{"API Evangelist"}

A Checklist For API Observability

I have had the Wikipedia page for Observability open in a browser tab for weeks now. I learned about the concept from Stripe a while back and is something that I am looking to help define APIs from an external vantage point. In this world of fake news and wild promises of artificial intelligence and machine learning, we need these black boxes to be as observable as they can--I am hoping that APIs can be one of the tools in this observability toolbox.

Stripe is approaching this concept from the inside, with a focus on stability and reliability of their API operations. I am focusing on this concept from the outside, to "measure how well internal states of a system can be inferred by knowledge of its external outputs". More of a regulatory, auditing, and journalistic way of thinking, but in the API way of doing things. Some of this is about understanding, but some of it is also about holding providers accountable for what they are peddling.

The other day I mapped out what API monitoring means to me, rebranding it as what I'd prefer to call API awareness. I went through the elements I use to understand what is going on with APIs across the sector, but this time I am thinking about them in terms of observability. Meaning, not what I'm doing to be aware of APIs, but what is expected from providers to meet (my) minimum viable definition of a platform being observable.

I do not think APIs being present automatically mean a company, it's data, content, or algorithms are observable. This list represents what I feel enables and empowers observability, and are the key areas where I think companies need to work on when it comes to ensuring their operations are more observable. I pulled this list of elements from my own toolbox for monitoring the API sector but is trimmed down specifically for platform operators who might be thinking about how they can make things more observable for everyone involved, internal or external. 

If you are like me and operating outside the corporate firewall, this list represents the areas you should be encouraging platform operators to invest in when it comes to pulling back the current a little bit more. Different folks will read this post and walk away with different takes on what observability might mean. Some will view it as a good thing, while others will see it is as API driven regulatory interference. Your take probably has more to do with your position in the conversation, than it does to the observability of anything. 

Now for some concrete examples. What is the observability of the Facebook news algorithm? What is the observability of the Twitter feed? What is the observability of our credit score? How observable is COMPAS as it evaluates its risk score for whether a criminal offender will reoffend? How observable is Chicago's food inspection algorithm? How observable is the RSA algorithm which drives cryptography? How observable is algorithmic trading across stock markets? These are just a handful of some relatable algorithms when in reality there are numerous micro impacts of algorithms felt each moment, of each day, as we navigate our worlds.

I do not think there are easy answers when it comes to algorithmic transparency. This is just my attempt to understand how APIs can be put to work making the algorithms that are increasingly governing our reality more observable. I spend a lot of time trying to make sense of the very obscure things like cloud and social. What are they? What can you do with them? Do they do what they promise? When do things change? I shared a version of this list for folks on the outside to develop API awareness. Now,, this is my recommendations for what companies, organizations, institutions, and government agencies can do to make the data, content, and algorithms they use more observable--not just for our benefit, for yours as well.