence Unit survey (PDF - please note that this is a separate report than the Forrester report mentioned elsewhere in this post) revealed that mashups were the most popular traditional web 2.0 technology in the enterprise, with 64% of companies saying they already use or planned to use mashups within the next 2 years.

Mashups come in three distinct flavors in the enterprise, says Forrester:

• Presentation layer mashup. This is the most simple variety. Presentation layer mashups present content from disparate sources together in a unified view. A start page like Netvibes would be an example.
• Data mashup. More complex than presentation layer mashups, data mashups "combine, manipulate, and tie together disparate data sources to present a unified view." An example would be Twittervision.
• Process mashup. Says Forrester: "The most complex of the three, process mashups allow users to mashup not just data sources but also business processes themselves, customizing process design and invoking business logic across multiple applications."

Forrester believes that the enterprise mashup market will hit a tipping points in 2009-2010 and will fold into the IT landscape by 2013. "As a result," writes Forrester, "we expect traditional collaboration and productivity vendors like IBM and Microsoft to ultimately come to dominate the mashup platform market, rolling mashup platforms into major products like SharePoint and the Lotus application suite."

This is just one of many mashups built using the Bit.ly Application Programming Interface, which recently saw a substantial upgrade.

When Google Maps launched -- with maps assembled client-side, in JavaScript! -- I was one of the band of tinkerers around the globe who poked at Google's obfuscated code until we figured out how to embed their maps in our own pages. It was a ton of fun, not only doing the reverse engineering, but seeing the various discoveries and hacks other people were making: embedding multiple maps in a single page, swapping out the map tiles, using custom map markers, making markers move, loading real-time data onto maps...These days, it's hard to fathom a Web without embeddable maps. Wasn't it always that way? To Google's eternal credit, instead of shutting these hacks down, they recognized the demand and legitimized it in the form of their mapping API.

According to the Google Geo Developers Blog post, nearly half of the almost 5,000 mashups on Programmable Web's dashboard use the Google Maps API. In an effort to show the widespread influence of the API, Keir Clarke from Google Maps Mania created a mashup of mashups that we've included below. From real-time tracking of buses in NYC to mapping out news down to a block-by-block level, the mashup shows where across the world Google Maps mashups have been created using the API.

Jeb Boniakowski is the co-founder of Proto and he emailed me to tell me more about the service. He calls mashups "situational apps" and his vision for them is "one in which users combine data and services from all over the place (web, desktop, intranet) and publish the results of their work back out to those types of endpoints, for further sharing, reuse, etc."

You don't necessarily need to be a programmer to develop Proto mashups, due to the graphical interface and drag-and-drop functions. A good example of a Proto mashup is this Salesforce.com app - which generates a report "through a series of calls to the Salesforce.com web services and then filtering, aggregating, merging results, and reporting on Proto's side..." Also check out this Post iTunes track to Twitter mashup, and a YouTube mashup.

What's more Proto is currently running a mashup contest, giving away everyone's favorite prize - an iPod - every day for a month. Jeb told me they're giving away 25 - and the chances of winning are high, given that Proto is still under the radar of most people. So mashup developers should get stuck in and send in an entry. Jeb has even suggested an idea based on a recent R/WW post: a custom tool to analyze the differences in the results from the top 100 alternative search engines for a given set of queries. Excellent idea!

1. Virtual Places
2. Weather Bonk
3. Diggdot.us
4. Flash Earth
5. Adactio Elsewhere
6. Where's Tim Hibbard?
7. Elicit
8. 2RealEstate Auctions
9. Flickrmap
10. Streampad

All the usual API suspects are included in these mashups: Google Maps, Flickr, del.icio.us, Amazon. I tried out Virtual Places and was impressed that it includes little old New Zealand. Here is my hometown of Wellington:

I've been checking out mashups from other sites too, particularly Ning.com. A couple of note are Restaurant Reviews With Yahoo! Maps and Anytown Marketplace With Maps, the latest version of Jon Aquino's craigslists-like mashup.

