22 August 2008

Flight Level 086

I went on a flying trip, this time on the right seat. As far as I can remember, this was maybe the second or third time that I was sitting to the right in a light aircraft and not pilot flying.

An opportunity so rare like this one needed to be captured in a photo :)

01 August 2008

Book Notes: Writing Effective Use Cases

As I read books, I note the contents that I think are most important - for me. You might find it useful, but consider reading the book for yourself: Alistair Cockburn: Writing Effective Use Cases.

This books describes, quite formally, how use cases should be written. Personally, I did not like the book because of that formalism. Most of it deals with highly detailed fields you should have in your use cases, which might be required in some large scale enterprise systems where there are many QA-infected people around who actually believe that such details really help anyone. I don't, but your mileage may vary.

Nonetheless, I found some valuable information in the book.
  • Stories (scrum) are de-facto use cases, but very informal ones.
  • Each story should have 2 ends: the success scenario, and failure scenario. Describe both.
  • Never write stories in passive form, always include WHO does WHAT. Example: "Error is being displayed" is bad, better is "Error view displays a list of all errors".
Use cases should always be easy to read, in simple grammar. They should cover and describe scope, actor and goal-level.

Pass/Fail test for use cases:
  1. Ask your user if he understands what is going on in the use case.
  2. Ask developers if they are able to implement a feature based on the use case.
Formally, the book mentions these parts of a use case:
  • Actor
  • Stakeholder
  • Scope
  • Preconditions and Guarantees (=Postconditions)
  • Main Success Scenario
  • Extensions (like failure scenarios, what can happen differently)
Final advice: Always control your scope. Make a in/out list of items that are in your scope and those that are not. This applies to everything, not only stories but also meetings, papers, documents etc.

30 July 2008

Very Special Off Time

My mother called me today to tell me to check the local media: a small airplane crash-landed after a failed attempted departure from my base airfield. Pilot and passengers were not hurt, but it still is a sad sight to see a fine aircraft in that condition. The damage looks significant, but the airframe might be repairable.

I took this as a good sign - it is highly unlikely that two accidents would happen within 24 hours. I left office, got to the airfield and put the clubs Cessna to good use for a short vacation - despite the thunderstorm that was rolling in from the west at 8:30am. Thunderstorms that early are highly unusual. It dissolved quickly, all that was left as we were airborne was some occasional light rain.

One colleague brought his cell phone and used it to film the video below. Mind you - this was done on a cell phone! This is better than what I get from my own digital camera (which, by current standards, is a dinosaur). This ought to change, and soon.

The wreck shown in the media picture above is still there, we saw it just off the runway as we landed and you can see it in the video, which - by the way - is a quick patch-together done in iMovie. What it lacks in quality, it should make up for in scenery!


28 July 2008

Book Notes: Scrum and XP from the Trenches

As I read books, I note the contents that I think are most important - for me. You might find it useful, but consider reading the book for yourself: Henrick Kniberg: Scrum and XP from the Trenches.

This one even comes as a free online edition, which I printed out and read in one afternoon. Then I bought the printed edition for $22.95 - worth every penny. If they had a hardcover edition in color, I would not hesitate a moment to get that as well.

I got the printed edition specifically to let it circulate in the office and do some damage. Within 4 days, 3 people have read it, and the 4th one has it on his desk now. One of our teams had its very first retrospective after that, and suddenly the quality of our stories is the subject of many discussions. Many, good results within less than a week - that's how good this book is.

Here is my brief of the content, in my own personal order of importance.


