I Can't Agree with Advice to Specialize

I follow John Sonmez from simpleprogrammer.com. His latest post "I'm Not Sure I Want to be a Specialist" got me thinking. John says:

"I’m definitely ... going to tell you specialize ... It’s going to be very hard to narrow yourself down so much that you’re pigeonholed into a place where you’re not going to have customers, especially if you’re just looking for a job."

I just don't agree. There are plenty of technologies that developers might have specialized in that just aren't in use anymore - the digital equivalent of buggy-whips.

Certainly some specialists have fantastic success, but aiming for that kind of success feels like aiming for a career in the NBA. If you are one of the few who can, that's great. But it's not the kind of advice I'm going to give out as a mater of course.

Specialization has real risks, for individual developers and for teams. For instance, as I mentioned, if I specialize in a technology that goes out of favor, my employability declines, potentially seriously.  Additionally, teams of specialists can face roadblocks when a particular type of work piles up faster than the team specialist can complete it. Or worse yet, when a team loses a particular specialist, predictable team capacity, a critical value for project owners, can tank until a replacement can be found and ramped-up (see Bus Number).

My employer, a consulting firm, seeks "Generalizing Specialists", developers who:

• have one or more technical specialties
• has a strong general knowledge of software development lifecycles
• has at least a general knowledge of business and a willingness to gain at least a minimally functional knowledge of their client's business domain
• actively seeks to gain new skills in their specialties and beyond, in technical and non-technical areas

If the team I am on suddenly needs an an extra push in QA, I am comfortable enough with testing that I can temporarily shift my responsibilities.

My advice to developers is: "Specialize in one or two things, stay on top of market trends without following fads, and always be learning, in your specialties, in tech areas you don't specialize in, and in non-technical areas."

How JIRA Helped Our Team

A colleague at a different client asked about what out-of-the-box JIRA reports we use and how they benefited our team.

Whatever you think about physical card walls versus electronic tools, the fact remains that some teams use them and JIRA is a common one.

Like many good charts, many conclusions can be drawn from the reports below. I have explained one or two specific benefits that my teams have reaped from them:

Burndown: We used it to track daily movement of cards (or lack of same) at standup and it worked to identify that developers we incurring risk by working on individual cards for several days.

Sprint Report: We used it to look at what work was added/removed from the current sprint, and what cards were finished. It worked because it enabled us to detect that developers were pulling work ahead while there were still incomplete cards in the sprint. This gave us an opportunity to have a conversation about swarming to get cards over the finish line and breaking down silos.

Velocity Chart: We used it to to compare committed/completed of current sprint verses past sprints. It worked because we detected a trend of increasing overcommitment and the team was able to adjust sprint commentments so that PMs had better projections and the team reversed the growing feelings that they were failures.

Version Report: We use it to project when our project will be finished. It worked because the PM learned that the team was not going to deliver all the scope by the needed date. The PM and the business were able to agree on reduced scope and the version report projected a good delivery date.

What do you expect from a tech lead?

I've been thinking about the role of Tech Lead. It is one of those titles where I think everyone "knows" what it means but most of us haven't tried to define it.

It goes without saying that a tech lead has to be strong technically, but I think what makes or breaks a tech lead is whether they have strong leadership characteristics. I see those characteristics falling in three categories (Self / Team / Business):

Self
Has an active desire to work with people
Possesses "confident humility"
Can accept a position with less than 100% of time spent working on code (sometimes a lot less)
Knows where the gaps in their knowledge are
Is not concerned with "looking good"
Can delegate work

Team
Can facilitate conflict resolution and team decision making
Wants team and individual members (including self) to grow and learn
Wants team members to succeed and actively works to ensure they receive recognition for it
When necessary, can combine their own experience with the thoughts of a team and render a decision
Understands that people have different learning styles (visual, tactile, auditory, etc)
Shields team from outside influences, so they can get their work done (much as a scrum master does)

Business

Understands the value that IT provides to business
Can speak to technical decisions in a way that addresses business partners' priorities
Understands and accepts that business may only wants to pay for "good enough" solutions, not "best" solutions.

What do you think? Do you agree/disagree with the characteristics I've identified?

What are the units for Story Points?

A co-worker and I were discussing estimation on agile software development teams (Yeah, I know - #noestimates. That was off the table here). I asked him what a story point represents. He said "effort", which PMBOK defines as "measurable work units", and pointed me to a post by Michael Cohn:

"...story points are about time—the effort involved in doing something. Because our bosses, clients and customers want to know when something will be done, we need to estimate with something based on effort. Risk, uncertainty and complexity are factors that may influence the effort involved."

