- What is XP?
- The more humanly we treat ourselves and others, the more productive we all become
- Use a "mentality of sufficiency"
- You can do your best work even when there are constraints. Fussing about the constraints distracts you from your goals.
- It's other peoples' job to manage their own expectations. Just communicate what you know to set that expectation.
- Learning to Drive
- Stay aware. Adapt. Change.
- Values, Principles, and Practices
- Practices: skill at normal activities; evidence of values
- Values: intuition about what we like and don't like; brings purpose to practices
- Principles: domain-specific guidelines; bridge gap between values and practices
- Values
- Difference between what we think is valuable and what is actually valuable creates waste
- Most important: how individuals behave as part of a team and as part of an organization
- Communication
- Others probably already know the answer
- Motion without communication is not progress
- Simplicity
- What is the simplest thing that could possibly work?
- Simplicity only makes sense in context
- Feedback
- Use feedback to get closer and closer to our goals
- Generate as much feedback as fast as possible
- If team starts to ignore feedback, slow down
- Courage
- Effective action in the face of fear
- Do the right thing, regardless of fear
- Respect
- Care about each other
- Others
- Could be: safety, security, predictability, quality-of-life
- Principles
- Should always be able to trace a path from the work done to a customer need
- Humanity
- Basic needs: safety, accomplishment, belonging, growth, intimacy
- Balance personal needs with business needs
- Trust goes a long way
- Economics
- Don't focus on a "technical success"; worry about the business
- Generate money sooner, spend money later
- Avoid investing in speculative flexibility
- Mutal Benefit
- All activities should strive to benefit everyone involved, including you in the future
- Write tests, carefully refactor to reduce complexity, choose good names
- Self Similarity
- Copy structures that work in one area to others (e.g. apply TDD to an entire year, quarter, or week)
- Improvement
- Nothing is perfect, but it can always be improved
- Do the best today, to improve tomorrow
- Start right away and refine later
- Don't wait for perfection
- Diversity
- Multiple ideas about a design present opportunities, not problems
- Resolve conflicts productively
- Reflection
- Analyze how you succeeded or failed
- Make sure reflection happens after action
- Flow
- Deliver value in small and frequent chunks
- Opportunity
- See problems as opportunities for change
- Redundancy
- Problems should be solved in multiple ways; cost of redundancy is cheaper than having a disaster
- Failure
- "If you're having trouble succeeding, fail"
- Failure isn't wasteful as long as it imparts knowledge
- Fail, instead of talk
- Quality
- "Projects don't go faster by accepting lower quality"
- Pushing quality to be high often results in faster delivery too
- People need to do work they're proud of
- Use project scope for controlling a project's time and costs
- Baby Steps
- "Momentous change taken all at once is dangerous"
- Take many small changes frequently
- Overhead of small steps is smaller than aborting from a large change
- Accepted Responsibility
- Person directly invovled in the task should be responsible for it, from planning to practices used to testing
- Practices
- Practices are situation dependent; choose the best for yours
- Primary Practices
- Sit Together
- Decrease barriers to open communication
- "No matter what the client says the problem is, it is always a people problem"
- Whole Team
- Use whole, cross-functional teams
- Find ways to fracture a big problem into smaller ones to use multiple smaller teams
- Informative Workspace
- Should be able to quickly and easily understand what a team is working through some visual means (e.g. cards on a wall)
- Energized Work
- Recognize when you've started being unproductive, or started removing value from your work
- Pair Programming
- Respect cultures and personal space, hygene, etc
- "Ideally, emotions at work will be about work"
- Stories
- "Plan using units of customer-visible functionality"
- Try to estimate the development effort of a story immediately
- Balance cost of features with their value
- Don't plan by focusing on appeal; focus on the cost and intended use first
- Weekly Cycle
- Start week by writing tests for features to be developed
- Have deployable software by the end of each week
- "Shifting the start of the [planning] cycle [to after Monday] makes sense as long as it doesn't put pressure on people to work over the weekend"
- Quarterly Cycle
- Focus on themes and the big picture
- Good for team reflection on process bottlenecks, etc
- Slack
- Add minor tasks to planning that can be dropped if necessary so important commitments can be met
- "Meeting commitments, even modest ones, eliminate waste"
- Ten-Minute Build
- Continuous Integration
- Team programming is a "divide, conquer, and integration" problem