Also the media companies are coming on board, led by BBC Backstage but also washingtonpost.com recently released an excellent site called mashingtonpost.com. Some interesting mash-ups that people have done already on mashingtonpost.com: News Cloud (a tag cloud), Ripped from the Headlines! (a daily news quiz), world map interface, thumbnail quiz of Arts & Entertainment stories, and washingtonpost.com search results via RSS.

This is an exciting time to be experimenting with mashups, whether you're a publisher, a 'user' or an API-wielding company. As Yahoo's Matt McAlister put it, "the idea is to make your content mash up ready and to build incentives for people to use your content." In that spirit, I've (finally) freed my own RSS feed under the Creative Commons, plus I'm currently trying my hand at mashups. Anyone can do it, so why not give it a try?

Mashups Mashups Mashups

We learned about the Daylife contest today on Programmable Web, the leading blog and database about public APIs and mashups. PW must have felt obligated to be polite and just report on the contest, albeit weeks after the winners were announced.

We were really excited to learn about the contest - Daylife is a company with some good technology, offering news content with some structure to it. What could make more interesting fodder for mashups than structured news data? It turns out almost anything could, if you judge from what came out of it.

If you can't see the video above where we look at the mashup contest entrants, here's a Flash version.

To take a tour of all the applications discussed in the video, you can visit this link.

There Is Still Potential Here

The examples that came out of the contest are all relatively dismal, with the exception of the touchscreen news reading interface. Over on Programmable Web's page about mashups built on the Daylife API though, we found one very cool one. TreeHugger's GRNDX tracks media mentions of a number of words related to the environment. (No one cares about the environment this week, apparently, the Olympics are all anyone cares about.)

That's pretty awesome - even if Treehugger calls it more fun than scientific. Fair enough, but let's see more apps like this instead of the wacky stuff that dominated the Daylife contest.

In addition, Mapstraction "fills some holes each provider's current offerings" - meaning if a certain mapping provider doesn't offer a particular feature that another one does, Mapstraction turns to open source solutions (for the most part) to provide the feature needed.

Those are the technical details, but what's really impressive about Mapstraction are the demos. Take, for example, this demo, a single page showing several maps. As you move around in one map, the other three also move simultaneously to reflect the changes that take place in the map you're using. This is a great way to view comparisons of the different mapping systems side-by-side.

Another feature of Mapstraction allows for tile layer support. This feature lets you toggle different overlays on top of a mapping provider's map, like this one. Clicking the links below this map let you toggle different overlays on top of the Google Map to see maps from 1950, 1912, and 1877.

This demo takes you to a full screen map where you can switch from mapping provider to mapping provider by selecting the name in the box displayed in the bottom-left corner. (Note: to exit the demo, you'll need to know your browser's keyboard shortcut to go "back").

Mapstraction isn't exactly new, having debuted back in 2006. It's currently in use in places like the UK's Nestoria, Mapufacture, Reuters Labs, and even powers a WordPress and MT plugin called GeoPress, which adds geo-tagging to your posts and pages.  However, we would like to see some more Mapstraction mashups - so if you've built one, let us know in the comments.

When two services like this are put together, it is likely that something interesting will emerge. However, not all possible combinations of services are actually useful. To bring order to the often chaotic world of Mashups, John Musser created a site called ProgrammableWeb that tracks mashups and APIs. John's site forms the basis of much of this post. So let's head over there to find out what the big trends are in mashups and APIs today...

High Level API Trends: Google leads the pack

The API section of ProgrammableWeb shows popular APIs. It also has an interesting scorecard that compares API offerings from web giants like Google and Microsoft. A quick look at the score card reveals that Yahoo! and Google lead the pack in terms of number of APIs - 22 and 20 respectively. Amazon, Microsoft and AOL have roughly half of that and eBay has 4. 

