Hackathons often fail on the measure that most of us consider primary: the prototypes. After all, the end of the day is a demo. But lots of hackathon-born experiments don't turn out quite like their creators would like, which can end the day on a sour note.

After a few of those, you're not so excited about hackathons anymore.

Veteran hackathon organizers and judges I spoke to say it's critical to change the way we talk about the goals of the hackathon so participants don't come away feeling frustrated.

Many hackathons succeed in making great social connections between members of the same community or company, and the cross-pollination between skills can jolt creativity. If that's the idea, then make it clear. If the goal is real, deployable projects, then provide a structure and plenty of prep time. Here's how to do both.

Either Get More Technical...

If you want hackers to actually build something worthwhile in the end, try serializing a series of short technical hackathons that build into one larger project.

"The process is the number one thing: it's all about learning this rapid prototyping--how you get from nothing to a totally finished publishable product," says Jonathan Marmor, the founder of Monthly Music Hackathon NYC. "If you can do it in a day, then certainly you can do it with a bigger project in a month."

Marmor, a composer and engineer, says the music hackathon series attracts people with a wide variety of backgrounds, from instrument builders to programmers to music entrepreneurs.

"They're sharing different pieces of what they're working on with each other," he says. "Maybe somebody's building a web app and ends up talking to a scientist [about digital signal processing]."

And they're each just one day each, so there's not enough time to build out a large-scale project.

"Eight hours is not really enough for me to start a composition project from scratch and finish it and get the score done and then get an ensemble to rehearse it and perform it," he says. "I will take some nugget of an idea and try to develop that."

But that time limitation can be a powerful benefit, teaching participants to quickly prototype an idea and get feedback on it, he says.

"If you do it every month, you're going to become fluent in this process, and when you have a big idea you're going to be able to execute on it and sort of know the pitfalls of creating a minimal viable product and turning it into something bigger," says Marmor.

Make sure hackers know the requirements and have time to plan and prepare well before the event. And if the plan is to let participants tinker with new technology and build simpler prototypes, make sure that's clear, too, so hackers know they have the freedom to experiment and practice new skills before the starting gun goes off.

...Or Be Clear Your Hackathon Is More Casual

Not every hackathon project is going to become a full-fledged product, and it's important to make sure participants understand that so there's no hard feelings after the fact, says Mike Curtis, the vice president of engineering at Airbnb, which holds regular internal hackathons.

"The purpose of this is to create lots of ideas and to have fun together as a team, but not everything is going to ship to production," he says.

A few internal tools used by the company did evolve from hackathon projects, says Curtis.

"There's a few examples of things that we've built in hackathons that have ended up just being disruptive internally, how we operate day to day," he says, including an internal link shortener and an improved company directory system. Airbnb's holiday card program, that lets hosts and guests swap virtual greeting cards, got its start as a hackathon project, too.

The goal isn't to build a complete product from start to finish during the hackathon but to let good ideas materialize that can later potentially be fleshed out, he says. The hackathons also let employees who wouldn't normally work together collaborate on projects, he says, something that was echoed by Facebook's Pedram Keyani.

"We create this kind of alternate social graph," says Keyani, forming ties that help different departments work together even after the hackathon is done. "The social connections there are insanely valuable."

Some high-profile features like Facebook Chat grew out of hackathon ideas, but projects aren't limited to engineering and code, he says.

"Nonengineering teams will do hackathons," he says. "They'll say, 'let's rethink how we run Legal.'"

Hackathon teams with promising ideas get to present them to Facebook executives including CEO Mark Zuckerberg, but it's understood that some of the most interesting ideas won't have a clear path to becoming products, he says.

"The instinct is to try to create some kind of prize system or incentive for people to hack or come up with cool ideas, but that sets up very shortsighted views of projects and gamifies it in a way that I think is unhealthy," says Keyani. "For us, hacking is the prize."

Hackathons let participants push their own limits as creators and developers and also get some experience pitching their ideas, says Prachi Gupta, a senior engineering manager at LinkedIn.

"You become a salesperson by the end of the day," she says. "It's an amazing transformation that you can go through in 24 hours."

LinkedIn organizes internal hackathons as well as public events like the DevelopHer hackathon for women and an annual hackathon for interns from companies across the tech world. The public hackathons have helped recruit potential employees, but that's not the only reason the company sponsors them, says Gupta.

"If you're only looking at hackathons as a recruiting tool, that's not good at all," she says. "We want people to come together and learn from us about things that we know how to do, and to teach us better about things that they know how to do."

Letting participants learn from one another was a big motivation for the Tribeca Film Institute in creating its Tribeca Hacks series, that brings filmmakers and other traditional storytellers together with designers, developers, and engineers, says Opeyemi Olukemi, Tribeca's manager of digital initiatives. (Full disclosure: I participated in Tribeca's Story Matter science storytelling hackathon last month.)

"It was really about understanding the lack of space for people to learn," says Olukemi, explaining that when Tribeca Hacks launched in 2012, filmmakers didn't have a place to learn to work with technical people to develop new kinds of content. Now, she says, participants from previous hackathons have gone on to apply for funding for larger projects from Tribeca's New Media Fund, and one hackathon project, a physical version of the iPhone game Flappy Bird, is raising funds through Kickstarter.

"The interdisciplinary collaboration across our hackathon is exploding, and I'm very proud to see that happen," she says.

How To Filter Hackers' Ideas

For hackathons where the primary goal really is to produce viable projects and ideas, it helps to make sure hackers are well informed about the problems to be solved, says Mike Mathieu, the cofounder of Walk Score and a sometimes hackathon judge who has previously written about the limitations of the format.

The Sunlight Foundation has also seen that giving people a clearly defined task leads to more useful code, as when the open government group asked attendees at the PyCon conference to write scrapers for their states' legislative sites.

"We had a very relatable task: iI was very well-documented and replicable," says Tom Lee, the director of the foundation's Sunlight Labs.

But even less structured events still have brought Sunlight some useful ideas and helped motivate supporters and potential employees, he says.

"It is also a way to bring in new people, whether that's from a staffing perspective or just sort of an organizational awareness and support perspective," says Lee. "People get a little more excited about our mission when they learn a little more about it, even if they're not going to become a coder themselves."

When Mark Zuckerberg's Harvard classmates first logged in to TheFacebook in February 2004, the site's servers ran PHP, which had beat out Perl to become the hottest language on the web.

Using a now-popular framework like Ruby on Rails or Django wasn't an option--Rails' first public release was a few months later, and Django wasn't unveiled until the following year. A decade later, PHP's been widely derided for having a sprawling library of inconsistently named and defined built-in functions, syntax and semantics just different enough from related languages to confuse multilingual programmers, and a history of design decisions that made it easy to write insecure code.

"Every PHP programmer is familiar with day-to-day tasks that can be tricky or cumbersome," Facebook developers Julien Verlaguet and Alok Menghrajani recently wrote on the company's engineering blog.

But PHP hasn't gone away--Facebook and other big organizations and projects have millions of lines of code written in the language, and programmers still appreciate it for rapid development and deployment, even as they try to steer clear of its messier features.

To ease the pain of PHP programmers without making them abandon the language and years of software development, Facebook developed Hack, a new, PHP-derived language that's largely compatible with existing code and augmented with new safety features derived from functional programming languages and academic research.

"It has been specifically designed to interoperate seamlessly with PHP," says Verlaguet, the technical lead on the Hack project, whose background includes a mix of formal academic study of programming languages and industry experience. Facebook's been using and developing Hack internally for about two years, and has recently made the project open source and scheduled a public "developer day" for April 9.

"What we're doing is basically making Hack available out there to hopefully gather feedback from the community, and work with the open source community to make Hack a good experience for people outside Facebook," says Verlaguet.