- The longer you wait to integrate, the more risky, unpredictable, and time consuming it'll likely be
- Test-First Programming
- Avoids scope creep
- Gives focus to the implementation
- If it's hard to write a test, you likely have a design problem (usually a coupling issue)
- Incremental Design
- Common wisdom: cost changes to an existing system increases with time
- May no longer be true; or there may exist areas where this is not true
- XP's practices may lower the cost of changes over time
- Introduce changes and implement at the last responsible moment
- "The most effective time to design is in the light of experience"
- Common wisdom: cost changes to an existing system increases with time
- Sit Together
- Getting Started
- "Change begins with awareness"
- "Metrics can lead to awareness"
- Change begins with yourself
- Corollary Practices
- Real Customer Involvement
- Make (visionary) customers part of the team and planning
- Build relationships and trust with customers with your transparency
- Incremental Deployment
- Make frequent, gradual replacements rather than singular, large ones (e.g. new systems, refactors) even if it means duplicating some services
- Working harder to prepare for something means they'll also be less productive after
- Team Continuity
- "Keep effective teams together"
- Shrinking Teams
- Alow teams to shrink and reform into bigger teams by not overloading them with work
- Root-Cause Analysis
- Eliminate both a defect's symptoms as well as its cause; never make the same mistake twice
- Add tests for the found defect
- Five whys questioning (ask "why" five times on the cause) usually ends up with the root people problem
- Shared Code
- Anyone can improve any part of the system at any time
- Code and Tests
- Maintain only the code and the tests, generate other documents from these two
- Single Code Base
- Only keep one version of the code
- Use configuration files / etc to maintain other necessary versions / builds
- Daily Deployment
- Negotiated Scope Contract
- Signing a series of small contracts rather than a large one decreases scope and risk
- Pay-Per-Use
- Force users to pay to use the software each time
- Real Customer Involvement
- The Whole XP Team
- Testers
- Help define what the acceptable functionalities are before they're implemented
- Interaction Designers
- Work with customers to write stories
- Architects
- Look for large-scale refactors, write system-level tests
- Keep complexity of architecture balanced with size and needs of project
- Project Managers
- Facilitate communication: internally, up and down, sideways, and externally
- Product Managers
- Picks stories and themes to work on
- A plan is what could happen, not what will happen
- Stories should be sequenced for business, not development reasons
- A project should be demo-able (for the most basic of tasks) in the first week
- Executives
- Articulate and maintain long term goals
- Health of XP teams:
- Number of defects found after development
- Time lag between initial investment and start of revenue
- Technical Writers
- Explains the features and creates closer relationships to the users
- Track which parts of the documentation is actually used by users
- Users
- Programmers
- Estimate tasks, break stories into tasks, write tests, write code, improve design
- Human Resources
- Evaluate employees based on the XP traits: acting respectfully, playing well with others, taking initiative, and delivering on promises
- Could use team-based incentives and raises
- Emphasis on hiring should be for teamwork and social skills
- Interview: let candidates work with the team for a day, pair program
- Testers
- The Theory of Constraints
- Theory of constraints: most systems' throughput are only constrained by one component
- To improve, find this constraint and improve upon it
- To find it, look for work piling up
- Focus on overall throughput, not micro-optimization
- The organization as a whole may change once the bottleneck shifts
- Theory of constraints: most systems' throughput are only constrained by one component
- Planning: Managing Scope
- Make goals and direction clear and explicit, so as to align them with others
- Analogy: grocery store
- Time is like money; can only spend as much as you have available
- Stories are groceries; their estimates are their costs
- Need to prioritize your needs and wants
- Estimates are uncertain; defer decisions to as late as possible to get the best info
- Three variables to manage projects: price, speed, quality
- Pick two and negotiate the third
- However, price and speed are typically decided already, so you can only change quality
- Lowering quality just leads to more work in the future; offloading responsibilities to someone else
- Missing variable: scope
- Acknowledge wishes, commit to needs
- Get more perspectives on your stories by laying them out by different criteria (e.g. riskiness, estimates, feature / tuning, etc)
- Estimate based on prior experience
- Get feedback and reiterate estimates as fast as possible (e.g. split one month into four iterations, or one week into five iterations)
- There is a real limit on how productive someone can be (creatively)