Also check out the mashup matrix, which gives insights as to how APIs cluster with each other. Each dot in the matrix implies that two APIs are part of the same mashup. One way to interpret the matrix is that a cluster of dots indicates logical belongings of APIs, or a natural fit with each other. 

Analyzing Mashups: Mapping continues to dominate

The Mashup dashboard on ProgrammableWeb contains a lot of interesting statistics about Mashups. On the right, we see the numbers for the past six months. We conclude that mashups have been growing steadily, but not spectacularly. This is not surprising, because creating mashups requires time and more importantly technical know-how.

The figure below shows the distribution of tags for mashups. This is calculated by analyzing the tags across all the mashups posted onto ProgrammableWeb. Mapping is overwhelmingly the most popular category, spanning 43% of all mashups. This is not surprising because it all started with the Google Maps API - and people are still very excited about putting information on a map. Photos, Search and Shopping tags are roughly equal - with 10% each.

Mashup popularity

ProgrammableWeb currently measures popularity by number of views. This is a bit unfortunate, particularly because there is no way to sort mashups by rating. As a result, the most popular mashup - Virtual Places - is one that does not have a high rating. It looks interesting and claims to be sophisticated, but it did not load for me. The second most popular mashup was called Wii Seeker. As the name implies, it helps people find nearby locations that have a Nintendo Wii. It must have been really popular at some point, but now it looks fairly primitive. The next mashup that caught my attention was the one in fourth place called Weather Bonk. This one had a high user rating and was actually very interesting, as it pulled together live weather, historical information, forecasts and web cameras.

Using Delexa for mashup popularity

The second most popular mashup according to Delexa was GeoGreeting, which is more entertaining than useful. This mashup showed geographical locations with buildings that resembled the shape of the letter that you typed. The third most popular mashup according to Delexa was Weather Bonk, the same one that caught my eye on ProgrammableWeb.

Analyzing APIs: GoogleMaps used in 50% of mashups

Back now on ProgrammableWeb and the chart on the right demonstrates the growth of APIs. As with mashups, the actual growth is not that impressive - but it is steady. As of now, there are close to 400 registered APIs. The way that Programmable Web measures API popularity seems to be more objective and reliable than is the case with mashups. The popularity equals the number of mashups that use the API and the chart below shows the distribution of the most popular APIs:

There are few surprises in this information. Google Maps is by far the most popular API, with 50% of mashups using it. Flickr and Amazon are next with 11% and 8% respectively. Also, it's notable that 5% of mashups are using YouTube - which probably corresponds to the recent rise of video aggregators like Vidmeter. Finally, 4% and total of 74 mashups use the del.icio.us API. 

One API that caught my eye was Cloudalicious, which shows the distribution of tags for a particular URL over time. This can be insightful in terms of understanding how people's perception about your site is changing (or not). Check out the del.icio.us tag cloud trend for Read/WriteWeb:

Conclusion

What can we derive from this analysis? It appears that mashups are certainly cool, but they are not burgeoning. The growth has been steady, but not really explosive. This begs the question: why? 

There are several reasons, the primary one being that most current mashups are created for fun and not for business. Enthusiasts with some spare time on their hands are building these during their evenings and weekends, without having monetization in mind. The second reason is that APIs, as with any software libraries, have a learning curve. Certainly Internet companies are trying to expose their services in the simplest possible way, but not everything can be made simple.

Is this what mashups will be - a playground for enthusiasts? I believe that the answer is 'yes'. Even though services like Yahoo! Pipes, Teqlo and Dapper are working to simplify the process of creating mashups, it will likely remain a fairly technical exercise done by enthusiasts. 

However, it is also likely that we will see companies and products taking ideas from many mashups and creating applications with the combined functionality. For example, taking ideas from the best mashups (like Cloudalicious) and creating a set of tools for bloggers and marketers would be very useful. So mashups will, I think, become the labs of the web - where rapid prototyping is done by enthusiasts, which gives rise to more integrated offerings by web companies.

