How technology bias impacts cloud architecture

In the summer of 2021, I posted a short article on social media titled “Ethics and Cloud/IT Architecture” that absolutely no one read. I’m not sure if it was the topic or the platform, but my attempt to discuss what people were thinking about technology bias went nowhere fast. 

Technology bias is something cloud architects, developers, security specialists, specific public cloud subject matter experts, CIOs, CTOs, CFOs, board members, and even stockholders must deal with all the time. The core issue is simple: Someone wants to select strategic technologies based on their specific bias rather than take a more objective view of what technology is truly best for the business. The bias could be caused by something as innocuous as a decision maker’s brand loyalty to a vendor they’ve worked with in the past, or the most obvious bias of a department head who received an all-expenses-paid invitation to a vendor’s winter “conference” in Hawaii.

This is what I term “implicit bias.” As I wrote in my article, “implicit biases and the resulting ethics concerns are a result of formal and nonformal relationships between those who select technology and those who sell technology.” 

We don’t have as many problems as we did when technology companies could purchase influence with extravagant trips and gifts. These days, the biases are more subtle and may not even be recognized or understood by those who possess them. We can be influenced by a personal or working relationship, a formal or informal partnership, a financial investment, or even by the technology press. 

Those who put together a cloud architecture, either as the architect or as someone involved with selecting and configuring cloud technology, often experience bias through one or more of these situations:

• The company already has a partnership with a cloud provider or other technology vendor. Someone (perhaps even you) has decided to limit technologies to those specific providers.
• Most of the architecture and/or development team has experience, training, or certifications with a specific cloud provider. They fail to consider that new providers might be a better fit or might be used together with other providers to create a more optimized solution.
• A certain type of technology might be considered as a negative. We saw this “no cloud here” attitude several years ago change to “only cloud here” today. 

If these biases cause us to make the wrong choices (as they often do), we’ll end up with underoptimized cloud architecture. This could end up costing the business as much as 10 times the cost and lost value of the most optimized solution, considering operational cost overages as well as strategic business value that’s being lost. 

Worse, the inefficient solution built from human biases does not have the decency to fail. The solutions work to some extent, so many businesses keep running the underoptimized cloud solution without any understanding of the value they leave on the table. 

You really can’t eliminate all the biases that people have because…well, they’re people. It’s important to watch closely for bias when we’re tasked to select the right cloud technology and the right configuration to get to the most optimized solution. The best approach to mitigate the bias problem is to ask probing questions about the suggested solution and its components. Review the terms of development and operational costs and then challenge each assumption and look for proof. 

I’m the “designated jerk” on my clients’ projects. Another approach would be to have a second independent team vet the solution and get a second opinion. Have them look for bias issues and proof that the cloud architecture really is the most optimal and considers all business and technology requirements. The extra cost makes this approach rare, so the “jerk” questions from the internal team are that much more important to ask and answer. 

At the end of the day, understand that we’re not perfect. We all have biases around technology solutions and providers, be they single cloud versus multicloud, databases, security approaches, etc. Some biases are easier to spot than others, but we must learn to recognize and deal with them. The alternative is to leave billions on the table in lost efficiency and strategic value. 

So, the next time you hear “it works,” ask if it works well enough. And wonder if you’ve made the right choices for the right reasons.