- Adjust your plans as soon as you have information from previous sprints (to the extreme: only plan to do as much as you did last sprint)
- If slipping in the middle of a sprint, try to realign with process changes (e.g. stopping work on less important issues) or communicate to customers what they'd like to have and replan
- Testing: Early, Often, and Automated
- Defects destroy trust
- Most defects are more expensive in the field than preventing them (in the long run)
- The sooner you find a defect, the cheaper it is to fix
- The more it costs to find and fix defects, the more defects will persist
- Frequent tests means that the person making mistakes is also writing the test
- The more stressed a team is, the more mistakes they make. However, with frequent testing, more stress results in more tests to verify behaviour
- Write tests first:
- Interfaces shouldn't be unduly dependent on the implementation
- Tests provide certainty, as well as a measure of progress (i.e. making failing tests run)
- Designing: The Value of Time
- XP is based on incremental design; design should be done every day
- Use prior experience to get to the next stage, while the software is still running
- Design just enough to get feedback, then improve the design with that feedback
- Real constraints on a design should become obvious after implementation; go after those in the next iteration
- "Design quality doesn't ensure success, but design failure can ensure failure"
- Can design by instinct, thought, or experience; if you don't need to spend more effort on the design, you can go with the lowest effort approach
- Design heuristic: "Once and Only Once"
- Data, structure, and logic should only exist in one place
- When confronted wtih a mess, improve only the pieces that affect you today
- Balance design with needs and relationships; delivery of functionality is what others want, design is not
- Simplicity of a design:
- Appropriate for the intended audience: should be understandable by those who need to work with it
- Communicative: the system's components should be clear to future readers
- Factored: little to no duplication of logic or structure
- Minimal: system should be as small as possible
- Scaling XP
- Number of people
- Factor large teams into smaller teams that work on smaller problems
- Initially solve a small problem with a small team and then add more teams, making sure to integrate frequently ("conquer-and-divide")
- Size of organization
- Make sure to maintain communication with others (in a format they're used to)
- Complexity
- Continue to delivery, but gradually reduce the complexity
- Even security audits should be done frequently and gradually
- Number of people
- Interview (with a manager using XP)
- Train one team, get them to master it, and then let that team teach another team
- Creation Story
- Consultant's job: find the person who has the knowledge (of the problem) and convey that knowledge to the person with power
- Taylorism and Software
- When picking a name, it helps if the opposite of the name is unappealing
- Micro-optimization doesn't always lead to macro-optimization
- Separating quality from engineering allows engineering to defer quality responsibilities to another department
- Toyota Production System
- "If you eliminate enough waste, soon you go faster than the people who are just trying to go fast"
- Make the quality of the production line good enough that you don't need to check quality later
- Using a part immediately (from another producer) gives you information about whether that producer is functioning correctly
- "The greatest waste is the waste of overproduction"
- E.g. large requirements documents, elaborate architectures, stale code, stale documentation
- Applying XP
- "The waste [in software development] is rooted more in what we believe and feel than in what we do"
- "Adopting" a process doesn't solve your problems; you solve your problems by tackling them with the help of the process
- Start changes with yourself, lead by example
- Only when you know how to improve, do you improve
- Practices that work for one team might not work for another; some practices contain hidden dependencies or a team might not be ready for them
- Conditions that can facilitate change:
- Aligned values: people are willing to accept the changes
- Pain: recent pain makes everybody more willing!
- Purity
- Saying your process is "x" sets other people's expectations of how to communicate and work with you
- The Timeless Way of Programming
- Keep in mind the needs of users, sponsors, but also developers
- Focus on integrity, balance, and harmony; hold yourself to high quality standards
- Believe in your process and try to make deep changes
- Foreword
- What matters is building the right product fast
- Invest in people so they can work quickly and effectively without formal process
- Preface
- "Planning is not about predicting the future"
- Development may not go as according to plan, but without a plan it will go awry
- "Keep the human beings focused, happy, and motivated and they will deliver"
- Why Plan?