Please let us know what your favorite mashup is and give us your take on where mashups are heading.

"...people actually use four main search engines for 99.99% of their searches: Google, Yahoo!, MSN, and Ask.com (in that order)."

Well it turns out the figure is closer to 95.45%, and that's counting AOL too! :-) Check out the actual blog post for the full table, with market share for each of the 100 alt search engines and more. But I especially liked these two summary graphs:

The above graph (large version here) shows the long tail of search engines, which is so small because Google, Yahoo et al are so dominant.

This graph (large version) shows the long tail "under a microscope (using a semi-log plot)". As SearchTheWeb2 explains, "it clearly shows that the long tail for the alternative search engines is again dominated by a few such as AOL, Digg and Del.icio.us".

Great stuff! Keep the mashups coming - you can find the other mashups, as well as Charles Knight's original Alt Search Engine posts, on his R/WW profile page.

The result is an app that lets users discover more obscure, independent and often homemade music, rather than a catalogue of music that's only been formally released by record labels. That's because SoundCloud hosts mostly user-generated music, including covers, mashups and original compositions. Some more established artists and labels have even embraced Soundcloud and uploaded commercially released albums to the service. It's essentially the YouTube of audio.

Surpisingly, tapping a song that you find via an artist search does not initiate a recomendation-based station like Pandora or Last.fm. Instead, it adds it to your personalized station. The recommendations come into play earlier in the search process. A search for "The Flaming Lips" returns music by that band, as well as several acts that are deemed similar by Last.fm's data.

Since the content on SoundCloud is so varied - some artists have full albums, some have 30-second clips, while others have only remixes, mashups and covers - the experience of using the app can be a bit varied as well. On the whole, it's a pretty effective way to discover new music beyond the large but inherintly limited catalogs of streaming services like Spotify, Mog and Rdio.

A Different Kind of Mash-Up: APIs Fuel Music 2.0

If nothing else, this is a creative fusion of two APIs from different music services that results in a new music-listening experience. It's the same kind of data mash-up we've seen on a variety of projects that utilize the massive Echo Nest music recommendation data set. For example, one developer created a recommendation engine for Spotify, which is a feature sorely lacking from the popular music streaming service.

Developers have been using the Last.fm API for years to build all kinds of new mashups and apps for discovering and listening to music.

Findory developer Greg Linden listed some problems with mashups as a business on his blog near the end of 2005. He noted that mashups have no service guarantees, there are usually limits on the queries of APIs and limits on commercial use of the APIs, mashups can be numbingly slow, and there are no barriers to entry.

Although Greg concluded that "there is no business model for mashups", in fact there are a number of ways in which mashup developers can monetize their products. The most obvious is advertising.

Advertising

Simplyhired.com is a 'vertical search for jobs' mashup, getting data from job boards, company pages, online classifieds, and other data sources. It also mashes in Google Maps, LinkedIn and PayScale.com. When a user searches for a job - say, a chef position in Seattle - they not only get a list of available chef jobs in Seattle, but on the right of the results are "sponsored listings"

I assume SimplyHired has agreements with the data owners, in which they licence access to the data for commercial means (if anyone can confirm that, please do in the comments). In any case I expect SimplyHired is earning a tidy sum with onsite advertising, because the results are highly contextual and job-seekers would probably be tempted by "send your resume now" ads.

Lead generation and affiliate programs

Another potential business model for mashups is lead generation and/or affiliate programs. Adam Trachtenberg, a developer from eBay, developed a mashup called Dude, Where's My Used Car? (previously known as 'eBay Motors & Google Maps'). It is a mashup of data from eBay Motors listings and - you guessed it - Google Maps, using the APIs of those two companies. What it does is enable users to find vehicles for sale in a location near them. The data in this mash-up is not stored on the host's server, unlike with Housing Maps for example, but is served up in real time.

