Machine Learning is Changing Developer Nation and the World

Breakthroughs and advancement in machine learning (ML) models, techniques, frameworks, and applications are having a growing influence on the developer ecosystem. Machine learning is enabling new experiences that allow computers to tackle more complex tasks that once only humans were able to. In the 14th edition of SlashData’s Developer Economics survey, the influence of machine learning can be observed throughout the collected data and analysis. Both the growth of new platforms and experiences as well changing infrastructure and programming languages are being driven by innovations in machine learning.

The State of the Developer Nation report 14th Edition summarizes some of the key findings from the survey and provides a snapshot of today’s developer ecosystem. Underlying these findings is the importance of machine learning is in driving future trends. As the capabilities of machines and humans begin to converge, implications extend well beyond the developer community. To be well positioned for future technology shifts, it is important to understand where and how machine learning is influencing the direction of the software industry.

ML is revolutionizing how we get around

52% percent of developers believe that advances in self-driving cars will have the most impact in the next five years. Machine learning is at the core of advancements in this area. The pivotal role that machine learning is playing in these projects is an indication of the power such advancements have in changing how people live their lives. Machine learning not only teaches cars to drive themselves but supports computer vision that can identify objects such as stop signs and pedestrians. The tremendous economics of this segment is attracting boatloads of capital that is leading to new advancements and new opportunities.

Machine Learning impacts AR

The growth of augmented reality is another area where ML will have an important impact. Deep learning is improving the simulation, localization, and mapping (SLAM) capabilities of leading AR platforms. SLAM enables AR platforms to identify objects and to overlay augmentations. It also recognizes and tracks features within a scene. These advancements are directly impacting the 15% of the developer community who are working on AR projects by supporting more advanced tools to create more fluid experiences. This takes the AR space beyond the Snapchat dogface mask. Given that mobile is the most popular platform for AR developers, 53% are targeting Android and 37% are targeting iOS, the developers in this space are also seeing a significant impact.

Image classification models are the #1 project that machine learning developers are working on so we expect continued advancement in this space to support driverless cars and AR. 22% percent of ML developers were working on image classification and object recognition. Another area where machine learning developers are working is conversational interfaces or natural language processing (NLP). 20% of ML developers were working on NLP/chatbots, ranking third in our survey of ML developers. Chatbots are already all the rage and NLP models have made significant advances. The next challenge is creating even more sophisticated models to make chatbots smarter.

ML drives Python and Serverless growth

While the work that machine learning developers are doing to create new experiences is having a profound effect on what developers can create, ML is also having a big influence on infrastructure and programming languages. Python rose to the third most popular language in our latest survey reaching 6.3 million developers behind JavaScript (9.7 million developers) and Java (7.3 million developers). Python supports many ML libraries and is easy to prototype and experiment with making it very popular with machine learning developers. The growth of serverless architectures is also being fueled by new machine learning models. While the development of models requires dedicated compute resources, serverless architectures can make implementing these models much easier. Not only can models easily be executed closest to the application but models can be tied together via functions that can span different languages and platforms, making applications even faster and smarter. Today the vast majority of workloads handled by serverless are web and mobile API calls but developers plan on using serverless for machine learning and conversational experiences more in the future.

The advancement and impact of ML is no news to developers. In our survey, 37% of developers believed that advancements in ML models would have the greatest impact over the next five years. Specifically, models that won’t require large training datasets, for example, using transfer learning or capsule networks. With less reliance on huge datasets, barriers to the exposition of new machine learning models are lowered and developers can create more models and smarter applications.

As the prevalence of machine learning grows, developers will need new skills that go beyond coding and computer science but incorporate, advanced mathematics, probability, statistics, and data modeling. Developers at the top of the food chain will be able to bring together skills, knowledge, and understanding from all these areas and apply them to next generation of problems.


Explore the data pointing to the influence of ML as well as data and analysis around additional developer trends by downloading the State of the Developer Nation 14th Edition report for more data and graphs depicting top developer trends.

Haptics and Sensors: The new toolset for handset differentiation?

[Haptics, sensors, gesture tracking, intelligent texting and pico projectors; a taste of the technology soup headed our way. Guest author Peter Crocker discusses how sensor technologies offer handset differentiation, and the challenges ahead for OEMs.]

Haptics and sensors - the new toolset for handset differentiation

Innovation is the name of the game for handset manufacturers. Not just for Apple who keeps expanding the envelope of hardware and UI capabilities, but all major OEMs who are looking to differentiate beyond software. Android and Windows Phone are now providing an end-to-end device recipe for device makers, from hardware to a developer ecosystem. As such, handset OEMs (Nokia, Samsung, LG, Motorola and Sony Ericsson) are finding themselves on the same playing field as PC assemblers (Acer, Dell, plus the likes of Huawei, ZTE and Visio). In the post-Android era, not only is the playing field leveling, but it’s also becoming more crowded. More importantly, unless handset OEMs can find ways to differentiate they’ll have to default to competing on price, which is exactly what they want to avoid; the OEM cost structure is not designed to withstand razor-thin margins.

