As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 11 years ago.
To quote wikipedia:
Scrum is facilitated by a ScrumMaster, whose primary job is to remove impediments to the ability of the team to deliver the sprint goal. The ScrumMaster is not the leader of the team (as they are self-organizing) but acts as a buffer between the team and any distracting influences. The ScrumMaster ensures that the Scrum process is used as intended. The ScrumMaster is the enforcer of rules."
Working on this basis, and the fact that most businesses are running 2-3 projects at a time, what actual work tasks does a SM do to fill a full time job? Or, is it not a full time job and that individual do other things such as development, sales etc?
Do any SM's out there have anything to share?
Unfortunately we don't have the luxury of having dedicated scrum masters. I am also a team leader and senior developer which more than fills the day.
I typically am on Stack Overflow all day. Oh, and I try to co-ordinate lunches.
The key to the ScrumMaster role is to remove impediments.
The ScrumMaster/ Iteration Manager
Builds the Release Plan
Builds the Scrum/ Iteration Plan
Plans and hosts the
Scrum/ Iteration Planning Meetings
Show & Tells
Release Planning Meetings
Retrospectives
Owns the blocker board and actively works with the team to identify and remove blockers
Updates the team WIKI
Updates Big Visible Charts in the team room including the story card wall
Participates in the daily standup
Participates in the daily Scrum of Scrums
The ScrumMaster/ Iteration Manager is also the sheep dog, that is they protect the team (herd). Finally, the ScrumMaster/ Iteration Manager is the point of contact for the team to external resources but primarily the Project Manager.
"acts as a buffer between the team and any distracting influences"
That is a full time job. There are a bunch of people who would love to get information from the team and it is the SM to handle those questions. To do that job well, it is important to be proactive, not reactive. Therefore they should be keeping all the wheels running smoothly. It is an amazing transformation when the SM is working well.
I think there will be as many answers to this question as there are people to answer it. On a small team with dedicated people who mostly know what they doing, the role of SM is almost invisible; on a larger team trying to cope with vague requirements and power struggles the SM will be highly visible and probably never have a moment to themselves, as they will become the lightning conductor for all the frustrations of the team (and stakeholders outside it).
There's no substitute for knowing what you want to achieve and having a small team of people who know how to achieve it. If you have that, and you "adopt SCRUM", you will probably be convinced quickly that being a Scrum Master is easy. But if instead you have a big mess of a team, and an undefined goal, and a lot of political fighting going on, and you "adopt SCRUM", you will probably come away thinking that being a Scrum Master is a full-time (perhaps impossible) job requiring a combination of very rare talents. Most real teams are probably somewhere between these extremes.
Please note: this question and answer is over twelve years old. The consensus understanding of the role of scrum master has moved on massively since then and so I no longer view this as a valid answer to the question, let alone one worthy of being the accepted answer. By all means downvote it. Beyond that, pay it no heed.
The Scrum Master will do things like ensuring scrums occur, organising sprint planning meetings, retrospectives etc. Also (s)he will be able to explain to management what the team is doing and why the team members cannot be poached off onto other projects until the sprint finishes. Beyond that, there aren't really any defined tasks for the Scrum Master. So one person should easily be able to be Scrum Master for 3 teams, and still have time left over to either do management type jobs (holiday requests, procedures, attending boring meetings with directors or whatever), or be free to contribute to the development resources of the team.
While ScrumMaster is a role within the Scrum framework, the individual fulfilling that role must be a member of the Team. In Scrum, Team members should at all costs be full time. Team members should be able to pick up tasks on the Sprint backlog. They might be development tasks, testing tasks, configuring the CI server tasks, etc... If you can't contribute to the burndown then why be on the team? Buggering off and joining another team is the last thing any self respecting ScrumMaster should do. ScrumMasters should be servant leaders that are embedded with and dedicated to their Team and product. ScrumMaster is a role on a Team, not a job title. I disagree with those that think you can be a ScrumMaster on more than one project at a time and still be world class. The fact is, that's just not Scrum.
First and foremost: remove impediments.