Explaining his mashup at the 2005 Web 2.0 Conference, Adam said it gave the user experiences they wouldn't normally get on eBay - primarily the visual mapping experience. But crucially the mashup still does a lot of things users can do on the eBay website, for example adding a watchlist onto their eBay profile.

There is also an eBay affiliate program, which is a possible source of revenue for Adam in the future. As of writing, he hasn't implemented the affiliate feature - noting in his To Do list that he needs to add Affiliate link tracking and more information on eBay Developers and Affiliate Programs.

How would the affiliate program work? Well this mashup essentially creates lead generations for eBay, because all results are directed to the eBay Motors website for details on purchasing the vehicle. It's similar to how HousingMaps directs users to craigslist to complete their transaction. Except that eBay actually has an affiliate program, which would allow Adam to earn a kind of 'lead generation' fee each time a user clicks through to eBay Motors and completes a transaction.

Transactional mashups

As mashups mature, we will probably see more examples of 'transactional' mashups - i.e. full-fledged web applications in which users can not only view (read) mashed up data, but do transactions with that data within the mashup.

Taking the eBay Motors example above, what if users could actually complete the purchase of a car from within the mashup itself - instead of being directed to the eBay website. The value would remain the same for eBay, who get the same cut of the action. Indeed they may find purchases increase, because the user can do everything from the one place (the mashup site). It would almost certainly mean more value for the mashup, because users would be able to do more things on the mashup site - which gives more monetization opportunities (e.g. advertising).

In an August 2005 blog post Technology VC Peter Rip speculated that advertising networks such as Google and Yahoo/Overture will eventually create "settlement network models" to enable transactional mashups. While he was talking specifically about advertising networks, there is no reason why this model can't be extended to web applications such as eBay Motors. The elements of a settlement network model would initially be:

"…bundling (1) contracts to police gaming and (2) payment settlement systems to enable the shared value model…"

Enabling transactions in mashups extends to enterprise applications too. In January 2006 Phil Wainewright wrote about a company called NetSuite, a provider of CRM software. NetSuite stores transactional data, such as orders and stock levels, which according to CEO Zach Nelson "makes it a natural choice for creating composite applications - or mashups, as the Web x.0 generation has it - with other services."

Other business models

Other possible business models for mashups include:

• subscriptions
• pay-per-transaction
• premium services
• charging businesses but not individuals
• etc. 

None of these business models I've outlined here has yet been fully proven. But then it can fairly be said that business models for mashups in general are still being explored and there are no easy monetization answers yet.

Flickr pics: The first one is by ancawonka, from the Yahoo! party at eTech 06. The second Flickr pic is by Dave McClure, from the 2005 Web 2.0 Conference session on mashups (which I attended). Pictured are eBay's Adam Trachtenberg, Yahoo's Jeffrey McManus and Google/HousingMap's Paul Rademacher.

Think of it as Trulia plus, with a social justice angle, or Zillow for people interested in more than just the price of a house.

I think this is a great idea and hope the project will succeed. Launching first in Chicago in Q3 of this year, it would be great to see MoveSmart go nation wide. MoveSmart is one of the first entrants in the NetSquared Mashup Challenge, an event that will bring nonprofit mashups together with developers, feedback and funding. $100,000 will be distributed by NetSquared to selected finalists at the event in May. Check out the work over there and consider lending a hand.

• Real words (23) - e.g. Sphere; sub-categories: Misspelled words - e.g. filangy (phalange ‘finger bone’); Foreign words - e.g. hakia (Finnish hakea ‘to fetch’)
• Compounds (20) - e.g. Dogpile
• Phrases (17) - e.g. liveplasma
• Blends (9) - e.g. collarity (collaborative + clarity)
• Made up or obscure origin (8) - e.g. onkosh (something derived from Arabic?)
• Tweaked words (7) - e.g. Clusty (cluster)
• Affixed words (6) - e.g. crossEngine
• Initials and acronyms (4) - e.g. Omgili (Oh my god, I love it!)
• Puns (3) - e.g. gnod (nod, n –> gnostic ‘relating to knowledge’)
• People’s names (3) - e.g. riya

In addition, The Name Inspector has identified some search engines names that belong to more than one category, which he labels "mashonyms". An example is Phrase + Misspelled Word + Domain Hack = alltha.at.

