Python and Artificial Intelligence – Growing Together
Technologies come about and evolve based on needs and opportunities that arise and often they are related to other technologies. As innovation sparks change, that change creates new obstacles and thus opportunities for other technological evolutions. We saw the mobile revolution set the stage for Cloud and now the Cloud is facilitating IoT. A similar, though less noticed, tech affinity is between Artificial Intelligence (AI) or Machine Learning (ML) and the Python scripting language.
Scripting languages existed in some form for nearly 30 years before web development spurred further growth adoption. Scripting language users have remained relatively steady as a proportion of the developer population, growing from nearly 57% (7.9M) in 2007, to 59% of the population (14.2M) in 2019. Among scripting languages, Python is now one of the fastest growing and that’s largely due to it’s ties to data, analytics and AI/ML.
The chart here shows the number of developers worldwide who are working on AI and ML algorithms as well as the much larger number who are using AI and ML in their apps, going back to 2013. Notice the change in the pattern that begins in about 2016 when the number of developers working with AI begins to increase dramatically. This same change is noticeable at about that time in a chart of the adoption of Python.
Most developers use multiple languages, so to measure language use we ask developers to tell us what percent of the time they use a particular language. This chart shows the percent of developers who use scripting languages who use Python any percent of the time. This chart compares three of the most popular scripting languages and, as we can see, while Javascript is the most used, Python is currently the fastest growing.
But Python is not just growing overall with casual users, the number of developers who use Python as their primary language; i.e., more than 50% of the time, has more than doubled since 2016. Python’s use for reporting, scientific, and mathematical purposes sets the stage for Python’s numerous libraries, frameworks, and tools with data science and machine learning. Notably, Theano is one example of where Python intersects not only with machine learning, but also with math library acceleration for CPU and GPGPU. More recently, PyTorch’s release (2016) and Tensorflow’s adoption of Keras (2017) are other AI-related changes that have created opportunities for developers to leverage Python. For purposes of AI and machine learning development, Intel has even included Python in its optimization technologies, including Intel Distribution for Python, which is supported by Intel Parallel Studio, and a Python-specific distribution of Intel’s Data Analytics Acceleration Library.
So, as more developers create and integrate AI and ML capabilities into their apps, we can expect to see more and more use of Python.
Data Source:
Global Development Survey, © Evans Data Corp 2021
Worldwide Developer Population and Demographics Study, © Evans Data Corp, 2020
Demystifying Developer Personas
On first look, software developers appear to be a more distinct and well-defined market segment than most others, which should make it easier to market to them. It’s always simpler to create messaging and positioning for a target market that is uniform and cohesive than it is for a very diverse market. And so developers give the appearance of being easy to market to; but is it that cut and dry?
There are important primary characteristics that set developers apart from the general population. They write programs professionally, and as a result a very particular type of mental acuity and skill set, is more likely to be found among this group than the general population. The very act of programming requires certain characteristics. The successful developer is logical, has a keen eye for detail, and responds to mental challenges with a kind of inquisitiveness that can be associated with analytical and creative mindsets. Developers are usually more cerebral, curious and way more literal than others.
However, while there is a measure of homogeneity amongst developers that can aid marketing professionals who are trying to reach and persuade them, there is also a level of divergence from the general population that makes developer focused marketing unique. Developers frequent and place confidence in different media than the general public; they appreciate different forms of touch, and different elements in messages are more likely to resonate with them. In addition, there is not just one form of development and the types of development this group of people engage in can be so diverse that reaching out to them requires a special understanding of what they do, in addition to an understanding of who they are and what media they trust.
You can do research to find out the specifics of today’s developer and we do. We’ve been studying developers for well over 20 years. We can give you lots of data that can aid in your development of a marketing campaign and strategy. Our annual Developer Marketing survey explores in great detail the various elements of subsets that make up the developer population – and since we’ve been doing it since 1998, we’ve learned a lot about developers and the various personas they exhibit. Some characteristics are shared across personas while others are unique to the type of developer you may be targeting.
For example, developers all answer to a variety of titles in their jobs, the most common being programmer, development manager, or project lead, though titles vary considerably by company size. They are overwhelmingly male. Although the female contingent is growing, males still comprise at least three out of every four developers almost anywhere in the world, though there are many more women developers in Asia. – the ratio varies according to geography, but both men and women developers think there should be more women involved.