My friend went on to write the following formulas:

(SumOfRemainingEpics + SumOfRemainingStories) / Velocity = Forecasted Duration (+/- uncertainty)

and

Forecasted Duration / Team Capacity = hours

I wasn't so sure that this demonstrated that story points are about time, so I re-wrote his formulas adding units (and dropping uncertainty):

SumOfRemainingEpicsAndStories(points) / Velocity(points/week) = Forecasted Duration(weeks) 

and

( (SumOfRemainingEpics(points) + SumOfRemainingStories(points)) / Velocity (points/week) ) x Team Capacity (hours/week) = hours

Management wants estimates of duration which we calculate according to the first equation by dividing points by velocity (Cohn mentions this too). In either formula, points cancel out. What that says to me is that units of story points don't matter. We can estimate in complexity, risk, time, or elephants! The upside of this is that non-timebased units for points prevents management from translating stories directly back and forth between points and hours, which can cause friction between the team and management

Agile Coach Camp Session - Every Tool Sucks

Yesterday, I returned from Agile Coach Camp 2014 in Indianapolis. Here are my recollections from one session:

Every (card wall) Tool Sucks:


Reasons for card wall tools:

• reporting / analytics
• distributed / dispersed teams
• backup / archive
• regulated industries?

Problems with current tools:

• hardware costs not low enough for large monitors to show the entire wall
• when not able to replace physical wall completely, keeping two things in sync is extra work
• teams often do not keep electronic cards up to date
• either restrictive workflows or so general-purpose as to make configuration difficult and costly
• commercial choices are costly
• continuous improvement of workflow not always easy or quick
• no participan

What might help:

• great reduction in price of technologies like large Microsoft Surface monitors
• QR / barcode recognition, OCR
• home grown tools? still runs risk of cost/time to improve workflow, process lock-in

Other ideas:

• 3M Post-it Plus App
• Interns to write/maintain in-house tool and/or keep tool in sync with physical wall
• Trello

Automated testing against physical iOS devices

Update the iPhone's iOS version to the latest

Install Xcode, Apple's IDE for developing iOS software:
http://itunes.apple.com/us/app/xcode/id497799835

Install Homebrew, a package manager that allows user to install, remove and updated applications and packages:
http://brew.sh/
ruby -e "$(curl -fsSL https://raw.github.com/Homebrew/homebrew/go/install)"

Download SafariLauncher, a simple IOS application that automatically launches Mobile Safari after a certain delay:
https://github.com/budhash/SafariLauncher
git clone https://github.com/budhash/SafariLauncher

If you are not already a member in Apple’s iOS Developer Program, spend $99 and join.
https://developer.apple.com/programs/ios/

Build SafariLauncher in xCode:
{ Tell about setting keys and missing field(?) }

Install ios-webkit-debug-proxy:
“The ios_webkit_debug_proxy allows developers to inspect MobileSafari ... on real and simulated iOS devices”
brew install ios-webkit-debug-proxy

Install appium:
http://appium.io/
brew install node
npm install -g appium
npm install wd

Launch ios_webkit_debug_proxy:
ios_webkit_debug_proxy -d
Go to http://localhost:9221/
You should see “iOS Devices” and nothing else
Connect iPhone
You should see “1. localhost:9222 - iPhone” where localhost:9222 is a hotlink
Disconnect iPhone.
Refresh browser. localhost:9222 is no longer a hotline
Reconnect iPhone
Refresh browser. Hotlink is back.
Click link and see: “Inspectable pages for iPhone:”
Launch Safari on iPhone and go to google
Reload browser on Mac and see a line added to the page: “1. https://www.google.com/” where https... is a hotlink
Click hotline and you will see ios_webkit_debug_proxy tool, which you can use like Firebug to inspect elements on the page.

Get the UDID of the iPhone:
http://whatsmyudid.com/
When you see your UDID in iTunes, you may need to right-click it and select Copy. Keyboard shortcut may not work.

Run appium server:
In a new terminal shell, type:
appium --session-override

Write a test that uses SafariLauncher to open your website and make Selenium calls to test the site:
{Beyond scope}

Run your test against the iPhone simulator:
In another terminal shell, type:
mvn -Dtest=com.saucelabs.appium.SimulatorSafariTest test
Watch as the simulator starts (SLOW)
You will see SafariLauncher start (SLOW)
You will see the website launch (SLOW)
You will see your test steps performed.
Look at output of appium window.

Stop appium server:
In the terminal shell where you launched appium, type ctrl-c

Run appium server and tell it the id of the iPhone you want to use:
In same terminal shell used for , type:
appium --session-override --udid e8fae...719
Use the udid for your iPhone, not e8fae...719 shown above.

Run your iPhone test:
In the same terminal shell you used to run the simulator, type:
mvn -Dtest=com.saucelabs.appium.iPhoneSafariTest test
Fix any compilation errors
Watch iPhone screen.
You will see SafariLauncher installed
You will see SafariLauncher run.
After a delay, SafariLauncher will launch your website in Safari.
You will see your test steps performed.
Look at output of appium window.

Updated Silde Deck from Agile and Beyond

My revised slide deck for "Refactor Your Software Career" is now available at the Transformative Networking home page or you can use this direct link.