Scrum Master Checklist
  • After Sprint Planning
    • Create Sprint Info Page
    • Link to it from the wiki
    • Print it, put it on the wall for bypassers to see
    • Spam everyone with sprint goal + link to sprint info page
    • Update sprint statistics (velocity)
  • Every Day
    • Daily Scrum Started + Ended on time
    • Stories added / removed to sprint backlog as needed to keep sprint on schedule
    • Inform PO about changes to the sprint backlog
    • Backlog + Burndown is up2date
    • Resolve problems + impediments, report them to boss
  • End Of Sprint
    • Sprint demo (everyone should know the date)
    • Retrospective
    • Update sprint statistics

Retrospectives

Goal of the retrospective is to improve. Answer the question: "What can we do better?"
  • Plan 1-3 hour retrospectives, in separation - without disturbances.
  • Show + Summarize sprint backlog
  • Do the round. Everyone gets the chance to say good+bad+ugly+ideas
  • Summarize ideas + suggestions
    • Good: what would we do again
    • Improve: vote on ideas for improvement, select only the top 1-2 ideas
    • Not all bad things need to be acted upon. Some problems solve themselves with experience.

Sprint Planning Agenda

  • 30min: PO goes through sprint goal and summarizes backlog.
  • 1h 30min: Team time-estimates, breaks stories down into tasks. Importance of stories is updated accordingly, items are clarified. How to demo is filled in for all high-importance items.
  • 1h: Team selects stories to be included in iteration.
  • 1h: Select time + date for scrum meetings. Further breakdown of stories into tasks.

Each Story has:
  • ID
  • Name
  • Importance
  • Initial Estimate
  • How to Demo (this is, in essence, a use case, or user acceptance test)
  • Notes
Optional:
  • Components (We don't have this)
  • Requestor
  • Description: We have that additional. This could probably be the same as "How to Demo", but sometimes contain technical details that a user story should not have.
Important: No technical language! Describe stories in user terms that anyone can understand!

Preparation for Sprint Planning
  • The backlog should exist.
That's it. Surprisingly simple. In more detail:
  • Each story that has a remote chance to make it into the iteration should have its own distinctive importance.
  • Product Owner (PO) understands all stories, and can clarify them.
  • PO is the only one who defines priority.
  • Team is the only one who defines estimates.

Sprint Planning

  • Output:
    • Sprint Goal
    • Know which team members are available
    • Commitment of the team
    • Backlog for sprint
    • Time + Place for daily scrum meetings
  • Agenda:
    • PO summarizes goal + most important stories
    • Team estimates stories, starting with the most important
    • PO answers scope questions for stories.
      Note: Code quality is part of the scope of any story that is absolutely not negotiable. Code must be clean, hack-free and, if possible, unit-tested. UI can be crude if it is going to be fixed later, but quick fixes in underlying code are a big NO-NO!
  • Agenda Detail, Example:
    • See Above

How to decide which story goes into the iteration

  • Use your gut feeling.
  • Once you did a few iterations, you know your velocity and can use that too. I don't know my velocities, so I didn't read those few pages.

Use Index Cards


Usually, teams sit around a computer or stare into a beamer while doing milestone planning. They go through story after story until they're finished. This sucks, big time - because it is a big waste of everyones time: while one person is driving the keyboard, all others are forced to watch and fall asleep. Using index cards not only keeps them awake, it also is faster and more fun. Update your excel sheets or whatever you use AFTER the meeting.
NOTE: I found this advice to be one of the best, and just used it in my second planning meeting. It works wonders, really is much faster and more fun.

Importance values are used only initially for the priorities. Strike them out as soon as you re-order your stories.

Time-Estimation is easier if a story is broken down into tasks first. Stitch tasks with small post-its onto your stories.


Definition Of Done

  1. Alpha-Tested
  2. Tester says it is OK
  3. Unit tested
  4. Documented (short)
Don't be too strict with your definition of done, and don't apply it to each story equally.
Bummer. We still have none of that, and it came back to haunt me today. Big time. Some day, there'll be a post in the "In The Pit" series about that.