Their median age is now 38 in most places in the world. They tend to be married, and to have one or two children. The typical developer has between three and 10 years of experience, and has a high-level academic degree – a bachelor’s degree or higher – though there many developers who continue to learn on the job in order to keep up with the ever changing technology.
These are valuable fundamentals on which to build a strategy, but you still need the insights that only experience in marketing to developers can bring, and that experience suggests you also need to define distinct personas within your target market.
Personas, including developer personas, are not well defined or rigid. How you define your personas should vary with the type of product you have, and the type of work and work environments you expect your users to be engaged with.
There are basically two ways to study developer personas: pre-defined and organic. Often marketers will define their own personas and then do research to profile them. For example, a marketing professional may have a product line that consists of an entry level development tool – maybe a text editor with color coded highlighting, an advanced IDE with a full suite of professional tools, and an RTOS. Obviously there are three different target customers and there may be subgroups within each of those. This observation lends itself to the creation of customer subsets that are expected within each group. However, in order to transform these subsets into personas you need to do research to profile them. You want to know basic demographics of each subset, but also factors such as language and technology use, where and how best they can be contacted and influenced, plus attitudes and intentions. You need to know the similarities and differences and only then can you flesh out your personas enough to make decisions based on them.
The other way to form personas is organically. Forming personas organically means not coming with a preconceived expectation of who your target subsegments or personas are. Instead a survey is done exploring all the relevant profiling factors, but cluster analysis is used to find the clusters that occur naturally within your target market. In essence you are letting your target market shake itself out naturally to reveal the personas that form distinct groups.
However you go about creating personas, marketers know that they provide an excellent foundation on which to base targeted messages and positioning. We do lots of persona work for some of the largest companies in the world and are happy to help with this. Let Evans Data apply the expertise earned over 20 years to help with this and other research needs.
Developers are Embracing Quantum Computing
Quantum computing has been the subject of speculation and anticipation since quantum algorithms and experiments into quantum computations first appeared in the 1960s. However, it hasn’t been until recently that quantum chipsets or quantum computers have been available to software developers. This spring Evans Data Corp explored this exciting field in our biannual Global Development survey.
Whereas classic computing deals with bits in 1s and 0s, quantum chips provide qubits, which can exist in 1 and 0, as well as a superposition of both states. Due to this, coding for quantum computing needs to take into account this superposition, and there are issues with that. Consider Schrodinger’s famous cat. The cat is in a sealed box with a vial of poison that will kill the cat if it’s shattered. Quantum mechanics implies that the cat is in a superposition and is thus simultaneously alive and dead until someone opens the box and observes the cat as being either dead or alive, which in turn implies that when the quantum superposition ends, as it does with observation, reality collapses into just one possibility. The fact of observation, or measurement, compels this change.
Quantum computing also represents a new form of hardware, and leads to the use of languages that are more similar to hardware description languages (HDLs) in how they manipulate quantum circuits. Also, due to the specialized nature of quantum computers and the apparatuses needed to cool quantum circuits enough to ensure stability, quantum computers continue to be rare and (up until recently) inaccessible.
However, some vendors now provide developers with access to quantum computers via the Cloud. This has helped spur a vigorous interest in quantum computing.
In our latest Global Development survey, conducted in the late spring, we look at how familiar software developers are with quantum computing, what their intentions are toward the technology, how they view the nascent landscape and what issues they foresee.
For one thing, two thirds say they are already familiar with the concepts of quantum computing, even though far fewer have tried any of the cloud services or simulators that major players in the market are now making available. Of those who have explored those services, Microsoft, IBM and Google are the vendors most likely to have been visited, but the relative popularity of each varies considerably by region. There’s other differences as well, such as in their roles, what they feel is needed to bring quantum into their own development efforts, and what issues they see, but one thing that’s very strong across all regions is the belief that one day quantum computing will eventually take over the mainstream computing scene.
Although it’s a science in it’s infancy – the future looks very bright indeed.
AI and Heterogeneous Systems
Artificial intelligence and machine learning are activities that are highly compute intensive. This means that the hardware that developers use to implement AI systems is very important for the success of the project. This is true of new projects as well and the retrofit of existing legacy applications. When thinking about the hardware requirements for AI projects, considerations about CPU use vs GPU use are always surfacing, as we see in our latest Artificial Intelligence and Machine Learning Survey. In this survey of over 500 professional developers actively working with AI and ML and conducted in May 2019, we explored the implications and usage of both CPUs and GPUs when working with AI systems.