It is best if a Scrum Master is dedicated to one team, so that impediments are removed as soon as possible. Some of this can be done proactively, for example by pushing the PO to analyze certain stories better for the next Sprint.
If there is extra time available it is convenient if the SM has some skills that can make him function as a developer or tester on the team. I've seen good result with SM's that delegate as much as possible to a (classical) project manager and focus on development most of their time.
To make a long story short, the Scrum Master is responsible for making things happen. And in practice it is often the case that the Scrum Master is actually a project manager in disguise. At least that's the case in my company.
Working on this basis, and the fact
that most businesses are running 2-3
projects at a time, what actual work
tasks does a SM do to fill a full time
job?
Anything within their skillset to help the Team achieve the goal.
Or, is it not a full time job and
that individual do other things such
as development, sales etc?
ScrumMaster was not originally intended to be a full time job. ScrumMaster is a role fulfilled by someone on the Team. That team member is dedicated to the product full time. So, when he\she is not doing ScrumMaster duties they default back to burning down tasks on the Sprint Backlog.
Everything and anything that developers need to keep being productive. Order pizza. Go talk to admins, management, other teams. Do bureaucracy kind of stuff. Fix the build server if no one else's available.
The key word here is that a Scrum Master's role is a facilitator's role. And as someone rightly mentioned up there his most important job is to ensure seamless distraction free environment for his team, which means removing impediments, making sure his team has whatever they need at all times. Scrum master is a link between the Product team and the Development team. The decision making is done by the TEAM and not Scrum master.
It is bad idea to share one Scrum Master between multiple teams as requirement of one team may be an impediment for the other team and hence defies the whole purpose of a Scrum Master.
Also it is very dangerous to have your Manager as your Scrum master as the pressure of delivery on the manager may force him to micro manage which is a killer for any scrum team.
Other than the regular stuff which is
Arrange Sprint planning and retrospectives
Facilitate daily standups Arrange
Demo's at the end of sprint iteration
Address team's concerns mentioned at the standups
A few important things that a Scrum Master has to manage on a day to day basis is
Foresee and remove any distractions for the team before even it hits the team.
Encourage team to communicate more
Maintain constant communication with product team to check what needs to be done in
preparation for future sprints
Make sure the team follows the processes they have collectively agreed upon as sometimes
during sprint busyness some processes slip through the crack
Constantly find ways on how to improve the processes followed by the team
Most importantly a Scrum Master has to standby and support his team.
All this work takes up a lot of time and does require a dedicated Scrum Master who performs no other role.
Scrum Master is like the mother bear for the team. They look after the team's health (project wise), protect them from pesky outsiders and remove any obstacles for the team. I play ScrumMaster for my team but I am also a development lead (for the same team!) who takes part in technical discussions, design discussions, coordinating between the developers and QA on our team(if they arent already doing it themselves). I do try and take on actual development tasks to burn the chart down when time is available.
Isnt it extremely distracting for the ScrumMaster to play that role in multiple teams. God I would find that confusing. Which impediment is blocking which team again?? Wait who was working on this task??
A Scrum Master role implemented correctly, is invaluable to the Project and should not be look upon as a Part time role. The most important aspect of the role is to act as an obstacle remover for any queries raised in the Scrum meetings by the Development Teams. A Technical Scrum Master (which is what most SMs tend to be) should not be a Developer on the team, but should be able to advise on design and solutions (an extension to pair programming if you will).
They are responsible for updating the ProductBackLog (stories should be created by the business), SprintBackLog and BurnLog and for liasing with the business and IT Management on progress. They also manage a SpikeLog for any items that require investigation that may evolve into Stories (again driven by the business).
As drivendevelopment implies, the ScrumMaster is a full team member and thus should be full time. I generally treat my role as "ensuring the team functions as a well oiled machine", which can have a number of meanings at different times. Frequently, a SM spends a lot of time facilitating the team's interactions with people outside the team, especially those related to business analysis and stakeholder expectations. Beyond that, it is a matter of meeting the mechanical items listed by Cam and looking after the physical and emotional state of the team.
Related to one of the earlier answers, one of the fundamental aspects I insist on is that no member of the team is a direct report to me, nor to each other. This precludes things like vacation time, expenses, etc from being part of my job, but goes a long way towards not cluttering the trust relationship that must exist.