Perhaps chief among Hack's innovations is the introduction of automatic type inference, a concept familiar to users of more esoteric programming languages such as Haskell and ML but less common in more mainstream languages.

Traditional PHP is dynamically typed, meaning the basic nature of a variable used in code--that is, whether it's a number or a text string or some other type of data--isn't specified until the program's actually running. Programmers enjoy that flexibility, but it creates room for errors that aren't possible in statically typed languages like Java or C, where the type of each variable is explicitly defined when code is written.

Hack takes a middle road: It lets programmers specify the types of some variables in their code and uses logic to infer the rest based on how variables are used together, issuing an error if the code's logically inconsistent. That concept itself isn't new, but it's previously been used in compiled languages, where programmers are used to waiting for their source code to translate into a form executable by the machine, and not in languages like PHP where programmers expect their code to be executable as soon as they hit save, says Verlaguet.

"The solution lies in building the type checker as a daemon," he says, referring to a background process that runs on a developer's computer. Instead of waiting for the programmer to explicitly run a compiler, the type checker asks the operating system to notify it when source code files have changed, similar to how services like Dropbox get signaled when a synced file needs an update.

"The kernel event that says that a file has changed is the starting point," he says. "Then, the new file is processed, and once the new file is processed, the two versions are compared to deduce what must be rechecked at a very fine-grained level: at the method level, not at the file level."

Individual methods that have changed are re-examined by the type checker, which makes sure they're still consistent with what it already knows about the rest of the code. Looking only at what's actually changed makes the type-checking process fast enough that programmers don't have to wait for it to run, even when they switch to new branches in a version control system like Git, Verlaguet says.

Hack also introduces other new features, like enhanced collection types such as vectors and sets to augment the PHP array and better support for short, anonymous functions used in functional programming. The new language lets Facebook gradually update its existing PHP codebase and still benefit from its longtime investment in PHP, says Ed Smith, the technical lead on Facebook's HHVM runtime engine, which will now support both Hack and PHP.

"Hack enables us to dynamically convert our code one file at a time," Smith says. "Switching to another language would be a lot more difficult."

It's too soon to say which other companies and projects will jump on the Hack bandwagon, with the project just made open source, Verlaguet says, though he notes the reception so far has been positive.

College students today stream movies from Netflix, queue up music on Spotify, and order late-night snacks on Seamless. But when it comes to buying textbooks, many students are still doing things the old-fashioned way: buying pricy paper copies from the campus bookstore at the start of the semester, then selling them back for a fraction of the purchase price when classes are done.

E-books were supposed to be a panacea, but the Kindle and iPad went mainstream and still relief never came. Companies trying to disrupt the industry say it has evolved slower than other content fields because the market is more indirect.

You see, textbook publishers market to professors who pick the books, not students who pay for them--where Apple and Amazon have traditionally directed their marketing. The key to innovation, these companies say, is to not try to beat the big publishing houses at their own game.

"Their customer base is not the student," says Nathan Schultz, the chief content officer at Chegg, which offers textbook rentals, e-textbooks and online study help. "Their customer base is the faculty member and, in some cases, the actual institution."

And every year brings a fresh batch of students looking to start college off right, making them wary of waiting for delivery of an online book, let alone experimenting with other ways of learning the material, says Texts.com CEO Peter Frank.

"Unlike with ordering dinner, students, especially younger students, are very unwilling to do what they perceive could put them at a disadvantage," says Frank, whose company operates a combination textbook marketplace and price-comparison engine. "They really just want to get off on the right foot."

And for students shopping with a parent's credit card, there's little incentive to shop around.

When Disruption Doesn't Work

So what's the solution? Find ways to offer students what the existing marketplace isn't delivering. "This has been attempted so many times, and the past is just riddled with failures," he says, even pointing to a list of "stupid fratboy business ideas" that includes the concept.

The problem, he says, is getting enough buyers and sellers onto a site so deals actually get made. Texts.com is attempting to get around that problem by offering buyers price comparison services for other merchants and making money off referral links as the marketplace grows.

But the students aren't the only source of inertia in the system, says Frank. Professors, many of whom grew up exclusively with print books, are often reluctant to experiment with alternatives.

"The textbook sales cycle is kind of like the pharmaceutical sales cycle," says Ariel Diaz, the CEO of Boundless, which develops interactive,cross-platform textbooks. "The one making the decision is not the one making the purchase."

And, says Chegg's Schultz, traditional publishers developing e-textbooks often contribute their own institutional inertia and just aim to render the successful print version on a screen, not build a new, interactive product.

"In general, publishers are saying don't mess with my book -- I just want you to create a digital representation of that same thing I sell in print," he says. "I'm losing the ability to create experiences for the student like 'turn my book into a flash card set' or 'turn my book into an image gallery, so I can just use the images to study from.'"

The big publishers are developing online supplements to textbooks, like Pearson's MyLab and Wiley's WileyPlus, he says. But those platforms also focus on the needs of instructors, he says.

"It makes the life of the faculty member a lot easier," Schultz acknowledges, offering features like online quizzes and reports on which students are at the top and bottom of the class. But those top-down tools might not match students' study needs, he says.

For that reason, Chegg tries to build student-centric tools, offering online study aids and expert help that supplement what students get from traditional textbook makers.

"We don't see ourselves as designing to replace that kernel," he says. "We see ourselves as an augmentation."

Considering The Professor As End User

Similarly, Boundless began by marketing its books directly to students as a low-cost supplement or alternative to their assigned textbooks, says Diaz. Boundless' textbook content is generally available for free under a Creative Commons license, and for $20 per book, students get the full interactive package with additional study tools.

"I think the key is to find innovative ways to reach the market," he says, explaining some students use Boundless' interactive books to study and then borrow a copy of an assigned textbook from a friend or the college library solely for the homework problems in the book.

"We started with a focus on students to make sure that we're building great products for them," he says. "In addition, along the way, we saw an increasing number of educators coming to Boundless to use the content in their classroom."

Now, the company's piloting customizable versions of its textbooks that professors can easily tweak to create particular reading assignments or add quizzes. With no print revenue stream to worry about or rely upon, it can easier for Boundless to innovate than a larger publisher, he says.

"I think that everybody in the industry sees that this is the future," says Diaz. "It's like any industry in transition: it's about figuring out what that transition looks like."

Every computer science student learns the basics of graph theory--a set of mathematical abstractions for modeling networks and the connections within them. But few developers ever work with network data on the scale of Facebook's social graph, with its more than 1 billion users and hundreds of billions of connections, or edges, between them. So in 2012, when Facebook engineers started thinking in earnest about ways to efficiently process all the network data they'd amassed, they found themselves in uncharted territory.

"As far as I know, there's been no work to be cited working with graphs as large as a half a trillion or a trillion edges," says Facebook engineer Avery Ching.

But Facebook still wanted to be able to measure basic statistics about their users' connections, and efficiently do computations like measuring affinities between users to suggest potential friends, or finding mutual friends of sets of users. To make this possible, they turned to an open source graph theory toolkit called Apache Giraph, derived from code originally donated to the Apache Software Foundation by Yahoo.

Making Giraph work at Facebook's scale required a bit of engineering, which Facebook documented on a company blog and contributed back to the open source effort, but Giraph's overall computing model proved to be the ticket to parallelizing huge network graph computations across Facebook's servers.

"It was really easy for us to use," Ching says. "The model itself was very expressive--it allowed you to do a lot of different things."

How Giraph Works

Giraph is based on what's called the bulk synchronous parallel model, a class of parallel computing algorithm developed in the 1980s by researchers led by Harvard computer science professor Leslie Valiant. BSP algorithms take place across multiple computers over a series of steps, and at each step, each computer does a bit of computation and, if it wishes, sends messages about its results to others in the system, which they can incorporate into their computations at the next step.