Many systems require the use of at least one CPU and one GPU, and sometimes other forms of accelerators. While these architectures are two different types of integrated circuits, they aren’t necessarily part of a heterogeneous hardware approach. Heterogeneous hardware is the inclusion of multiple types of architectures on the same memory bus that can run instructions in parallel and share from the same pool of memory. The proximity and sharing of resources reduces the memory needed for scheduling, but also requires particular attention to how memory is shared.
Measuring and Forecasting Developer Populations
Some market data is current and measured. Some looks to the future and is forecast. While very different from each other, each has its own value and reliability.
For example, Evans Data’s recently released Global Developer Population and Demographics Report , shows a total worldwide population of approximately 23 million developers. This is forecast to grow to 27.7 million by 2023 which is based on a growth rate that varies slightly by year but turns in an average of 3.86%. Growth projections are based on a very long and solid regression analysis going back to 2006 and, barring a huge economic event of some kind, they are quite reliable.
In the same study we find that 7.3 million developers are using Clouds for their development environment (either public, private or hybrid), meaning they write their code, debug, build etc in the Cloud itself. Cloud vendors of course encourage this and often supply tools and incentives for developers to work in their Cloud environments. These efforts have been very successful and we find that an additional 4.9 million developers report that they EXPECT to be developing applications in the Cloud within the next 6 months. That’s a highly significant growth rate and if that were achieved and repeated every six months for 1.5 years the Cloud would be the ONLY environment that developers would use in just two years.
But there’s a big caveat. Developers are wonderful beings, but they’re not prescient. And for this question they are using their intuition rather than any type of data analysis. They have been predicting a similar adoption rate for Cloud development for several years; and though the Cloud as a development platform has definitely grown, it hasn’t grown to that extent in any six month period.
That’s why we think of the future intentions that developers tell us as just that – intentions. But make no mistake, intentions are important. They not only tell you what the person thinks he’s going to be doing, but in most instances they imply what he wants to be doing or what technology he thinks will be successful. This glimpse into the psyche is essential for product strategy, as well as marketing, messaging and positioning. Current attitudes and perceptions highly influence actual future adoption and can they can be an invaluable resource to inform a product strategy for great results.
For more information see:
How Fast do Developers Move Along?
There’s a tendency in our industry to assume that software developers immediately migrate to the newest version of any tool or technology as soon as it’s announced, and there’s a bit of truth in that. Certainly many developers are passionate about technology and adopt “bleeding edge” innovations even as they’re in beta before production release. But while developers do tend to become excited about technology evolutions and love to try them, there are actually a lot more considerations to be taken into account when abandoning a tool, platform or technology than when adopting a new one.
Embracing the newest tech is initially easy because there’s no background or canvas to be worked into the project unless and until the project has to integrate with existing technology. But giving up older versions can cause a realm of trouble because newer versions may not support the same functions as previous versions or may not be compatible with software created with and/or around the older technologies.
In our newly released Global Development Survey 2018 -2 report we found that while just over 53% of developers will continue to use a tool or other technology no longer than 6 months after it has been discontinued, another 29% will continue to use it for up to 2 years after discontinuation, 8% will use it up to 5 years, and 10% will use the discontinued tool or technology indefinitely – basically never abandoning the old technology.
This survey was conducted last month in 6 different languages worldwide so we can see some regional variations in the length of time a developer will keep using an obsolete technology. Latin American developers, in particular tend to want only the latest and greatest, and this may be because this emerging region doesn’t have the historical legacy baggage that slows down other regions. On the other hand, the EMEA region is most likely to be where developers will never give up their old tools or technologies. APAC region also has lots of early adopters, and North America is middle of the pack.
So, just because a vendor discontinues a product, don’t think usage will dry up overnight – it could take a long time to move all developers to a new version.
Developer Outreach and Marketing
On first look, software developers appear to be a more distinct and well-defined market segment than most others, which should make it easier to market to them. It’s always simpler to create messaging and positioning for a target market that is uniform and cohesive than it is for a very diverse market. And so developers give the appearance of being easy to market to.