Breaking Stories into smaller ones

  • Story: Stuff that is visible to the PO or user.
  • Task: Stuff that is not.
Example: the story "User Management" could be broken down into "Add User", "Edit User", "Query Users"


Possible Default-Tasks for stories

  • Clarify Requirement
  • Write test cases
  • design GUI
  • implement core
  • implement UI
  • integrate, debug, refactor
  • test + document

Where to draw the line

The result of a sprint planning should be, in order of importance:
  1. Sprint Goal + demo date
  2. Stories accepted for this sprint
  3. Estimates in all sprint-stories
  4. How to demo filled out
  5. Velocity + resource calculation
  6. Time + place set for daily stcrums
  7. Stories broken down into tasks
If the planning meeting drags on and on, cut it after your defined time-box expired. Brutally. Let the sprint suffer. Adding extra hours to the meeting is not going to help the team to be able to break stories into tasks or clarify them, It is only going to wear everyone out.


How to deal with tech stories
  1. Try to avoid them. Transfer them into normal stories.
  2. Do it as a task in another story
  3. Treat it as a normal story. This will be difficult to sell to PO and users.

Communicate Your Sprint
  • Spam the company.
  • Create a wiki dashboard, keep it up to date.
  • My own addition: make sure people know the backlog, especially management and sales.

Demos

Team members should at least attend the demo, even better: give the demo themselves.


25 July 2008

Book Notes: Applying Family And Group Psychology to Companies

As I read books, I note the contents that I think are most important - for me. You might find it useful.

I found these notes in my notebook, but I did not note down which book contained them. The book would not make much sense to the casual reader anyway, it did to me only after I talked to psychiatrists, psychologists and mediators. Some of the notes below come directly from those discussions, and not from any book.

I use the term "system" several times here, which refers to the "whole" as an entity in the sense of systems thinking, but with a psychological background. For this purpose here, you could use the term as a synonym for: the company, people owning or running the company, the purpose of the company and the history of the company. I know, it is complicated. This is psychology, not technology.


Dignity
  • If a young person puts himself before an older person, the younger person violates the older persons dignity. The only relevant factor here is age - not competence, education or pay.
  • You can preserve your own dignity best if you put yourself in your systemically (read: hierarchically) appropriate spot.
  • Your own dignity is deeply connected to humility. Humility is accepting that there exists something bigger than you are. (Note: this is in no way religious. "Something bigger" can, in the case of a company, be the purpose, history and dignity of the company itself, in essence: the system.)
  • People cannot bestow dignity.
  • Violating someones dignity always comes back to haunt you and the system.
  • If your job puts you into an "inappropriate" spot (for example, you find yourself being the boss of someone older than you are), you can remedy that only by honoring the people which whom you are in conflict (i.e., by honoring the older person and treating him with the utmost respect).
On appropriate positions
  • In a job, the "appropriate spot" is where you can apply and extend your abilities best.
  • Someone who never is utilized to his full capabilities in his job has to leave or switch into a different position. So does someone who is permanently used beyond his capabilities.
Lasting damage to the system
  • Someone who inflicted lasting or ruthless damage to someone else or to the system must leave it.
  • If someone is wrongfully fired, or in an other, offensive way removed from the company, he remains a part of the system - even though he officially is not. This has serious consequences: it affects processes and employees in a negative way, blocking them. It is not even relevant if the person who was wrongfully fired suffered any damage or not.
  • Drawing profit from other peoples harm has, in the long term, a disastrous, lasting effect - especially in the next generation of the owners. It leads to the demise of the company and the personal demise of the owners.
Hierarchy
  1. Someone who leads always has priority.
  2. Seniority (length of service)
  3. Competence
Leadership must be justified by competence and achievement, only then will the leader have authority and respect. A leadership position always stands in service to the company and to subordinates. It is no means to the end of a fantasy of power. The leader is grown up and is ready to take responsibility.

Hierarchy based on seniority has a purpose: the exchange of experience.

Leadership and seniority sometimes stand in conflict with each others. Leaders have to hold seniors with respect.

Competence needs to be considered, irregardless of the hierarchy. Ideas that come from the base can be very valuable, but are often ignored because they come from the wrong people.