In Giraph's case, the individual computations are done as if from the perspective of the individual nodes, or vertices, of a graph. And at the end of each step, each node can choose whether to send a message to its immediate neighbors. Before Giraph, that model had been proposed in a Google paper describing an internal tool called Pregel, and Facebook engineers found it's an easy way to describe many graph theory problems in a way that makes them possible to parallelize, by having different computers handle simultaneous computation on behalf of different sets of nodes.

"For large-scale graph computation at Facebook prior to Giraph, people were trying to [use] frameworks such as Hive and MapReduce," says Ching. "You can do it, but it's just really, really slow--think 50 to 100 times slower than it is today."

Finding The Degrees Between People

As a simple example, Giraph's manual presents an implementation of the single-source shortest path algorithm, which measures the distance from any particular node to every other node in the network.

Logically, the length of the shortest path to any particular node is the length of the shortest path to one of its neighbors, plus the distance from that neighbor. So, under Giraph's model, each node begins each step by figuring out the shortest path it knows to itself from the start node, notifying its neighbors if it's learned of a shorter path since last messaging them. Those messages let each node redo its own computation in the next step and again see if it's learned of a shorter route; when a step ends without any nodes needing to send revised paths to their neighbors, each node knows the best path from the source.

And Giraph works for more complex problems, too: A Facebook blog post from earlier this month described how Giraph can be used to assign Facebook users' data to particular database servers. Facebook runs more efficiently when users' information is stored on the same servers as that of their friends, since it means fewer cross-server lookups for common operations, but maximizing the number of users on the same server as their friends is hard with a billion accounts.

The Giraph approach assigns each node to a particular set, representing one server, and lets it tell its neighbors where it's located. At each step, it randomly decides whether to move to another set, with the probability of a move proportional to the number of neighbors in that set.

After a few dozens iterations, that algorithm improved handily on simply assigning nodes to sets based on geographic location, according to the post.

"An early adopter within the company was a service in which typical queries involve fetching a considerable amount of data for each of a person's friends," according to the blog post.

And Facebook remains committed to improving Giraph and contributing the changes it makes back to the open source project, so that it can be used by others dealing with graph data, although some of the algorithms it runs on top of the platform are only shared internally, Ching says.

Before any of Facebook's improvements to Giraph itself are deployed internally, the changes are pushed to the Apache project's open source code repositories, he says.

"We contribute everything we do first to Apache, and then we pull it back through trunk," he says, referring to the project's development branch.

"We work with some researchers and other folks outside the community to just advance Giraph generally," he says. "If people are going to use Apache Giraph, they should use it in the best way possible, which is the version that we're using."

Since the days of the Old Masters, it's been difficult to tell an authentic masterpiece from a fake. And once artists like Andy Warhol began using mass production techniques in their own studios, authentic originals became even harder to distinguish from knock-off copies, leading in Warhol's case to high-profile disputes about what constitutes a genuine work. (And that's to say nothing of the added complications of copyright to authorship.)

With purely digital forms of art, what it means to "own" a work gets more complex still--where an entire work is ultimately a series of bits on disk, what is the gallery actually selling?

"I've seen firsthand the difficulties that artists working digitally have in participating in the traditional art markets," says artist and New York University professor Kevin McCoy. "Non-object-based art doesn't have the same traction."

It's so difficult for digital artists to sell their work through traditional gallery channels, in fact, that some leading some artists have changed their practices, adding tangible components to their art to make it more marketable, says McCoy.

The more digital art sells, however, the more likely it will be counterfeited. To solve that authentication problem, McCoy and serial entrepreneur Anil Dash created Monegraph, which allows artists to digitally sign their work using the cryptocurrency Namecoin. The signature, along with a digital title recording who owns the work, is preserved in Namecoin's public shared transaction registry, or blockchain, which is essentially a modified version of Bitcoin's blockchain letting users publish arbitrary data. If owners choose to transfer the work, they can publish a record of the title change to the blockchain as well.

"Only one copy of a digital image can ever have a valid monegraph signature," McCoy and Dash wrote on the Monegraph website. "Monegraph images are just ordinary image files, so they can be duplicated and distributed like any other images, but only the original file will pass validation against the monegraph system."

Namecoin is primarily designed to serve as a decentralized domain name registry service, where users can register, renew, and transfer Internet domains without the need of any central authority by publishing to the blockchain. But the blockchain is designed to let users store arbitrary data in key-value pairs, not just domain name records, so it can be used for other purposes, too, like recording the title to artwork.

To register a work through Monegraph, the artist records a URL pointing to the piece, the piece's cryptographic hash, a brief explanatory, and a second URL pointing to a more detailed announcement of the claim. That claim gets posted on a site known to belong to the artist, such as his or her Twitter account, verifying it's indeed the artist registering the work.

McCoy and Dash launched Monegraph at The New Museum in New York on May 3, as part of the digital art group Rhizome's Seven on Seven, essentially a hackathon pairing influential figures from art and technology. Now, they're working on making the system easier to use for artists, McCoy says.

The initial version was designed with images in mind, but once the platform is further built out, there's no reason the same concept wouldn't work for authenticating and tracking ownership of any kind of media file, from music to film to the written word, he says.

"Everyone in the art world that's dealing with technology, they get it right away," says McCoy. "Now, we're at the beginning of the process of trying to roll it out, from a protocol to a platform, which is a bigger job."

The web charting library Highcharts isn't widely known outside of data visualization circles, but it's used by organizations like IBM and NASA, the BBC and the Financial Times. Why? It's easy to use, well documented, and--most importantly--compatible with browsers from Internet Explorer 6 forward. That's a big deal, because IE compatibility hamstrings a lot of other sites and tools that work fine on Chrome, Safari, Opera, and Firefox. But that kind of cross-compatibility means extensive testing on all sorts of devices. Here's how they do it.

First, a primer on what Highcharts does. Developers can use it to add interactive elements, line and scatterplot graphs, as well as more a few more exotic visualizations like adding speedometer-style gauges to their website, by filling in the blanks in a JavaScript object which specifies everything from the chart margins to the formatting for the tooltips on the individual data points to the dataset itself.

Then, when a user visits the website, the library draws a chart to a specified element on the website, using SVG graphics or VML, the pre-SVG vector markup language used in older versions of Internet Explorer.

"When we first talked about it some four or five years ago, IE8 support was a big selling point," says Highsoft CTO Torstein Hønsi. "It's becoming less of a selling point now, but it's still very important because there's so many enterprise systems and in-house solutions than still run on IE8."

Big Time Testing

To make sure new versions of the software still work with browsers from the latest versions of Chrome and Firefox to IE6 to the native browser from Android 2.0, Highsoft runs an extensive series of tests before each release, Hønsi says.

"We run automated visual tests for the SVG-enabled platforms," he says. "We have a test suite of about 1,000 samples and we compare programmatically rendered version of the charts with the previous version of Highcharts."

The automated tests typically finish in a few minutes, but for other platforms, the company will often spend a day or two running manual tests and fixing regression bugs, he says.

"For VML and for different browsers, we don't do all the automated tests, but we do selected tests manually, so we have a recipe for manual testing like IE8, IE7, even IE6 and also mobile browsers," he says. "For each feature, like a certain touch gesture or a graphical visualization in legacy IE, we have a manual test."

And for each testable feature, the company also provides at least one complete demonstration in a JSFiddle online debugging environment so coders can experiment before deploy to their own sites, he says.

The extensive set of built-in options for basic charts and copious documentation makes Highcharts appealing to corporate customers, since it means new developers can more quickly get up to speed than with more abstract libraries like D3.

"Versus D3, Highcharts has a much lower entry point, so if you hire someone to create some charts on your website, then they will get started much quicker with HIghcarts than with D3," says Hønsi.