- Business and software changes too quickly for plans to be able to perfectly predict the future
- "Plans are about figuring out a likely course of events and the consequences of the inevitable changes"
- "The more obvious it is that you should do something, the more important it is to ask why"
- Before planning, figure out why you think you should be doing this task (and how you can measure its success)
- Planning is useful because:
- We need to be doing the most important things
- We need to coordinate with others
- We need to understand the consequences on the first two when unexpected events derail our plans
- In the face of unexpected events, make sure to adjust our plan and coordinate with others
- Coordination requires that you know where you are in your own plan
- Set up clear milestones that anyone, including the customer, can understand and trust
- Don't plan to control events; you can only control your reaction to events
- Don't pretend that a plan is on-track when it's not; this is only trying to turn reality into an illusion (requiring more effort and inflecting more pain later)
- "If things seem to be going exactly according to plan, that's usually a sign of trouble"; expect plans to change
- Fear
- To develop effectively, we must acknowledge our fears of things going wrong
- "Unacknowledged fear is the source of all software project failures"
- Don't build useless political structures and processes aimed at protecting yourself (and then hide behind them)
- Aim to work with each other and help out as needed (including customers)
- Driving Software
- "Driving [software] is about making lots of little course corrections"
- "We are trying to maximize the benefit of a process," not trying to hit a target
- Continually assess your direction frequently, compare it to your ideal direction, and course correct
- "If you get lost driving, it isn't the car's fault, it's the driver's"
- Balancing Power
- Balance software decisions against business decisions
- The business people should be making business decisions
- Deadlines
- Feature scope
- Feature priority
- The technical people should be making technical decisions
- Development estimates
- The business people should be making business decisions
- Create a set of rules that encourage technical people to make technical decisions and business people to make business decisions
- Have a customer, represented by one voice (e.g. the product manager), that is a part of the team
- A good customer (representative):
- Understands both the domain and how it works
- Understands how software can help in the domain
- Delivers value regularly, bearing on the side of too little than nothing
- Makes decisions about what's needed now and what's needed later
- Accepts ultimate responsibility of the project's success or failure
- Asks, at all times, "What is the most valuable functionality to have next?"
- Never slip a deadline! (It's a slippery slope)
- Always provide a little functionality each release, and release to real customers as often as possible
- Balance software decisions against business decisions
- Overviews
- Recognize that there are two cycles to accomodate and synchronize: the business cycle and development cycle
- Any activity that goes between the engineering of a release and then delivering it represents a risk, so each development iteration should result in a production-ready system
- The business side determines the most valuable tasks to work on at the start of an development iteration
- Small iterations mitigates implementation risk by allowing plans to be affected as quickly and as visibly as possible if something goes wrong
- Too Much To Do
- When you're overloaded, give yourself less things to do (rather than trying to get more time, which is impossible)
- "We always ate Korean; it seems to be good for the brain"
- Not having enough time breeds hopelessness, which in turn, breeds frustration mistakes, burnout, and failure
- Having too much to do breeds hope, we can:
- Prioritize what to do
- Reduce the scope of some tasks
- Delegate tasks to someone else
- Four Variables
- Interrelated variables for projects (changing one affects the rest):
- Cost
- Quality
- Time
- Scope
- Delayed reaction after moving one variable
- Cost
- People: non-linear effect and long delay
- Brook's Law: "Adding people to a late project just makes it later"
- Tools: some delays
- Don't be afraid to spend money to keep motivation up; well-motivated people are well worth it!
- Try your best to keep overtime low, make your employees do something else if you have to!
- Results in lack of motivation, bugs, mistakes, etc
- People: non-linear effect and long delay
- Quality
- External: how the customer perceives the product
- Internal: the quality of the system's internals
- Dangerous: don't let this drop; you might get a small increase in speed at first but it'll come back as a huge decrease later
- "Nothing kills speed more effectively than poor internal quality"
- Push non-functional requirements (e.g. performance, aesthetics) to scope
- Time and Scope
- Scope is easy to increase without realizing the ramifications while time is hard to see
- "Every time you add scope, you [should] immediately see the effects on time"
- Short iterations force you to reconcile the consequences of changing scope with time
- Usually best to assume cost and quality and fixed levers
- Planning should:
- Preserve developers' confidence in completing the plan
- Preserve customers' confidence in getting value
- Be as efficient as possible
- Each iteration has stories, costs of the stories (their estimates), a budget (in time), and constraints (business constraints)
- Costs on stories can go up or down depending on their actual complexity, and stories may have to be delayed to meet the budget (but never modify the budget!)
- Interrelated variables for projects (changing one affects the rest):
- Yesterday's Weather
- Assume you'll do as much this week as you did last week
- Avoids habitually overestimating a team's abilities
- Forces teams to focus on finishing at least some features rather than having them all "in progress"; imagine telling a customer they can have 0 features next week
- If you commit to too much, everyone will know the plan is doomed and will do everything they can to protect themselves and hide their failing schedules
- Assume you'll do as much this week as you did last week
- Scoping a Project
- First estimate coarsely, to get a feel for the project
- Does it make any sense (in regards of time, scope, quality, and cost)?
- It's OK to guess, and leave lots of padding
- To make the initial estimate:
- Break the project into smaller pieces
- Estimate each piece (don't get too into the details!)
