FunctionSource, Your Source for Developer News

Some conference talks are interesting, some are entertaining, and others are topical. Now and then you get a talk that doesn’t just give you some new info, but changes the way you think about a topic.

The talk that comes to mind for me is one by Rich Hickey (of Clojure fame) talking about flaws in Object Oriented design and the roles of time and state. He tends to come up with great stuff all the time, even when spending time in his hammock (or, because of it).

I just saw Dan North give a talk that resonated greatly. The talk may have the tag “Agile”, but he shares his concern in how that word has lots its way. When you look around and see people talking only about the process versus the product, you are in fail territory. When you see priopriatary language to make things seem so special, and certification instead of just working on making the delivery process get better over time, then you begin to worry if the train is off the tracks.

Dan left one of the temples of Agile (ThoughtWorks), and has found himself at a trading company that was delivering “against the rules”, and it got him questioning the doctrine.

I feel a little bit of a kindred spirit in that I have ended up at a larger company that does things a little differently too. At Walmart Labs we get to deliver software pretty much as we wish, and are changing our process over time. One area of the mobile world that I live in is to do with conversion of mCommerce. That piece aligns squarely with the trading universe that Dan is in. When you make a change to the software you can measure the effect in simple turns. How did it affect conversion? This is a perfect storm to wind a team up and A/B their way to constant improvement. With the right system in place, anyone on the team can have a hypothesis, and the team can decide which ideas to measure and go after.

We also have other products (mainly on the “in store” side of the coin that is Walmart Labs Mobile) that are more in a “what is our vision?” vein. We also have constraints such as “people buy most of this stuff for the holiday period.” This means that we have to take care to have a process that doesn’t have us delivering great holiday software after the fact, and we need to be ready for the scale…. something I am sure Dan’s team are constantly aware with in their field.

Now, the title of Dan’s talk is “embracing uncertainty”. This is one of those “duh” moments. You knew this to be true. It is common sense when you hear it explained, but it was never in your mind in those two word terms.

The core fact that I agree with and have seen time and time again, is that people want hard dates even though they are wrong. They don’t want to deal with uncertainty, and would rather choose something that is going to be wrong.

I have long thought of a “roadmap” in a way where it starts to fade over time. You may be close to opacity:1 for today, but there is a sharp fall. This is why I have liked the premise of a tool like LiquidPlanner that allows you to be honest with yourself with respect to putting a much bigger range on tasks that you know how little you know how to estimate. No matter how honest you are with yourself, and how great the tools are, you still end up with the “unknown unknowns”, which will screw you anyway.

In general my feeling is that as long as you see improvement over time, everything else works out. Rather than pushing for a crazy rate of change, can you build a smaller rate into your lifestyle…. so it can be embedded for the long term (sounds like losing weight slowly vs. fast. Huh :).

How many times have you put something on paper that you know isn’t reality? As soon as I see that, I am going to put a mark on it for “UNCERTAINTY”. I think we need a logo! It’s OK to have uncertainty, so lets work with it and not against it.

Anyway, definitely give Dan a good watch:

I want to grab a beer with him to talk about how he posibly could have taken a look into faith and ended up with the conclusion that he mentions in this talk. I have done similar soul searching in the past, and the conclusion has been the exact opposite. Well, it would also just be good to compare notes on software delivery.