What Testing Gets You: Love

Media organizations praise the library for the same reasons: Journalist Malik Singleton tweeted last year that the "best time to master @Highcharts API is on deadline," and editors and producers from organizations from the Washington Post to Southern California Public Radio have blogged about using Highcharts to build visualizations for news stories.

More advanced developers can write more complex code, though, adding JavaScript code that gets called when users interact with points on the map or drawing graphical elements like lines, points, and arcs atop the chart.

An interactive mapping module for Highcharts, called Highmaps, is currently in beta--Hønsi says that despite the wealth of web-mapping toolkits already available, from Google Maps to Leaflet, mapping was one of the company's customers top requests.

"What our clients are saying is they don't want to use two different tools if they can do it with one," he says. "They prefer not to use another tool for the mapping, so we when we released a Highcharts map product it can also run side by side with Highcharts as a Highcharts module, which, of course, is much more efficient than learning two different libraries."

At the moment, the Highcharts JavaScript library needs to be installed on a customer's site and invoked through developer-written HTML and JavaScript. The library's free to use for noncommercial projects, with various license and support options available for commercial use.

But in the near future, Hønsi says, the company will be rolling out a cloud-hosted version of the tool that won't require any coding at all. Highcharts Cloud, currently in alpha, generates charts that can be embedded on a website or shared across social media, he says.

"We have created a beta now, which we launch in two weeks," he says. "We're also very, very close to coming up with a business model and a subscription model."

For years, specialized tools like R and SPSS have been standard for anyone working in statistics and analytics, whether in private industry or the academic sector.

But for many in the business world, those languages are unfamiliar, which means companies haven't been able to leverage these languages the way that, say, universities have. But that could change thanks to an open-source Python data-analysis library called Pandas, which offers many of the same analytics tools as R in a language developers are already using, but in languages businesses can work in.

"One of the reasons we like to use Pandas is because we like to stay in the Python ecosystem," says Burc Arpat, a quantitative engineering manager at Facebook. "We have a lot of systems inside Facebook, or infrastructure that allows us to either use Python to talk to those systems or integrates with Python very easily or is written in Python."

Many of the engineers working on analytics projects at Facebook are well acquainted with R, which the company also uses for certain tasks, but Facebook's existing Python codebase often makes Pandas easier to work with. That's a common reason for developers to choose Pandas, says the library's creator, Wes McKinney.

"In companies that have an engineering culture, just because of the general growth of Python, it's made Python and Pandas an easy choice," he says.

McKinney began work on the library while working at the financial firm AQR Capital Management, where he was using R for quantitative finance projects and basic "data wrangling," he says.

"I was frustrated on multiple fronts," he says. "I felt that R was not strong enough for software engineering--for building big software, R left a lot to be desired as far as the tooling for debugging and building big systems."

McKinney started building his own toolkit in Python, which he saw as better suited to larger projects, and that evolved into what was open sourced as Pandas.

Like R, Pandas is oriented around manipulating "data frames"--two-dimensional matrices of structured data similar to a well-organized spreadsheet or a SQL-style relational database. Pandas has built-in support for quickly reading in data frames from Excel spreadsheets or comma-separated value files, filtering rows and columns, and generating aggregate statistics like sums and means.

Those data frames can also be passed to other Python number-crunching libraries, like the powerful statsmodels package that handles linear regressions and more esoteric statistical tasks, or the Scikit-Learn machine learning toolkit.

And having stats-intensive coders work in Python means their work can more easily be integrated with production code, says Dave Himrod, the director of ad-quality engineering at the Internet advertising exchange AppNexus.

"In a lot of places, your data scientist or your quants or analysts, whatever you call them, they analyze the data and create this model in R or SPSS, and then an engineer has to take that and translate it into whatever your production system is," he says. With Pandas, that's often not necessary, says Himrod.

"It's nice that you can have your production environment and your researchers all using their same tools," he says. "They just say, 'Here's what's going on in this code,' and everybody just knows the basic operations of Pandas and some of the [Python math] libraries like SciPy or NumPy."

A new version of Pandas due out this month also offers better integration with a variety of SQL databases, says Jeff Reback, one of the lead developers on the project. Reback says Pandas' ability to translate between different structured data layouts, like SQL, Excel, and more obscure formats like HDF5, is a crucial advantage in a world where Python has come to be the intermediate language connecting companies' various computer systems.

"Python is now the new glue language," replacing earlier choices like Perl, says Reback.

Still, Pandas users and developers emphasized, there's still a place in the statistical world for R, which McKinney says has really improved in terms of engineering tools since he began work on Pandas, thanks to the team behind the IDE RStudio and library developers like Hadley Wickham.

"The work that R-Studio has done to build a better development environment for R has really changed the game for R programmers," he says. "R and Python are both rapidly growing--and taking market share away from proprietary tools like SAS, SPSS, and Matlab."

Today, Firebase is a popular backend-as-a-service company, letting developers sync their apps and websites to Firebase's cloud without having to worry about database management and scalability.

But the service didn't appear overnight--Firebase actually evolved from two prior startups launched by company founders James Tamplin and Andrew Lee. Tamplin, the company's CEO, says that evolution wouldn't have happened without their working hard to find out what their customers actually wanted.

Tamplin and Lee first started SendMeHome.com in 2008. The service started by generating printable tracking tags for users' valuables and evolved into a social networking site geared around real-world objects.

"To begin a story, a user must register an object at SendMeHome.com," that company explained in 2009. "The object is then passed from person to person; each person who receives the item can use its unique SendMeHome ID as a key to add a chapter to the online story."

SendMeHome wanted to let its users chat in real time but struggled to find a reliable chat tool to drop into the website, Tamplin says.

"There were absolutely no good ones out there, so we ended up building our own," he says. Then, Tamplin and Lee realized their problem probably wasn't unique, so they decided to focus on making embeddable, real-time chat widgets for other sites.

The two founded Envolve, which launched its chat-as-a-service platform in 2011, with backing from Y Combinator.

"Their main advice is go and talk to your customers and figure out what their needs are," Tamplin says of the incubator. "This is actually one of [Y Combinator founder] Paul Graham's famous theories, or maxims: You're always one or two hops away from something huge, and you just need to invest in good teams who can execute and make those leaps."

And, after releasing its chat API, Envolve discovered a lot of the messages passing through its system weren't actual chats at all--developers had started using Envolve's chat API to as a way to sync data across their users' computers.

"Our gaming customers were trying to send game data through the chat system," says Tamplin. "Instead of just letting the users talk to each other, they were sending character hit points through the chat system in these private rooms."

It turned out many of Envolve's customers had skilled front-end developers but just didn't have the resources to set up real-time backend syncing for their apps.

"There were very, very few developers who could build these scalable real-time systems, real-time backends," Tamplin says.

So the founders decided to pivot again, and separate the real-time synched database backend, which became Firebase, from Envolve.

"We ended up divorcing the interface of chat from this real-time architecture," says Tamplin.

And Firebase is still taking customer feedback into account, he says. The company even holds a weekly in-person and online office hours in its San Francisco headquarters and goes out of its way to be helpful to developers, whether they're existing customers or just potential ones.

"We go to hackathons and we'll help people build their projects, whether they're using Firebase or not," says Tamplin. "We'll stay up all night and always have someone at our sponsor booth when all the other sponsors have gone home."

Firebase also just added hosting for web assets, like HTML, JavaScript, and image files, so developers don't have to set up separate accounts for webpages that are simply going to communicate with Firebase's backend. That, too, was in response to user requests, says Tamplin.

"We built the database, but we just didn't have the method to deliver the HTML, CSS, JavaScript, images, et cetera," he says. "And that's where people were getting hung up."

Not that long ago, being able to quickly glance at a powerful handheld computer to pull up customer information during a meeting or sign off on a deal while on the treadmill would have seemed like science fiction. But now that we take smartphones for granted, scrolling through a phone menu during an intimate meeting can look awkward, and reaching for an iPhone while out for a run just feels like a pain.