- Defer less valuable pieces
- Make sure to separate actual business functionality from "context" items (e.g. performance, aesthetics, reliability)
- Turn generic ideals about the product (e.g. "easy to use") into actual functionality (e.g. "personal profiles")
- You don't have to accomplish everything in this first plan, priorities and requested features will change!
- "A pep talk is no substitute for a plan that everyone believes in"
- First estimate coarsely, to get a feel for the project
- Release Planning
- Synchronize the date and the scope of the project with the business (may be external factors, e.g. run way)
- Allocates user stories to releases (and development iterations)
- Effort between the customer and the developers; customer drives and developers help navigate (business prioritizes, technical gives estimates and technical risks)
- Does not need to be stable; it's just a snapshot of of the current view on what is (and should be) being built
- The further ahead the plan, the less accurate it will be, so plan only a few iterations and releases ahead
- Evolve the infrastructure as you build functionality, not ahead
- Use past performance ("velocity") to determine how much you can do
- Writing Stories
- "A user story is a chunk of functionality that is of value to the customer"
- Emphasis on the simple and providing value
- Make them understandable to the customer (the end ones)
- Should be short, and act as a contract for the developers to talk to the customer during implementation
- Only add more details when you actually need them (usually when you start implementing)
- A few stories should be implementable in an iteration, otherwise they're too large
- Developers should not be writing stories (though they may suggest them); the responsibility lies solely on the customer
- Customers write the stories, developers estimate them
- Estimate sooner than later; this allows you to refine stories so that they become estimatable (e.g. turning "make it easier" into "make it easy to start the process," etc)
- Stories should be independent, should be able to be completed in any order
- Stories should be testable (and an implementation is only complete when it successfully passes this test); automated is better
- Avoid lumping too many things in a story: not only is it not simple or clear, it bundles different stories' priorities together when they may be separate
- When a split is needed, split tasks along priorities; the customer should drive this
- Developers and the customer should iterate on this, to make sure the stories written by the customer are small enough and estimatable
- Exploratory programming is sometimes a good way to investigate and come up with an estimate
- "Writing stories is not the point. Communication is the point."
- Start developing when the most important stories have been identified and enough work is available for several months (so the custom can make choices)
- Estimation
- Use a comparable you've previously done
- Keep it simple
- Use what happened in the past
- Learn from experience
- The team should estimate, and favour optimistic estimates during disputes (to avoid the team from getting too optimistic; those who get burned by optimisim will probably learn)
- "Estimates are not commitments"
- Don't predict the effect of external influences (e.g. changing team size), measure it and react accordingly
- Instead of worrying about short term data fluctuations in real, calendar time, use ideal time (assumed 100% productive) to estimate tasks
- Encourage better estimates, and make sure to record your actuals!
- Ordering the Stories
- Do the stories that have the highest business value first!
- Don't worry too much about technical dependencies
- Focus on minimum technical commitment for stories
- "Most ordering dependencies are false"; you probably only think you need them when you actually don't
- Don't look for dependencies to order things, look for business value!
- Business value is determined by the business guys
- Even if you can't release, people are happy if you've done the most important things first since it looks like you're listening and incorporating their feedback
- Technical risk: "When developers feel nervous [after thinking about a story], everyone should listen"
- "Programmers want to do higher risk stories first"
- The programmer should make the risk visible, and the business side should factor this in to their ordering of items
- Keeping technical commitment low should prohibit risk from impacting too much
- "Not all usable behaviour has to be usable by the user [at first]... the key criteria is that it is testable"
- Release Planning Events
- Requirement creep is OK since you're iteratively defining the requirements (rather than all up front)
- Build a new plan when:
- You have too many deferred items
- Your development velocity changes
- To rebuild:
- Re-estimate stories (you have more experience now than before)
- Re-prioritize and pick items for release
- Should be done every 3-4 iterations
- The First Plan
- "The first plan... is always high on expectations and usually low on delivery"
- Guessing velocity:
- Measure progress while writing and estimating first stories
- Story estimating:
- Start with the ones you're most comfortable with estimating first
- Use data from other teams, but don't get attached to comparing your results with theirs
- Iteration length:
- 2 - 3 weeks generally
- Getting development started
- Mob program for a bit to evolve the design and begin initial implementation
- Get some zero functionality tasks (e.g. infrastructure) ready
- Release Planning Variations
- Shorter releases: every iteration (or story) becomes a release
- May result in optimizing short term features over long term ones
- Longer releases
- First, question if you really, really can't make it shorter
- Create internal releases that are shorter
- Smaller stories
- Divide stories up; offers finer control for more cognitive load and lower flexibility
- Shorter releases: every iteration (or story) becomes a release
- Iteration Planning
- The iteration plan is the developers': they decide how to do things and in what order
- Break down stories into tasks, have developers sign up for tasks, and then estimate
- Report progress mid-iteration to adjust scope and get a sense for what will actually happen
- Deal with dependencies inside the iterations!
