What to read and watch in 2014 if you are building a system, not a career in hype
A closing post for the year that gives the reader a map of directions, not just a list of fashionable words.
The last week of December is when dozens of "best books of the year" and "technologies to watch" lists appear online. Most of them are assembled on the principle of "what was popular" rather than "what will help you think more clearly".
I want to do this differently. Not a list, but a map. A few directions, each with a concrete orientation - what to read, what to pay attention to. For people who build systems and make architectural decisions, not people who collect fashionable words.
Infrastructure and cloud
If you have not yet understood how public cloud platforms are built internally, that is a gap that prevents sound decision-making. You do not need to become an AWS or Azure engineer, but you should understand what a region and availability zone are, how object storage works, and what separates IaaS from PaaS.
Read the technical documentation from the providers themselves - it is written better than most textbooks. Follow the engineering blogs of Netflix and LinkedIn: they publish analyses of real architectural decisions, not marketing copy.
Data and engineering
The main question in 2014 here is not "which tool to choose" but "how to build data discipline". Tools change fast; discipline persists.
Martin Kleppmann has not yet written his book on data-intensive applications, but the ideas he will eventually put there are already being discussed in conference talks and technical blogs. Look for material on data pipelines, on consistency in distributed systems, on the difference between batch and stream processing.
Watch what comes out of Etsy, Facebook, and Twitter in their engineering blogs: they are solving scale problems that most companies will encounter two or three years from now.
Security
In 2013, events around mass surveillance made security a public topic. For an IT director this means two things: pressure from above will increase, and there will be more room for a serious internal security conversation.
Read not about specific vulnerabilities, but about principles - defence in depth, least privilege, zero trust as a concept. Bruce Schneier writes clearly for a non-technical audience without dumbing things down to meaninglessness.
Machine learning
ML in 2014 is not yet a mass-market tool, but it is no longer academic exotica either. The one thing a manager needs to understand: ML is not a magic button, it is an engineering discipline with its own requirements for data, infrastructure, and process.
If your field has ML applications, study not the algorithms but the implementation case studies: what worked, what did not, what were the mistakes. NIPS publishes its materials openly, but to start it is enough to read practical application analyses in industry publications.
Robotics and physical systems
This is still the topic furthest from most businesses, but 2013 showed that autonomous physical systems are leaving the laboratory. Amazon and warehouse logistics, self-driving vehicles, industrial robotics - all of this is beginning to intersect with real business.
Worth following not for hype reasons, but for this one: where might new operational opportunities or threats appear in your industry five years from now.
The main filter
For every new source or topic I use one question: does this help me think about the system as a whole, or does it only add another term to the vocabulary?
A good piece on cloud architecture helps you understand trade-offs, not just describe services. A good article on data explains why some approaches are brittle, not just list tools. Good security reading changes your threat model, not just adds new CVEs to a list.
If after reading something you think differently about your architecture - that was useful reading. If you only know a new name for something - probably not.