Salesforce,com is betting that the future lies in even less obtrusive wearable devices, like Google Glass and the Samsung Gear and Pebble smartwatches. On Tuesday, the customer relations management and business cloud computing company unveiled a new platform called Salesforce Wear that lets developers connect wearable devices to the company's cloud systems, letting users access business data through a quick tap of a watch or simple voice commands.

"We really think wearables are the next wave," says J.P. Rangaswami, Salesforce's chief scientist. "The rate of change is itself quite surprising, quite amazing."

One sample app lets users page through calendar and contact entries for a day's entries using a Samsung Gear watch; another lets resort operators use the Bionym Nymi biorhythm-based identification armband to authenticate and cater to VIP customers as they move around a hotel.

A Google Glass example lets industrial inspectors verbally and photographically log inspection results and open trouble tickets in the Salesforce cloud, even in situations where dirty hands or protective gloves would make it hard to use a touchscreen.

Those example apps, along with developer documentation, are open source and available in GitHub, so Salesforce's customers can build their own wearable device apps for their own needs.

"Most importantly, 1.5 million developers will be able to drive their ideas, their ingenuity into this new device while also making our customers' lives easier," says Rangaswami.

Salesforce already provides smartphone and tablet apps and APIs through its Salesforce1 platform, but the company wants to let its customers access and build the right tools for different wearable devices, not just smartphone apps delivered on different screens.

"All of these new devices have different application architectures, UX patterns, and data flows," said Adam Seligman, the company's vice president for developer relations, in a statement. "The Wear Developer Pack handles the identity, secure API access and plumbing necessary to connect the device to the Salesforce1 Platform, letting the developer focus on innovative new use cases."

When users first moved from desktop to laptop computers, they used the same software on smaller screens, says Rangaswami.

"All that was really changing was form factor and portability," he says. "The nature of the information you engaged with remained the same. You carried your desktop experience on your laptop in the early stages."

But with wearables, Salesforce plans to work with hardware and operating system makers and with its own customers to develop the right interfaces, notification systems, and levels of detail, he says.

"It becomes like an iterative learning process for the whole community," he says. "And we will become a part of that."

The founders of Kiwi Wearables set out about a year ago to build a wearable computer controlled through state-of-the-art motion sensors and gyroscopes, says cofounder Ali Nawab.

"We basically were looking to build a product that relied on the power of all the sensors that were available," he says. "Kind of like how voice became popular with Siri, we were looking to push motion as the source of interaction."

Of course, they also wanted to build a machine consumers would actually find useful. In the validation process, they found potential customers were definitely interested in some of the features a wearable smartwatch could deliver, but they had major misgivings.

Their research indicated people wanted to be notified of phone calls and texts without having to pull out their phones, and they worried about missing calls from their phones when deep in a pocket or purse. They wanted to be able to interact with other electronics--dim lights, advance PowerPoint slides, find a missing iPhone--using something always available at their fingertips.

"We spent a fair amount of time with focus groups and people just trying to figure out what problem or issues they have on a day-to-day basis," Nawab says.

But, Kiwi found, a big part of their target audience already wore a traditional wristwatch they were quite happy with and they weren't necessarily looking to trade it for a newfangled gadget--even it came with "smart" features. So, Kiwi created something it calls Glance, billed in an ongoing Kickstarter campaign as "a smart accessory for your watch" that attaches unobtrusively to an existing timepiece's watchband.

"We thought there needs to be a product that works with your watch that you already have to give you all the notifications and features that you would expect with the smartwatch," Nawab says. "It shouldn't disturb the aesthetic of the watch that you have or the comfort level that you already have with the watch that you've been wearing for a little bit."

The Toronto-based company plans to start shipping the Glance in October 2014, along with Android and iOS apps to let the devices interact via Bluetooth with users' smartphones, says Nawab. Kiwi's actively talking to Kickstarter supporters about what features to prioritize for the apps and to developers about a growing SDK to let third-party tools interact with the devices, he says.

Developers will ultimately be able to access the Glance's API from smartphone apps or through web apps through a web socket interface, and train the device's motion detection systems to listen for certain user movements. A Unity3D plug-in will eventually help programmers integrate the device as a game controller, Nawab says.

"A large portion of our motion recognition tools have been built up incrementally as we were getting more and more input from the developer community," Nawab says.

But at launch time, Kiwi plans to focus on the iOS and Android apps, and on features most exciting to end users.

"With our early backers, we've already started the dialogue with them and we're collecting their insights or their feedback on what they're most excited about," he says. "A typical consumer is not really that interested in the machine learning engine or how it works--what they're interested in is use cases: How is this going to help; what am I going to do with it?"

It's a little more than a year since Google launched Blink, a custom engine used by Chrome to turn HTML and CSS code into what you see on your screen. Before that, Chrome was powered by a tweaked version of WebKit, the Apple-led open source engine used by Safari.

That split sounds like a headache for web developers, who now have to check that their pages look right on four independently built engines: WebKit, Blink, Microsoft's Trident, and Mozilla's Gecko. But developers and browser makers alike say cross-browser development is actually less painful than it's ever been, thanks to efforts by browser providers to keep the tools as functionally compatible and compliant with published standards as possible.

"I think that browser compatibility is actually way ahead of what it used to be," says Rey Bango. "If you look at the most modern versions of browsers, things are coming off really nicely." Bango is now the director of developer relations at Telerik, which builds tools for cross-platform app development. Before that, he worked at both Microsoft and Mozilla and was a member of product team for jQuery, an open source library that simplifies cross-browser JavaScript coding.

Far from creating silos or havoc, this move by Google shows how "competition" in the technology sector can be a much more nuanced concept than usually thought. On the web, businesses, distribution networks, and services operate across so many layers of abstraction that practically no game is zero sum.

"Having more rendering engines offers checks and balances and kind of a sanity check," Bango says. That is, a competitive marketplace keeps individual browser makers from rolling out major features that aren't supported by their rivals, since developers won't make much use of a feature that only works in one browser.

In the early days of the web, that just wasn't so: Rivals at Netscape and Microsoft introduced a slew of incompatible customizations into Navigator and Internet Explorer, leaving the world with "features" like Netscape's support for blinking text and untold numbers of aging but mission-critical web apps that load only on ancient versions of Internet Explorer. To create Blink, Google "forked" WebKit, meaning it created the new engine with WebKit's source code but doesn't intend for the two sets of code to remain compatible.

"We're not really competing on feature differentiation as much any more," says Lars Erik Bolstad, the senior vice president of web technology at Opera, which dropped its custom Presto engine for WebKit last year, then followed Google down the Blink fork. "There's not much focus on competition between the browser vendors on the browser engine side, at least not compared to how it was a few years back, when you had a lot more competition in terms of who would be the first to have the feature, who would be the most feature complete."

Now, says Bolstad, browser makers are more likely to collaborate through groups like the World Wide Web Consortium to standardize new feature sets and push more for performance boosts than unique sets of features, since speedier browsers help the web itself compete with native mobile and desktop apps.

"The last six to eight months, the main focus, the main priority, both at Google and for us has always been on improving performance," says Bolstad.

Better performance was one of the driving factors behind the Blink fork, says Alex Komoroske, a Google product manager on the Chrome team. Moving away from WebKit let Google discard parts of its codebase that weren't actually used by Chrome and re-architect how parts of the browser fit together, he says.

But the Blink team is still committed to keeping the web open and standardized--the browser takes its name from the infamous nonstandard Netscape blinking text effect--and discusses feature releases on a public mailing list where web developers and rival browser makers can chime in.

"The beauty of open source is WebKit can watch what we're doing; we can watch what they're doing," Komoroske says.