Eight months ago we wrote about the launch of oAuth 1.0, asking if the standard would lead to a flood of mashups across the web.

Google's Support

Last night the Google Data API blog announced that oAuth is now available for all Google Data APIs, everything from Gmail contacts to Google Calendar to Docs to YouTube. This means that 3rd party app developers now have one easy, standardized and secure way to authenticate that their users really own the Google accounts they say they do - without the apps asking users for their Google passwords. That data from Google can then be mashed up with any other application interested in leveraging it.

Google had included oAuth into the OpenSocial framework, but there was little indication that app developers were making use of it. Google's recently launched FriendConnect offered website developers disappointingly little access to their users' data - partitioning the Google functionality into an iframe inside participating pages.

Other Support

We've wondered recently whether oAuth was just a good idea that wasn't really gaining any traction. The list of sites with live oAuth support has been much smaller than we hoped. Now that's changing fast. PhotoBucket offers oAuth support and today SmugMug announced it as well.

We expect to see oAuth authenticating and relying parties spring up all around the web now that coveted Google user data is available through oAuth.

What This Means for Users

There is now no good reason for new applications to ask you for your Gmail username and password in order to access your list of contacts. Don't give it to them - there's a standard, approved way for them to access that data now that doesn't require giving them unlimited access to your entire account.

Apps that don't use the approved Google user authentication method in short order will be acting like a mail carrier who says they have to have a key to the inside of your house to pick up your mail because they aren't familiar with the mailbox on the front porch.

Furthermore, we as users can now expect a thrilling new wave of mashup options that can take secure advantage of our Google data. Google's adoption of oAuth is one of the most significant, tangible moves in support of authentic data portability that we've seen in a long time. App developers should be tripping over each other to make use of this data so that our use of their apps can be made richer, more powerfully useful and engaging. While they are developing to take advantage of Google's oAuth APIs, why not offer some oAuth back out to the world as well? Google's validation of the standard should start a snowball of standards enabled mashups.

We're very excited that Google has taken this step to un-silo our data and support the mutually beneficial ecosystem of mashup developers and users. We're very happy too for the community of oAuth supporters, who have done a great job building and spreading something so needed around the web. Today is a good day for the future of the web.

- CES Storylines (nice quote: "By the time Vista hits the street... it will be less about the OS itself, and more about the vast array of services surrounding it.")

-  Consumer Internet, Digital Media and Technology Conferences for 2006 (very useful list)

- Dave Winer defines Reading Lists: "Mashups" of feeds, forming virtual publications (I'm really excited about what Dave comes up with for Reading Lists!) 

- Neil Budde from Yahoo News talks to paidcontent ("I think the area that's growing and is most interesting, particularly in news, is behavioral.")

- Shore: Neil Budde Looks at Content Blown to Bits by Users ("This will be an exciting year in news aggregation, and tools and deals being developed by Yahoo will be a key part of that excitement.")

- It's all in the mix (Guardian article about mashups... I liked this quote from Ben Metcalf: "rather than simply combining data feeds, the future of mashups is with people providing interfaces that enable you to display data in ways you wouldn't normally see it.")

- TechCrunch: The Memeorandum Hunters (Excellent round-up by Mike. I'll be investigating some of these in-depth soon...)

- Mark Evans ponders Memeorandum's next move ("I suspect he's trying to build a viable business rather than positioning memeorandum for an acquisition by Google, Yahoo, Microsoft, etc.")

- Don Dodge on the meme trackers ("There will be a few winners who make a lot of money, and a lot of "also rans" who will fade away. ")

Flickr pic by cpurrin1