[poll id=”11″]

One way to differentiate is with phone features – not GHz figures, but the type that would have a major impact to the user experience. Many OEMs would crave to break into the market with innovations such as what the Palm Graffiti handwriting recognition was at its time.

Feature innovation comes today in many forms, as manufacturers try to evolve smartphones into smarter phones; haptics, predictive texting, gesture recognition technology, inertia sensors, digital compasses, and the emergence of pico projectors to name a few.

A taste of feature innovation
– Next Generation Haptics: Haptics, or the process of using motion or vibrations to create tactile feedback on a users hand or finger, has been around for quite a while, with solutions available from Immersion and Synaptics. As an alternative to using touch-sensitive screens, companies like eyeSight and GestureTek are using the built-in phone camera to analyse hand motions and recognize gestures.

– Inertia and direction sensors: handset makers are following Apple’s lead with the integration of accelerometers, digital compasses and gyroscopes into the phone. These sensors can be leveraged to support for example improved location through dead reckoning and gesture recognition. Gyroscopes and compasses are also providing precise data on the location of a device in the three dimensional plane opening the door to augmented reality applications. Companies such as Layar and Wikitude are helping developers walk through that door with AR software platforms.

– Predictive Text & Gesture Tracking: Predictive texting has seen limited innovation beyond plain-old T9; as such a range of vendors have emerged to provide significant improvements in prediction and correction accuracy, namely Keypoint, EXB, TouchType, Cootek, Keisense (now Nuance) and BlindType (acquired by Google). New forms of predictive texting combined with gesture recognition technology such as as Swype and ShapeWriter (acquired by Nuance) is enabling quicker text input on a touchscreen – for example tracking the movement of a finger on a touch screen, a phone moving in space with inertia sensors, or tracking hand movements with infrared technology hand gestures.

– Pico Projectors: While the integration of pico projectors, or mini video projectors, into mainstream phones is still a ways off, the technology from the likes of TI and Micorvision claims to overcome one of the biggest UI challenges of mobile device, small screens.

What’s more, combining such features can yield more than the sum of the parts. For example, gesture recognition technology combined with haptics could allow users to effectively navigate applications. Similarly, the combination of pico projectors, gesture recognition and image tracking technology could eventually enable interfaces that will resemble Sci-Fi movies.

Integration challenges
As easy as it may sound, innovative features are not just about shopping components off the shelf. Cost is an important consideration, especially for technologies that require specialized components that do not enjoy economies of scale. For example, the green laser required in a pico projector represents one third of the cost of the entire system due to the fact that the part has no use beyond a pico projector.

Integrating new technologies into handsets is a further challenge for handset designers. Digital compasses are sensitive to electronic interference and need to be carefully positioned within the phone to avoid interacting with neighboring electronics. The design of haptics mechanisms also presents many problems. In a typical haptics system design, touch screens float in their frames and are held in place by flexible materials that allow the screen to vibrate creating haptics effects. These designs can fail letting dust inside the device or the screen can separate from the frame if the device is dropped. OEM’s are still learning how to effectively incorporate such features into their designs.

A number of start-ups are working on overcoming these barriers in addition to creating new capabilities. Senseg in Helsinki is eliminating the need for moving parts in haptics systems and has created a system that it claims can pinpoint tactile feedback. InvenSense has brought to market a motions sensing MEMS chip that integrates a gyroscope and accelerometer in one chip, making it easier for OEMs to integrate and reduce cost. Light Blue Optics has developed a pico projector that creates a holographic image and infrared sensors to turn any surface into a virtual touch screen. The company also just raised $13 million to shrink the technology.

Innovation of course requires risk-taking. OEMs are finding themselves in a chicken and egg scenario; design cutting edge features first, or wait for the apps to leverage the features? Samsung and HTC seem to be comfortable taking such risks. Samsung was the first to introduce a phone with an integrated pico projector in 2009 and the Galaxy S sports a gyroscope, Swipe technology and an Augmented Reality browser. HTC is also pushing the envelope having developed and launched devices with home grown haptics.

Undoubtedly users will be the biggest winners as OEMs battle to wow new customers. A close second will be application developers who will stretch their imagination to build new applications and businesses around emerging features.  While these opportunities are compelling, progress will not happen overnight. Gyroscopes are still only available in high end smartphones and next generation haptics will only appear in niche devices next year. If you’re interested in building an app for a pico projector, you may be waiting a few more years.

The question is: is this new roster of sensor technologies going to allow OEMs for once to out innovate Apple?


[Peter Crocker is the founder and principal analyst at Smith’s Point Analytics (, a full service market research company helping innovators in the mobile and wireless market better understand emerging opportunities. Peter has been in the mobile and wireless industry since 2003 and holds an MBA from the College of William and Mary. Peter can be reached at]