Chrome's also moved away from "vendor-prefixed" features, where a new web feature is introduced marked in different browsers with the maker's name, so "-webkit-featureX" and "-moz-featureX" can have different semantics until the feature is fully standardized. That sounds logical in theory, but in practice it leads to more fragmentation and complexity, as developers wanting to use a new feature leave up websites with now-abandoned vendor-prefixed features.

"You had a lot of developing leveraging features that were nonstandard and they were leveraging it through vendor prefixes," says Bango. Now, Chrome lets developers try out experimental new features on their own machines by activating them through a largely hidden menu, effectively keeping them out of the browsers of the general public.

And as browser makers do roll out new features, they and other developers have begun providing what are called polyfills--little snippets of code that emulate the features in browsers that don't natively support them. For example, the Google-developer-heavy Polymer project and Mozilla-led X-Tag both aim to provide cross-browser support for Web Components, an emerging standard for allowing developers to introduce new HTML tags with custom interfaces, such as a "" tag for interactive maps.

"Both Polymer and x-tags build on the emerging Custom Elements standard, which means their components are interoperable by default," according to the Polymer site.

"When you look at what's possible, the fact that you can create these elements that are reusable and you can distribute them, for example, using [the Web package manager] Bower, that's pretty powerful," says Bango.

Doctors aren't big users of social media--their schedules don't leave a lot of time for casual status updates, and a state-by-state patchwork of ethics rules limit what they can post and who they can friend, says Toronto critical care physician Dr. Joshua Landy.

But, Dr. Landy says, they are signing up for a specialized smartphone app called Figure 1, created by a company of the same name he cofounded. It's a specialized photo-sharing hub for physicians and other medical professionals, letting them share photos of medical conditions for teaching and diagnostic purposes while incorporating safeguards to ensure patient privacy and consent.

"From my users' perspectives, the way that privacy gets dealt with in the app is essentially, if you can think of the phrase, the best way to keep a secret is not to have it," Dr. Landy says.

That means doctors are prompted before they share an image to make sure they have proper consent and to delete any identifying features from their photos and captions, like patient faces, tattoos, or potentially unique facts like the dates of a hospital stay. Redacted parts of a picture are actually scrubbed from the image file, not merely obscured, he says.

"After that you submit the images, before they're released for anyone to see, they're reviewed by our privacy moderators, and then once they've been moderated or released, if anyone has any concern about any images, they can be flagged or removed immediately," Dr. Landy says.

Dr. Landy, who as a visiting scholar at Stanford University studied doctors' use of mobile devices, said health care workers were already snapping and sharing pictures of patients' medical conditions with their colleagues before Figure 1 launched about a year ago.

"But what they're not doing is saving those pieces of information: those interesting cases, those teachable moments, that sometimes happen at 4 in the morning when you're alone," he said. "There's no current way, or there hadn't been [before Figure 1], any way to archive these great educational assets."

The company formally markets Figure 1 as an educational tool, not a diagnostic one, and, Dr. Landy says he's seen doctors use the tool to quiz students and residents about medical conditions. About 15% of U.S. medical students use the app, estimates cofounder and CEO Gregory Levey.

Dr. Landy says Figure 1 lets doctors and other medical professionals see a wider range of cases than they might see in ordinary practice.

"There's so much just outside the usual circle of regular diagnoses," he said. "You only learn how to make that diagnosis if you've read a case or seen that case in person."

While the site is open to any licensed medical professional, Dr. Landy says it's definitely not intended for patients looking to self-diagnose.

"People get called out if it seems like someone's trying to hide their own medical condition as a patient's condition that they're trying to help, and I think that makes sense," he says. "People are here in a professional environment, and they want to talk to each other and learn from each other."

But, he says, patients he's spoken to are generally okay with doctors anonymously showing their colleagues photos of their medical conditions through the app.

"Almost every patient I've asked is interested in having the images shared as long as they're being shared in the interest of education and being able to help people who are in similar circumstances," he says. "Obviously if someone says no, that's pretty much the end of the conversation; the first principle of medical ethics is pretty much patient autonomy."

Most 3-D printers are great at turning designs into solid structures, but can only build with one or two types of solid plastic. But a new Kickstarter campaign from Ontario-based Structur3D Printing offers an aftermarket add-on letting common printers work with a wide variety of gel materials to produce everything from silicone-based orthotic shoe inserts to custom cake toppers printed from icing sugar.

The Structur3D team decided early on not to try to jump into the crowded 3-D printer market by producing their own complete unit and to focus instead on expanding the capabilities of existing hardware, says cofounder and R&D director Andrew Finkle.

"One of the first things that we decided, because we are materials engineers, [is that] we wanted to focus on just the system that deposited the material," says Finkle, who's completing a PhD at the University of Waterloo. "Because they all run on very similar electronics, we were able to design an add-on."

That add-on, called the Discov3ry extruder, can be swapped in for the plastic-filament extruders included with printers from companies like MakerBot and Leapfrog. The extruder allows the printers to print with a range of gel-like materials, from meringue to hardware store silicone, loaded in refillable syringes.

"We can print things like silicone and latex and polyurethane that are extremely flexible," says Finkle, explaining those materials can be cheaper than traditional 3-D printing plastic and usable where a more rigid material isn't.

"Imagine sliding a hard plastic orthotic into your shoe," says Finkle, explaining that orthotic makers have previously had to build or print custom rigid molds, then use them to cast the actual shoe inserts from a more suitable material.

And working with edible materials can make experimenting with 3-D printing more fun for kids, says Finkle.

"They can eat their mistakes and have a lot more fun doing it over traditional plastic printing," he says. The company plans to work with an Ontario youth maker group to experiment with printing food and other fun materials like Play-Doh, he says.

Structur3D recently had success printing 3-D images out of Nutella--one of the rewards offered to Kickstarter backers is a custom photo printed from the hazelnut spread onto a graham cracker--and the company found bakers at a local cupcake shop were excited by the printer's confectionary applications, Finkle says.

"You could also print very custom cake toppers for your child's birthday cake or a special company occasion," he says.

Once the first units ship to backers in the fall and winter, Structur3D plans to set up an area on its website where company engineers and customers can swap material ideas and tips.

"We're going to have a very open forum," Finkle says.

The Kickstarter campaign helps to stir up consumer interest while raising money to manufacture the first set of extruders, Finkle says. At the same time, the company plans to talk to potential industrial users about potential manufacturing applications, such as for custom health products, he says. And, they hope to tweak some commonly used gel materials to make them work better with the Discov3ry system.

"Using our materials background we can tune some of these materials to make them better for 3-D printing," says Finkle.

The basic idea behind Bountysource seems easy enough to explain--it's a crowdfunding site for open source software. But when the site first launched about a decade ago, those were still fairly esoteric concepts for potential users and investors. Even the founders, then fresh out of college, had never heard the term "crowdfounding," says cofounder and COO David Rappo. The project died fast.

"It ran for a few months before we realized this wasn't gonna pay our bills, and we needed to move on and get real jobs," says Rappo. But about a year and a half ago, Rappo and CEO Warren Konkel decided it was time to focus full time on Bountysource once again.

"Nowadays, we can say it's a crowdfunding platform for open source software, and people are like, we get it," Rappo says. "The time is right: people not only understand crowdfunding, but they love it."

The company's recently hosted successful and well-publicized funding campaigns for Neovim, a modern update to the venerable Vi used by generations of Unix hackers, and for RVM 2, an enhanced tool for Ruby developers managing libraries of third-party code.

Bountysource helped the RVM 2 team plan and distribute the rewards it offered backers and often helps software developers organize and even write copy for their funding campaigns, says Rappo.

But the other advantage of raising money for software projects with Bountysource, as opposed to a general purpose crowdfunding site like Kickstarter or Indiegogo, is that open source projects can publicly offer "bounties" payable to any developer willing to contribute certain features or quash particular bugs.