Compensation
  • Service and achievement MUST be appreciated. If it is not, either achievement withers or the employee leaves the company. The opposite is true as well: lack of competence must be recognized, and be dealt with in a constructive way.
  • Special achievements from employees require a special compensation from the employee - so do lacking achievements.
  • People who, in the long run, give too much and get too little in return will leave. If they don't, they will burn out. People who take too much and give too little have to leave. If they stay, others will follow the example and begin to demand too much from the employer - without showing adequate performance.
Seating arrangement
This is very old knowledge which was common place until a few decades ago, but somehow got lost over time. But it is still true: the seating arrangement on a meeting matters. It should be hierarchical, and the person sitting to any one's right should be senior. The only exception is at the start and end of the sequence - if you sit in a circle.

20 July 2008

Book Notes: Agile Project Management

As I read books, I note the contents that I think are most important - for me. You might find it useful, but consider reading the book for yourself:
Jim Highsmith: Agile Project Management

This is one of the best books I've read about agile development so far. I highly recommend it to anyone who works with agile projects. It is a must have in any book shelf.


Linear thinking, prescriptive processes, and standardized, unvarying practices are no match for today's volatile product development environment.

We value:
  • Individuals + Interaction over Process + Tools
  • Working Code over Comprehensive Documentation
  • Customer Collaboration over Contract Negotiation
  • Responding to Change over Following a Plan
Documents are no suitable replacement for direct customer contact. Meeting protocols are, best case: noticed - worst case: an impediment.

Encourage exploration.
Build adaptive teams (self organized + self disciplined)
Favor Simplicity.

Deliver Customer Value on each iteration.
Employ Iterative, Feature-Based Delivery
Champion technical excellence (which leads to adaptability)

You manage things, you lead people. Management deals with complexity, leadership with change. Both are necessary. The more exploratory the project, the more important is leadership.

Leaders are leaders because of who they are: they depend on influence derived from respect, which in turn is based on integrity, ability, fairness, truthfulness - in short: character. Authority is earned bottom-up, not bestowed from above.

Managers fear change, leaders encourage it. Exploration can cause anxiety and fear. Good leaders create a safe environment, where it is OK to voice outlandish ideas.

PMs who focus on diagrams, budget and histograms are dooming their team to mediocrity. Innovative teams are led, not managed.

Teams need a purpose, goal, vision. That does not change - execution of the task does. They need encouragement to explore, make mistakes and regroup.

Shared space: Interaction by using demos, prototypes, simulations. Managers, customers, developers, marketers have a better time discussing when there is something tangible to base the discussion on. Requirement documents are too abstract - not visual, not common. Create tangible artifacts - you have to know who needs to interact, and what those people can base their interaction on. Not all artifacts are common-place and can be understood by anyone.

Vision: Targeted deadlines (+ reasons for those deadlines), high priority features (only few + reasons for them)
Architectural overview, component + interface description. Not too detailed.

Interaction: Trust and respect lead to meaningful debate, and vice versa.
With too little coordination, teams will diverge so far that integration becomes a nightmare. Too much coordination, and teams will be clogged by meetings and information overload.

Cooperation first, documentation second. Useful information is based on cooperation 70%, documentation 30%

Insist on Accountability: Individual commitments to delivering a feature in an iteration -> accepts accountability when PM is involved, provides requirements + information, resolves impediments.
Commitment -> hold accountable.
Hold each others to commitments.
Meet commitments -> builds trust.
Trust is the foundation of collaboration.

Create an adaptable, self-organizing team: grant as much autonomity as possible, destructure to the extent possible, then coach individuals and get rid of those wo don't fit.

Make unilateral decisions at times if necessary, make decisions with team involvment, but primarily delegate decisions to the team.

