Thursday, April 21, 2016

Beyond the 'happy path'


In the context of software or information modeling, a happy path is a default scenario featuring no exceptional or error conditions, and comprises the sequence of activities executed if everything goes as expected

- Wikipedia

Software is possible because we have clear, well-defined expectations in a lot of common situations. Given a problem and a particular state of the world, software can reason through known information, and perform according to the expectations.

People who invent software are enterprising, excitable, focused on delivering a working solution for a problem.  Which mostly translates into building software that is great about performing the ‘can do’ portion of the expectations. It is the so-called happy path, where the software works as desired and designed, in the perfect world for the well-behaved user who happens to think in perfect lock-step with the inventor of the software.

However, the world is hardly ever perfect. People and users think differently. Not all users are both well-behaved, and can or will follow the detailed vision of the software designer.

Consider a bank application that allows read access to one’s account balance. It is important that it does NOT

  • allow read access to other people’s account balance.
  • allow write access to any account balance.
  • allow any user behavior to affect other user’s experience and access.

These kinds of expectations are less likely to be included in the specification, because we like to think about software in terms of what it can do, and not what it doesn’t do. These expectations are less likely to be designed into the software architecture, because we like to think of achieving our goals must more, than we like to be concerned about preventing bad things from happening. Finally, these expectations are less likely to be tested, because it is both harder and does not demonstrate a job-well-done even if successful. 

Better handling the non-happy paths requires a more systematic approach to specification building and testing, and range verification.


Tuesday, April 12, 2016

Leading by good looks



… you don't get trusted positions just because of your ability. You also have to attract the notice of superior[s…]. You have to be liked. You have to fit in with the system. You have to look like what the officers above you think that [leaders] should look like. You have to think in ways that they are comfortable with.
The result was that you ended up with a command structure that was top-heavy with guys who looked good […] and talked right and did well enough not to embarrass themselves, while the really good ones quietly did all the serious work and bailed out their superiors and got blamed for errors they had advised against until they eventually got out.

Ender's Shadow, by Orson Scott Card @1999


Likable people that fit within the system do a decent job advancing the system toward success. The kind of success that is easy to imagine, easy to prognosticate, and mostly means being ahead of the competition that does the same thing.

However, if one wants to shoot for the stars, to change the paradigm, to redefine success – expecting good-looking people who think in ways that are approved by the existing system to lead us there is doomed to fail. Occasionally, success is not so much ‘the same thing, just more and better’, but rather a long shot that could turn out to become something different, powerful and amazing. And then there is a need for a different kind of leadership, the leadership that does not work as hard to look presentable, to say the expected thing, to be liked by the superiors.

There have always been unlikeable leaders who changed the world in big and small ways. A few went against the system, and others simply disregarded the system and built their own way. The society at large considered most of them failures during a significant portion, if not the entirety, of their journeys. Others have reached the official definition of success, and occasionally even became liked and popular. Still, leading for change is a risky proposition with high potential for failure.

So when picking a leader, or aspiring to lead, put some thought into what kind of leader you want to follow, or to become. Good looks may not be required. 

Thursday, February 18, 2016

Following Plato Dialogues: conversation about estimates


-            So, why do we need estimates?


-          Well, as a customer, I want to have estimates.

-          Great! In what ways are estimates useful to you?

-          Say, I order from Amazon, and if my purchase does not arrive on the estimated day, I get to request a refund.

-          Huh, so you are rooting for the sender to fail, and that’s why you need estimates?

-          I guess so.

-          Any other reason to want estimates or tracking?

-          Well, if as a customer I was promised a certain delivery, and it is getting really close, but the work is less than half done, I can make a determination that I am not getting that delivery.

-          So, the reason you want estimates is to feel reassured that work is being done? To make you more comfortable with the delivery that you do not trust?


What are good and useful reasons to want estimates?