"When you come to Bountysource and raise money, you can keep the money in the system, start paying it out to different developers for different versions of things," Konkel says.

Even outside of a major funding campaign, anyone can post a bounty offering to pay for improvements to a favorite open source tool, and other users are able to pledge their own funds until the bounty's high enough that a programmer is willing to take on the task. Then, once the requested feature is implemented to the backers' satisfaction, the developer gets paid by check, PayPal, or Bitcoin.

Bitcoin's proven especially popular with programmers overseas in countries where paying by check or PayPal can be difficult, says Rappo.

"It's absolutely the preferred method of payment for a lot of developers these days, especially international developers," he says.

Projects with successful bounties range from the Linux-based operating system Elementary, which has used Bountysource to commission a variety of interface enhancements, to the D programming language, related to C and C++. And backers range from individuals looking to improve favorite software tools to organizations like Mozilla and the cloud provider DigitalOcean.

Letting backers fund iterative improvements to software projects instead of simply raising large sums of money up front aligns better with how open source projects run, says Konkel.

"Kickstarter and Indiegogo are very much around raising one lump sum of money," he says. "With software development, there's really a long life cycle of software."

Ca7ch founder and CEO Rom Eizenberg started thinking deeply about cameras after the birth of his son a little over two years ago.

"I discovered that new parents tend to be some of the most enthusiastic photographers," he says. "If not the parents, then the grandmothers who are pressuring from far away to take pictures."

It occurred to him that while people are still buying point-and-shoot cameras, they're nowadays more likely to actually be using their cellphone cameras to take and share quick snapshots, instead of taking photos with a traditional point-and-shoot, transferring them to a computer and uploading them to cloud storage or a social network.

So, when Ca7ch designed its Lightbox wearable camera, billed as "a camera fit for our times," it built it as a mobile phone accessory, not as a standalone unit.

"Lightbox works with your smartphone through built-in Bluetooth and Wi-Fi, moving the camera's storage to the cloud and letting you instantly connect with friends and followers," the company explains on the webpage for its successful Kickstarter campaign.

That is, Eizenberg says, the 1.5-inch square camera, which can be attached with a two-part magnetic clip to a shirt, hat, or other piece of clothing, captures video and still images to its internal memory, then automatically syncs them via Bluetooth or a home wireless connection to a companion smartphone app and the company's cloud storage.

"The camera storage is the cloud," says Eizenberg. "As soon as it hits the phone, the app comes up and uploads the media to the cloud and clears the local phone memory." Once the images hit Ca7ch's servers, user settings allow the images to be posted to a social site like Facebook, where they can be made public or shared with a custom group of users.

The goal is to offer a simpler alternative to point-and-shoot cameras and even to older wearable cameras like those from GoPro. "GoPro is really really cool--we really love our GoPros," says Eizenberg, adding that they're harder to mount and wear than the Lightbox.

"Building the Lightbox, there were really two key words we had in mind that kept coming up: simple and fun," he says. And one advantage to raising funds on Kickstarter is that Ca7ch has been able to get feedback from fans even before the first Lightbox cameras are set to ship this fall. The company already boosted the amount of onboard memory in response to customer requests, and Kickstarter backers have even debated features they'd like to see in later models, he says.

"One user says I would love to see a wider view angle," he says. "Other users jumped in and said yeah, but we don't want the lens to become big." The company's also looking into making it easy to mount the camera on remote-controlled aircraft after getting inquiries, he says.

And as developers play with the Lightbox's API, they'll be able to build their own apps to help customers find new uses for the cameras, he says. "One of my friends is really into writing a baby monitor app," says Eizenberg, who's himself been able to see the world through his 2-year-old son's eyes by having the boy wear a Lightbox prototype. "The possibilities are really endless."

When you think about household appliances that it would be nice to control from your smartphone, lightbulbs might not immediately rise to the top on the list. After all, home lighting generally comes down to a couple of basic rules: Turn the lights on when you enter a room and remember to turn them off when you leave.

But smart-bulb maker Ilumi hopes to change that, with Bluetooth-enabled LED lights that let you turn them on and off and configure brightness and even color from an iOS or Android device.

"I think it's a really good time for this type of technology," says Ilumi cofounder Corey Egan. "By throwing in some intelligence and really using the mobile device as a platform, we can make ordinary things extraordinary."

The bulbs plug into ordinary lightbulb sockets, and multiple Ilumi bulbs in a room can form their own mesh network, so as long as any device is within a phone's range, it's possible to control them all. Since the bulbs are equipped with their own flash memory, they can remember settings even if a wall switch gets turned off or a lamp unplugged.

And letting customers use their phones to configure the bulb instead of a custom control panel makes adoption that much easier, Egan says.

"It's actually done through a device the user already has," he says. "They're already used to interacting with it."

The company has received more than $200,000 in funding from crowdsourcing campaigns on Indiegogo and Kickstarter and began shipping bulbs in May, Egan says.

The bulbs can be set to automatically turn on at a certain time and even made to pulse and change color in time to music. And once the lights in a room are all set to the right color, that configuration can be saved and recalled later, Egan says.

And as far as more complex possibilities, Ilumi's providing a software development kit to let programmers build their own apps, he says.

"We've seen a lot of unique things," he says, including researchers looking to use the bulbs for scientific purposes.

"One researcher is doing research in how light can affect folks with autism," Egan says.

And, the company says, users will soon be able to have Ilumi bulbs automatically turn on and off as they move through their homes, meaning energy-conscious Ilumi customers may no longer even have to remember to turn off the lights when they leave the room.

A lot of wearable fitness trackers can help you keep track of how many steps you've taken or how many hours you've slept. The creators of the Stonecrysus aim to go further, using data about how particular sleep or exercise patterns or dietary choices affected your body in the past to offer recommendations to help meet your fitness goals.

"Stonecrysus then learns how specific foods, activities, and sleep affect each user's physiology uniquely," says Stonecrysus CEO Matt Landers. "It collects over 35 data points and analyzes those data points to make those recommendations."

Daily measurements help the device, which can be worn on a lanyard or wristband or carried in a pocket, and a connected smartphone app continually improve their estimates of how different exercises and sleep habits affect individual users, he says.

"Stonecrysus doesn't define a mile run as 125 calories burned," the company wrote on the page for its ongoing Kickstarter fundraiser. "It looks at how running a mile has affected you in the past to determine how it will affect you today, based on your current health and fitness."

Then, the app can make predictions and recommendations for changes based on specific goals, like weight loss or muscle gain and continually refine its estimates over time, he says.

"If your weight is even slightly different, it's going to relearn about how an apple and three miles [ran] affect you and what your metabolism is," he says. "Even if it corrects only slightly, it's still gonna correct."

The wearable component can distinguish between different motions related to different exercises, and the app includes a wide range of graphical menu options to let users specify down to the slice exactly what foods their meals included, not just nutrition label calorie counts, Landers says.

Users will also have the option to share their information with fitness advisors, such as doctors or personal trainers, through a secure cloud system.

"We created this online dashboard where all this data gets uploaded, and the user can give access to their health professional to access this data," says Landers, who worked with his cardiologist father to develop the product.

Landers says Stonecrysus lawyers say the app doesn't have to comply with strict HIPAA health privacy requirements, but the company plans to do so anyway.

"We've done an extensive amount of research into how to best protect this data, and to whether we need to meet certain regulations," he says. "The data privacy is of the utmost importance to us."

The company will also let users optionally submit their data for anonymous research into how certain foods and sleep and exercise patterns affect different demographics, he says.

"We're going to look at how lifestyle habits affect specific health events," from weight loss to various illnesses, he says.

Everyone's seen automatic lawn sprinklers watering the lawn on a rainy day. It's funny, as long as you're not the one paying the water bill.