Self Disciplined Individuals:
  • Accept accountability for results
  • Contemt reality through rigorous thinking
  • Engage in intense interaction and debate
  • Work willingly in self-organizing framework
  • Respect colleagues
If you get the right people, self-discipline comes more easily. Get the wrong ones, imposed discipline creeps in, destroys trust, respect and egaliterian atmosphere (p.72). Note: this is emphasized because it was the problem on my team.

Simplicity:
When you want your boat to go fast, it is easier to cut anchors than to add horse power.
Simplify your process by taking out the detailed tasks and compliance work -> force people to interact. The only rule: Us your good judgement. Be quick, but don't hurry.
  • Simplify the process
  • Simplify the activities
  • Simplify compliance

Agile Development + Framework:
Encourage
  • Thinking
  • Acting
  • Interacting
If your goal is reliable innovation, your process must be organic, flexible and easy to adapt.

Explore:
  1. Deliver planned features
  2. Create a collaborative, self-organizing project community
  3. Manage interactions with customers and PM
After your envision phase, interactions will consist of the following loop:
  1. Speculate
  2. Explore
  3. Adapt
  4. Speculate...
This requires good judgement!

Envision:
Create a clear, compelling vision
  • is hard
  • takes work
  • takes leadership
Elevator Test Statement:
For (target customer) who (statement of need or opportunity), the (product name) is a (product category) that (key benefit, compelling reason to buy). Unlike (primary competitive alternative), our product (statement of primary difference).
Product Vision Document:
  • Mission Statement
  • Target Customers + their needs
  • Customer Satisfaction Measures
  • Key Technology
  • Critical Product Constraints
  • Competitive Analysis
  • Key Financial Indicators
Get The Right People
The top 20% of people produce about 50% output.
Key Qualities are: a) Capability b) Self Discipline

Speculate/Planning
  • Formulate Features that are understandable to customers: features act as interface to customers
  • Feature Cards: Front: Feature information for planning, Back: technical Task information to estimate effort
  • Have a Product Feature List (backlog)
  • Feature = Piece of a product that delivers value to the customer
Feature Cards:
Name, ID, Description (in customer terms), type (customer/technical), dependencies, estimated effort, acceptance test

Iteration planning:
  • Take high risks early, before high values
  • Have iteration themes (a.k.a. sprint goals)
Risks:
  1. Inherent schedule flows <--> Iterations, Regroup
  2. Requirement Inflation <--> != Evolution!
  3. Employee turnovers <--> Collaboration
  4. Specification breakdown <--> Decide!
  5. Poor productivity <--> Morale + People
Project people need to be visionary, positive and brutally honest about the risks.

On Paper...
APM advocates less up-front planning, architecture and requirements gathering because information is gathered as the project unfolds. Up-Front paper is lying to the management.. but that's just my take on it.

Arichtecture...
...needs to have a good mix of up-front and just-in-time.

Get the right people
  • Appropriate technical ability
  • right behaviour
  • getting the right people on the bus is more important than where the bus is going
  • most companies build bueraucracy to manage the small amount of wrong people, driving away the right people, creating the need for even more bureaucracity -> wrong people work on delivery, right people work more on compliacne work
Release/Iteration Plan
Work- or task driven plans often degenerate into very detailed, prescriptive plans.
Teams often get lost in intermediate artifacts, that have little bearing to progress.
Feature driven planning helps overcome this problem.

Technical Dept
lip service about technical excellence and being pushed into hurrying lead into technical dept.
CoC (cost of change) raises slowly first, but rapidly later...
  • no allowance for periodic refactoring
  • poorly maintained test data
  • constant pressore to cut time
are DEADLY.
Without firm dedication to long term technical dept management, development teams are pressed into the trap:
  1. dept leads to delays
  2. delays lead to pressure
  3. pressure lead to hurrying
  4. hurry leads to dept
The secret to long term dept management is: do it early, at lower cost, easier to justify.

Handling technical dept can result in architectural changes. Holding the CoC down by not changing architecture means that when change has to happen, neither code nor people will be ready for it.

