What We Learned in 2018
A look back at the tech that defined 2018 and some cautious outlooks for 2019.
We learned crypto-currency may not be the holy grail, but blockchain can be useful.
Bitcoin saw a rather bumpy ride in 2018 as it started high around $15,000 and proceeded to fall down to around $3,000 where it sits as of this writing. The rise of regulation, theft, and merchants pulling out of accepting BTC due to its volatility have put a damper on the coin’s value.
But not all is bad, though. Major players in the cloud-computing space — AWS, IBM, and others — have successfully rolled out blockchain-based products. Blockchain, the technology that underpins crypto-currencies like Bitcoin and Ethereum, allows for relatively secure distributed transaction logs that are difficult to forge. This kind of security is extremely useful in industries where products or items need to be tracked throughout their lifecycle, like packages, vital records, and identities. Being able to track a diamond from the moment it is unearthed in a diamond mine all the way to its final purchaser ensures that legitimate stones are not swapped for or stolen or illegitimate ones — aka, “blood diamonds” or “conflict diamonds” — allowing governments and merchants to ensure they are not trafficking in illicit goods.
Kubernetes hit its stride as a major force in the scheduler wars. But for how long?
The rise of Kubernetes has been on the horizon for a long time now and its prominence was just about a foregone conclusion due to its backing and heritage (Google). Born out of Google’s Borg scheduler which it uses internally, Kubernetes is a “container-platform-in-a-box” designed for modularity. Recruiters are now looking for people with Kubernetes experience because it has become the de facto container runtime, with even AWS and Google Cloud now offering a managed Kubernetes service.
Stop me if you’ve heard this before: new open-source virtual infrastructure platform gains massive popularity due to its hyper-scale ability, giving companies the ability to run their own internal virtual platforms. After years of sunk budgets, millions of developer hours contributing back to the project to fix its myriad issues, and the growing complexity of its demands, companies finally pull the plug on their efforts and move to more stable and sustainable platforms.
Sound familiar? It should if you’ve followed tech for any length of time. Because that’s the story of the rise and fall of OpenStack, and Kubernetes is right on its heels. Like OpenStack, Kubernetes’ modularity and scalability comes with a complexity and human-hours cost. Also, with Kubernetes having its first major vulnerability this year, its only a matter of time until security researchers find more. While its popularity probably will last through 2019 due to companies just now getting their implementations up, its sunset will come sooner than you might think.
The Go language was a huge hit.
We’re seeing more and more products developed in Go these days. Even Elastic, a historically JVM-centric company, is starting to release apps written in Go. And who can blame them? Go is a strongly-typed C-like compiled language that has concurrency baked into it. Its relatively shallow learning curve for those of us who came to it from JavaScript or Python helps drive its adoption. So far it has been stable, performant, and popular among developers.
But that stability comes at a cost: the language maintainers have stated that it will stay rather static and they don’t plan on adding any of the niceties that make languages like Python easier to work with. How long will developers put up with not having generics, a reliable dependency management system, having gotosand labels in their code, or any of the other advancements in computer science post 1995?
Security is still a problem.
Once again, security vulnerabilities continue to hang over our heads like so many boogiemen. From state-sponsored cyberwarfare compaigns to engineers being complete idiots, 2018 saw its fair share of security stories. We saw the government report on the Equifax hack that took place in 2017, and it wasn’t pretty. Additionally, high-profile companies like Ticketfly, Orbitz, Delta, Sears/KMart, Macy’s, and Adidas all disclosed security breaches of their own — and that was all before the end of July. The US Government didn’t remain unscathed either. The Department of Homeland Security disclosed a breach in which 247,167 current and former DHS employees had their personally identifiable information stolen by a former DHS employee. And these are just the ones that we hear and know about; there are countless other breaches happening all the time that are either never found out, never disclosed, or never reported on.
Part of the problem is that cybersecurity is trying to prove its value by proving a negative: good security is worth it because nothing happens, but does nothing happen because of the security or because no one targeted the company? (This is an over-simplification but it serves to prove the point.)
The other part of the problem is the talent pool. Cybersecurity is not a young field, some of its core principles have been around since the inception of computing, but modern security engineers are tasked with understanding those principles but also understanding modern computing as it grows ever more complex. The net result is a corps of security engineers that either understand the principles and try to apply them to everything (“security by checkbox”) or that understand modern computing but don’t understand good security practice. Its a demanding field but its one that we cannot afford to continue with in its present state. The next year should be interesting.
Artificial Intelligence
The artificial intelligence (AI) hype-train is real. Everyone from IBM to Jabra have jumped on board and are offering “AI-powered” products, some with more realistic claims than others. And that’s certainly part of the problem with the AI buzz: there’s a lot of snake oil out there from companies just trying to ride the hype. Personally, I take issue with the use of the term “AI” to describe anything that cannot pass an unscripted Turing test, but I digress from the larger point: that not every problem is a potential application for machine learning (what today’s “AI” really is). Machine learning doesn’t need to, and probably shouldn’t be, used to cancel out background noise on a pair of headphones when we can do that just fine as it is, nor should it probably manage our finances, yet we see these types of products popping up all over the place.
So what does this mean for 2019? Expect a huge wave of AI-powered IoT startup extinctions. Consumers are going to start realizing that the AI-powered TV remote they just shelled out an extra $50 a month for doesn’t help them change channels any better than their free one, or that maybe managing an old analog thermostat wasn’t that hard after all and they don’t need a wifi-enabled AI-powered IoT device that costs $300 do it for them.