The trouble is that most sprinkler systems just run basic timers, so unless someone's home to override the default settings, they'll happily pump out water in the middle of a storm. And even for homeowners who keep a careful eye on the weather report, sprinkler controllers can be difficult to configure, and it's often hard to calculate exactly how much water a section of lawn needs on a given day, says Skydrop cofounder Clark Endrizzi.

Skydrop makes Wi-Fi-enabled smart sprinkler controllers that automatically pull in online weather report data to help them decide when to turn on and when to stay off, conserving water.

"My frustration just originated with my own experience with my own sprinkler controller," says Endrizzi. "I pay for my water like a lot of people, and so it's an expensive thing that I have to pay monthly for, and I went and talked to three different contractors wondering how should I be watering my lawn and how can I do it most efficiently, and I got three different answers."

Skydrop's configuration panels, which the company says will begin shipping in mid-August, couple live, zip code-level data about precipitation and temperature with user-provided information about sprinklers and soil to keep grass properly watered and not over-saturated, he says.

"Where the real savings come through is on that day-by-day, season-by-season basis," he says. "Utah, where we are based, is a very dry state. In May, we just happened to have this crazy spell of rain, so my Skydrop didn't have to water for three weeks."

And if users want to override Skydrop's recommendations for any reason, they can set custom schedules either directly on the control panel or configure their sprinklers remotely through a mobile or desktop browser, he says.

"We give our users as much or as little control as they want," he says. "If you happen to be seeding or something like that, it needs to be watered four or five times a day, that's where you might set up a custom schedule or something like that."

Unlike some Internet-based smart sprinkler systems like those from Droplet, Skydrop's systems are designed as drop-in replacements for existing sprinkler controllers. And, since the controllers pull weather information from Skydrop's cloud servers, not from clouds in the sky, there's no need to install weather monitoring equipment, according to the company website.

Endrizzi says it's an exciting time for garden automation in general, with more digital tools coming on the market to tell homeowners when to put down seeds and when to fertilize their plants.

"This is a huge area where it's ripe for innovation," he says. "You look at places like California where it's like 95% dry. There's some major things that can be solved in this space in terms of water use and water conservation."

As an avid weightlighter, Dr. Eric Hensen knows how annoying earbuds can be at the gym. Hensen, who is also an ear, nose, and throat doctor, watched as the tiny headphones slipped out of sweaty ears and got tangled in machines. For athletes and gym buffs, traditional earbuds just aren't cutting it. So Hensen decided to reinvent them.

"There's a guy who I work out with--he does bent-over rows, and he always puts the wire in his mouth while he lifts," says Dr. Hensen, who is the founder and chief medical officer of wireless earphone maker FreeWavz. "Inevitably, he'll bite through the wires once a week."

Hensen says he wanted to design a safer and more comfortable earphone for athletes that would fit snugly and comfortably in the ear and work without tangling wires. He took a look at current listening device design and ultimately came up with a prototype for the FreeWavz headphones, which connect to an audio source via Bluetooth and are designed to stay in the ear even through rigorous exercise.

"I took a set of these really old hearing aids, and I kind of retrofitted them to our idea," he says. "From there we actually drilled [them] out, made our own housing, put all the components that we wanted in it."

Cutting The Fat: How The Product Evolved

In an early iteration, the earphones included a built-in MP3 player, but FreeWavz soon heard from potential customers that they'd prefer to listen to music they already stored or streamed of their phones. FreeWavz can pick up audio signals from a phone up to about 33 feet away through Bluetooth networking, he says.

Bluetooth technology's being integrated into a number of wireless headphone designs lately, from the tiny Earin earbuds to other exercise-focused tools like the Dash earbuds. Dr. Hensen says FreeWavz, in particular, were developed to include all the exercise-centric features the company could pack into its ear-fitting form.

"The MP3 player got dropped because none of our customers wanted it," he says. "That allowed us to incorporate even more stuff that we wanted."

Taking out the MP3 player and storage left room to add more fitness-oriented components to the earphones, including an accelerometer to track motion and a pulse oximeter to track heart rate.

Smart Earbuds For The Self-Tracking Generation

Through a connected smartphone app, users can program the earphones to automatically announce stats like distance and heart rate throughout their workout or when they cross certain thresholds.

"There's a lot of power in getting information and having it fed to you audibly without any visual distractions or safety issues," says FreeWavz president Harry Ericson. A software development kit will ultimately let other developers deliver audible notifications through the headphones, too.

The earphones also include support for hands-free phone interaction and calling, and each earpiece can be configured independently to blend outside noise with audio content in a particular ratio.

"There are two microphones--one you talk into and one that's on the earbud that allows you to pick up sound in front of you, behind you, or to the side of you, depending on how you program it," Ericson says.

That's of particular importance to cyclists, who want to make sure they hear traffic sounds while listening to music, says Dr. Hensen.

"When you're going with traffic, you can program the left side to pick up as much road sound as you want," he says.

Users can also independently configure volume and equalizer settings on each earphone, which Dr. Hensen says will be useful to anyone with different hearing levels in different ears. And, they can save multiple profiles for different exercises, so they can tune out outside noise and skip updates on distance traveled while lifting weights but still hear traffic sounds and mileage numbers while riding a bike.

FreeWavz is in the midst of a Kickstarter campaign and hopes to ship the first earphones by October, says Ericson.

"We've got the core functionality done and it's really now moving toward production," he says.

Businesses can gather more and more real-time data on everything from sales to web traffic to social media buzz, but actually accessing those stats often requires individually logging in to lots of different analytics sites.

That's something a lot of busy company leaders just don't have time for, especially when they're traveling and only have Internet access on a smartphone, says Davorin Gabrovec, CEO of Databox.

"They have Salesforce, they have other apps, but they don't have those apps installed on mobile," he says.

To make real-time data from providers like Google Analytics, Salesforce, and Zendesk actually useful in real time, Databox aggregates selected company stats from across the web and creates simple, customizable charts and dashboards all within one mobile app.

"Imagine it like the first app you will open in the morning, and you will see what's really going on with the business," says Gabrovec.

While doing market research, Databox found that many executives only received an emailed spreadsheet of company data once or twice a day, while the live numbers were effectively only available to analytics staff able to devote a desktop browser tab to each analytics provider.

"Email was still the only way they accessed their business data," he says. "We started to see there is a huge need for the whole enterprise for people who spent huge amounts of time outside the office to bring them key insights and metrics in a way more appropriate than opening Excel files."

For common data sources, users can provide their login credentials to the app and select from some of the most commonly accessed metrics, Gabrovec says.

"We asked, first, for example in MailChimp, what are the top 10 metrics that we care about as a MailChimp user," says Gabrovec. "A user can change those metrics and personalize based on his or her needs."

Different types of users naturally pull from different data feeds, says Gabrovec: Developers often look at info from GitHub and Trello, while sales and marketing staff look at data from sites like Salesforce and Oracle's Eloqua, he says.

Users can also set up a daily scorecard that displays a morning summary of the data they're most interested in, says Gabrovec.

"You will receive a push notification before you even started your breakfast saying these are five things that happened yesterday that deserve your attention," he says.

And push notifications can also be set up to automatically notify users when certain stats cross designated thresholds, he says.

An SDK also lets users add custom data sources, and display customized charts and scorecards, Gabrovec says.

"When we expose that part, real users feel the whole potential of really plugging in different data sources presented in a way they would like to see," he says.

The company's in the midst of rolling out an Android app, which should be available in the next few weeks, and tweaking the layout options for that and the existing iPhone app, he says, in the hopes of making company data simple to use and analyze.

"Everyone is talking about big data, but data itself is not the problem--we have huge amounts of data already," he says. "But how to deliver that data to someone who needs to make big decisions, that is what we call the last mile in data work."