If not done:
  • Loose customers to more responsive competition
  • Good agile team will work faster with technical excellence
  • Excellence costs a little more, but reduction in compliance work will more than pay off
Simple Design
Remain grounded on what you know rather than what you anticipate. ADAPT to change, don't ANTICIPATE it. Good architectures encompass aspects of both. Anticipate known change only.
Value adaption over anticipation, but value both.

Coaching + Team Development
  • Focus on delivering results - revisit key constraints and tradeoffs
  • Cheerleader: members want a boost occasionally
  • No blabla: facts, even negative ones
  • Key motivator: understand what outcome is expected
  • Right people find their own way, if they know where to go and how to travel
Core elements needed to attract, focus and keep key employees:
  1. Understand expected outcome
  2. Understand why what they do is important
  3. Work with a team that is committed to quality
Don't give a vision that is too far away, otherwise there might be doubts about its possibility. A 2 year plan can be daunting, a 2 month vision easier.

Molding a team
  1. Trust. No trust -> superficial interaction
  2. Interaction. No interaction -> focus on individual goals
  3. Satisfactory Conflict Reolution. Not done -> reduced trust
  4. Participary Decision Making. If not participary -> undermine commitment to team
Trust and respect are closely tied together.
If, fundamentally, you don't trust people, then there is no line, no pointin time, beyond which people suddenly become trustworthy.

Some failures will occur, don't impose strict controls. Get rid of untrustworthy individuals instead of abusing the entire organization with rigurous controls.

Channel debates to build trust and respect rather than undermining them. Focus on issues, not individuals.

Manage team mood by managing your own mood.

Interaction drives innovation. No interaction -> no synergy

Each team member must take initiative to confront others when they are not performing or behaving according to team rules. This includes calling the PM if situation dictates. No exemptions.
Discussions need to be focused on getting all relevant information onto the table.
Mood, or "emotional intelligence" in leaders has much larger impact on performance than imagined. Team members pick up the mood of the leader within a couple of hours.

Rules of engagement:
  • Everyone has an equal voice
  • Everyones contribution is valuable
  • Attack issues, not people
  • Keep privacy within the team
  • Respect each other and your difference
  • Everyone participates
Very difficult, but important: confront team members who violate behavioral or performance standards. If they aren't confronted, mistakes aren't identified, and no learning occurs. Failing to address these issues is one of the greatest complaints against project managers.

Reviews and adaptive actions
  1. Reflect, Learn, Adapt
  2. Change Pace: Relaxed Review Day
Customer Focus Groups
focus on discovering + recording desired changes, do not gather detailed requirements.

Technical Reviews, Team Performance Evaluations: Retrospectives.
  • What went well?
  • What didn't go well?
  • How do we improve next iteration?
  • What don't we understand?
Project status reports

The Agile Vision
  • Are you delivering innovative products to your customers?
  • Are you excited about going to work every day?
Agile Processes and Agile People lead to Agile Products.
Project cultures and management controls for exploration usually are at odds with each other. Exploration is key.
Don't work in Dilberts company.
Don't be Dilbert.

APM Style:
Leadership and Collaboration
(NOT: command and control)

Leading is more difficult and rewarding. PM + P leaders champion the vision.
Leadership includes staff selection, deployment and ongoing encouragement. Leaders help the team focus on delivery while minimizing compliance work.
Steer people towards role that match their abilities + talents, develop technical and behavioral skills, encourage people through frequent feedback.
Create environment of "no fear".

Soft skills are hard.
Hard skills are easy.
Use your god damned common sense.

10 April 2008

One Six Right

I've posted some of my pictures, because they say more than words. I found a movie that says more than pictures - One Six Right, an independent production that makes my heart bleed every time I see it. It took me a year to locate a shop that sells the DVD, but I found one. Now, I have the entire film on my iPod and can watch it when the job gets over my head - to remind myself why I am doing it: no job, no flying.

The movie below is a feature on the DVD, called "flight".
Enjoy, and get the DVD!