- Never slip release dates; defer functionality (let the customer choose what he cannot live without)
- Let the customer decide if a release date can be moved back
- Timeboxing is not about time; it's about functionality trade-offs
- Iteration Planning Meeting
- Summarize stories to get everyone empathetically on the same page
- Break down stories into tasks each taking a few days
- You never plan exactly when these tasks will be finished; dependencies will usually sort themselves out (or people will)
- Write tests for the tasks
- Try not to set priority of tasks solely by technical reasons (slippery slope)
- Absolute values of velocity =/= productivity; it's more useful as a predictive tool
- Don't use other people's work to estimate your tasks
- Tasks are only complete when tests are written and passing
- Let customers remove stories and functionality when you come up short on capacity
- The sum of tasks is more granular than their parent story's estimate; they provide better (and more accurate) information
- Tracking an Iteration
- A designated person should ask each developer how many days they've spent on a task and how many days are left
- If the actual and estimated (to finish) days left aren't similar, adjustment might be necessary
- If someone reports not being able to spend as much ideal time working on their tasks as usual, maybe they need some help (with e.g. clearing something up, getting familiar, etc)
- Let the customer decide when a task is complete (if possible)
- Known bugs are the customer's choice to accept
- Let customers demo the features (if the developers don't mind); this ensures the customer knows what's going on
- If you know you probably can't finish something, let others know (they might be able to take over that part or help significantly)
- If the team absolutely can't handle it, the customer needs to descope
- Don't use overtime to catch up on a release
- Stand-up Meetings
- Let everyone know what's going on, and what's not
- What you did yesterday, what you'll do today, problems, annoucements
- Communicate problems, not solve them
- Visible Graphs
- "You can't manage what you can't measure" -- Denning
- Don't let the measurements become the ends (keep them as means!)
- Use intuition:
- Smell a problem
- Devise a measurement
- Display measurement
- If it doesn't go away, keep devising measurements
- "Many an insight comes when idly staring at a graph when you're half doing something else"
- Dealing with Bugs
- There will always be defects
- Remove the emotion; consider changes to be enhancements
- Customer decides if time should be used towards fixing bugs vs new features (business decision: cost of bug vs value of feature or earlier deployment)
- Treat the defect as a story if it's big
- Prioritize small bugs before iteration planning
- Rotate some development members to production support
- Customer makes the call on what's a critical bug that has to be fixed now
- Changes to the team
- Scale teams by dividing by half after ~10 people
- Could try putting management tasks for each iteration to give visibility and maybe let others try
- Tools
- "Communication is more important to success than whizbang"
- Problems in project management:
- Tracking data
- Maintaining communication
- Keeping everyone honest is usually the most difficult problem
- Everyone understanding the current state of development is better than keeping whatever tool you're using current
- Business Contracts
- Don't put the interests of the supplier against those of the customer (work together instead)
- Outsourcing: try to let the scope be visible (and adjusted) within quick iteration cycles (e.g. by applying this book)
- In-house: find a good customer who can speak for business
- If you have a committee of customers, make sure their individual goals are aligned
- If you have a number of product managers, make sure they reconcile their priorities and present themselves as one voice to development
- Red Flags
- "Slow down until you're under control, then speed up again"
- Missing estimates
- Don't guilt trip yourself into overestimating
- Iterations should be finishing comfortably
- Make better estimates (and by adjusting with prior work)
- Customers won't make decisions
- Find a better customer, or reassure them that the process is iterative and they can fix any mistakes later
- Defect reports
- Slow down and fix defects until you stop seeing them (or see less of them)
- Not going end to end
- Don't stop developing if you get uncomfortable; do your job
- Failing builds
- Make your production environment similar to your test environment
- Customer won't finish
- Make better release plans that are visible by upper management
- Your Own Process
- Start by trying the whole system (synergies), and then adjust with time through each iteration by experimenting
"If you’ve got a dream, you have to try it; you must get it out of your system. You will never get another chance. If you want to change your life, do it"