After all, there are important primary characteristics that set developers apart from the general population. They write programs professionally, and as a result a very particular type of mental acuity and skill set is more likely to be found among this group than the general population. The very act of programming requires certain characteristics. The successful developer is logical, has a keen eye for detail, and responds to mental challenges with a kind of inquisitiveness that can be associated with analytical and creative mindsets. Developers are usually more cerebral, curious and way more literal than others.
However, while there is a measure of homogeneity amongst developers that can aid marketing professionals who are trying to reach and persuade them, there is also a level of divergence from the general population that makes developer focused marketing unique. Developers frequent and place confidence in different media than the general public; they appreciate different forms of touch, and different elements in messages are more likely to resonate with them. In addition, there is not just one form of development and the types of development this group of people engage in can be so diverse that reaching out to them requires a special understanding of what they do, in addition to an understanding of who they are and what media they trust.
You can do research to find out the specifics of today’s developer and we do. We can tell you lots of data that can aid in your development of a marketing campaign and strategy. For example, developers answer to a variety of titles in their jobs, the most common being programmer, development manager, or project lead, though titles vary considerably by company size. They are overwhelmingly male. Although the female contingent is growing, males still comprise at least three out of every four developers – the ratio varies according to geography, but both mean and women developers think there should be more women involved.
Their median age is 36 in most places in the world. They tend to be married, and to have one or two children. The typical developer has between three and 10 years of experience, and has a high-level academic degree — a bachelor’s degree or higher — though there many developers who continue to learn on the job in order to keep up with the ever changing technology.
These are valuable fundamentals on which to build a strategy, but you still need the insights that only experience in marketing to developers can bring.
Providing that insight and understanding for marketing success is what motivates us at Evans Data to host our annual Developer Marketing Summit. This year it’s on September 17 and 18 in San Jose. We’ve got two full days filled with insights, networking and knowledge headed by the top developer marketing professionals from virtually all of the major players in the industry. Intel, Amazon, Microsoft, Facebook, Salesforce, and many more will provide powerful insights into how to successfully reach and motivate software developers. Don’t miss this very important event. https://devmarketing.evansdata.com/
Augmented Reality Use Cases
Imagine going into a store, tacking your steps on a visual display map on your phone and seeing bright red stars over everything you’ve bought in the store over two weeks ago and thus might need again. Or strolling through Paris and pointing your phone at a building and getting a visual history of it from the 17th century to the present, then turning around and pointing at an insect on a flower and getting the names of both and maybe some animation depicting how they pollinate.
These are commonly dreamed up implementations of augmented reality (AR) for consumers. They’re fun, interesting and easy to understand, but they’re probably not the first area where AR is going to flourish and come into its own.
According to our recent Evans Data Global Development survey, software developers who are actively creating AR applications today are most commonly targeting industrial design for their apps. Almost a quarter of the world’s software developers claim to be kicking the AR tires, trying out AR in projects, experimenting, or actually working on functional apps. Of those, 17%, the largest group, are targeting industrial design.
The survey, conducted in over 100 countries and in six different languages, showed that augmented reality functionality within apps is being adopted particularly strongly in emerging market regions in APAC and Latin America, but it was the EMEA region where the strength of industrial design was most clear.
But what does that mean? Industrial design today is done with CAD programs, drawings, applied mathematics, electrical and mechanical engineering tools, models, and so on. For each widget there is a lot of time and resources required as well as diverse steps and tools. But imagine AR applied to industrial design.
This would mean a product design environment where your design exists as a 3D object, projected virtually in the space before you – sort of like a hologram. This 3D image is constantly fed information from the physical world as well as the virtual one, so your virtual product behaves just as it would if it actually existed. It would simplify and speed product design as we’ve never known it. New and better products would be able to come to market in a fraction of the time.
Now, it should be pointed out that the number of developers targeting industrial design is a very small plurality. Commerce and engineering are next, followed by numerous other use cases that get single digit responses. So this new emerging market is still fractured and forming, but it’s exciting to think about all the ways AR will be able to change our lives in a million different ways – through healthcare advances, retail, travel, entertainment, you name it!
Are Hackathons Losing their Luster?
For the last several years, companies who want to attract software developers have been betting big on hackathons. Set up a little coding event in a school or clubhouse – or better yet your own facility if you have something that can be used – publicize it, buy some pizza and sodas as motivational refreshments, and just sit back and wait for the developers to stream through the doors, start coding and become so enamored with your company and products that they’ll not only thoroughly adopt your APIs or platforms but they’ll also tell all their friends to do that too. Right?
Well, maybe or maybe not.
The world has become totally interconnected which means that more and more companies outside of the usual realm of platform and tool companies are aware of the need to connect with developers. To connect with those developers companies are opening up APIs. Open APIs are proliferating everywhere and if you’ve got an API you’ve got a platform, and if you’ve got a platform you need developers to adopt and support that platform.
We’ve seen numerous companies trying to adapt to this new reality by putting together a small team and charging them with populating their platform. For many that has meant putting on a hackathon, or maybe a whole series of them and that’s about it. As Evans Data continued to measure developer sentiment, actions and technology adoption we watched the number of hackathons that developers attend rise each year and saw our clients devoting quite a lot of resources to them.
But now the tide seems to be turning. The number of developers who attended a hackathon declined significantly in 2017 according to the newly released Evans Data Developer Marketing 2018 survey report. The survey showed that 70% of developers in 2017 had gone to at least one hackathon, which is a significant 19% decrease from the 83% that went to at least one in 2016. At the same time, those who went to only a small number of hackathons increased while those attending many hackathons showed a marked decrease.
And it wasn’t just vendors who were putting these on. Of those who attended hackathons, 46% said they went to one sponsored by a commercial vendor while 42% attended one sponsored by a local developer organization, and 37% went to one sponsored by a school or university.
It’s not entirely a surprise. One had to wonder how long the hackathon phenomenon would go on in its current state. Hackathons have become a crutch for many developer marketing professionals who don’t really understand what to do, but while they are good for getting limited grass roots support for platforms or tools, they are also labor intensive for marketers, they involve cost, and by their very nature are extremely parochial and thus limited in reach. When one considers the amount of resources it takes to reach probably less than 100 developers it really doesn’t add up.
So what’s next? Online contests deliver far more bang for the buck, but for in-person events, try meetups. Last year almost three quarters of developers went to at least one meet up.
The new Developer Marketing 2018 survey is exclusively focused on tactical marketing outreach efforts and understanding developers. The 221 page reference covers topics such as: Demographics, Firmographics, Psychographics, Purchasing Authority, Outreach Vehicles, Motivations, Using Social Media for Recruitment, Training, Conferences and Hackathons, Encouraging Participation in an Online Community, Using Search Engines for Marketing and more. Margin of error is 4.2%.
More on developers……..
This past Monday and Tuesday we celebrated the 14th year of bringing top developer relations professionals, evangelists, and strategists together in a one of a kind event. The conference highlights the methods and insights of professionals in the largest software companies in the world, and allows free and easy discussion amongst peers. While some little shows have sprung up lately geared for program managers in small companies and startups, the Evans Data conference is designed for the biggest and the best. Some of our speakers and keynotes included executives from Intel, Amazon, IBM, Google, Facebook, HP, Ford Motor Company, Salesforce, General Electric and more! We also featured Guy Kawasaki, legendary trail blazer in the developer relations space.
The conference was packed – sold out, and we think we’ll need a new venue next year. Plenty of great insights being shared. Throughout the two days attendees were able to benefit from stimulating conversations, workshops and sessions addressing the hottest issues in this area.
One of the biggest concerns of professional at even these large companies was how do developer relations professionals prove their worth to the other departments/ management of the company. Of course some companies are more enlightened about this than others, but it’s nonetheless a concern everywhere. Another way of putting it is how do developer relations professionals demonstrate an ROI on their developer programs?
Counting active users, showing an increase in frequency of developer web site visitors, attendance at hackathons or other events were discussed, but these only measure changes in reach or effectiveness. To demonstrate an ROI requires connecting the program’s activities to company revenue in some way. We talked about following a developer’s journey from awareness to evaluation to purchase or adoption and then measuring the difference between those who join a program and those who do not – a long and arduous process that may or may not show true indications.
It’s far easier to show the worth of a program to the developers themselves – ask them. We’ve found that 69% of developers say they won’t adopt an API that isn’t supported by a program. Sixty percent say a program helps improve the quality of their code or provides their personal productivity. Forty percent say being in a program typically cuts their development time by 30% or more.
THE VOICE OF EVANS DATA 