As generally understood priority # 1 on scrum-master list is to remove impediments as reported by team. But this should not stop here, he should constantly look out for potential impediments.. and more importantly impediments that are there but not yet identified. Ken said Impediments are opportunities. So scrum-master should avail these opportunities all day along to bring his team(s) to hyper productivity.
Ultimately purpose of scrum is to bring success to projects. Purpose of having a scrum-master is to ensure that scrum succeed in fulfilling purpose of scrum. Now to to fulfill purpose of scrum-master, he/she must think & act at strategic level as well. This is full-time job.
Related
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 2 years ago.
Improve this question
I am just curious, since for a scrum-enabled development process the Team Leader (TL) role is merged with Scrum Master (SM) role and the team is supposed to be self-motivated, self-organised and self-driven.
Without a Team Leader (and architect!) inside a team - who is making decisions on a future-proof ways on the implementation? Say, which libraries to use, client data-handling, etc.
Based on the scrum concept, such decisions should be made by the team collaboratively or by the individual team member. Chances are they will not be qualified/experienced enough and the decision could cost millions in the years to come (even one or two years perspective).
How in scrum concept is this addressed, please?
Thanks
Scrum is only the delivery process. Technical decision making should be done by relevant responsible people. Those decisions may come after discussions and team meetings. Team members can also participate to these meetings if required. Role of the scrum master is to make sure the backlog items taken into the sprint is delivered. It has nothing to do with technical decision making. (Scrum master is responsible to remove the blockers and make the team move towards the sprint goal) The team can allocate time for architectural decision making or design/redesign process when they take in a task during the planning meeting.
For example, the TL may have a 4 hour task in scrum board to do the design. If this task involves a meeting with the architect, and the architect is offsite, scrum master should take necessary actions to schedule a meeting and make sure this blocker is removed. May be he can arrange a call between them.
Also it is wrong to think technical leader should be the scrum master. Those are two separate roles and may or may not be the same person.
In Scrum there are three roles defined:
Scrum Master
Product Owner
Development Team
The Scrum Master helps the team to follow Scrum and to remove impediments. The Product Owner looks after the backlog of work and prioritises things.
The Development Team does everything else. There are no defined roles in the Development Team, but instead we have capabilities.
For example, the Development Team has a capability to do development. It also has a capability to do testing.
There is nothing to stop a Development Team having a capability to do architecture. For example, it might have a team member who is from an architecture background or perhaps several team members have architecture experience.
As for the Team Leader role, there is no need for that as everyone in Scrum can act as a leader. For example, if somebody in the team is really experienced in database work, they might show leadership when the team is working on database work items.
who is making decisions on a future-proof ways on the implementation?
The team does.
An experienced Scrum team will get together regularly to talk about 'the big picture'. They will think about the future of the implementation, about the way its architecture, etc.
which libraries to use, client data-handling, etc.
Who has experience making decisions on which libraries to use? They might be a good person to suggest an approach on which libraries to use. The team can then discuss it and agree on an approach. The same goes for client data-handling.
This is collaboration. Everyone has a voice and everyone is a potential leader.
Chances are they will not be qualified/experienced enough and the decision could cost millions in the years to come
If the team is concerned that they don't have enough experience to make important decisions then they should raise that as an issue. Some possible solutions include:
Get additional training for team members
Bring somebody into the team with more experience
Use communities of practice to get advice
The Agile Manifesto directly addresses your question in the Principle that mentions self-organizing teams: "The best architectures, requirements, and designs
emerge from self-organizing teams." Relative to the other well-known Agile methods, Scrum perhaps does the best job of trusting "motivated individuals" (per another Principle) to do so using what scientists calls the "group mind." My coaching practice is based on my review of scientific studies related to teamwork. These have shown repeatedly that cross-functional small groups with diverse backgrounds will out-perform a single expert in decision quality over time. Scrum facilitates this by making people plan everybody's work together and holding them accountable to the group through the scrums and Demos.
That doesn't mean you can't have a technical expert like an architect on the team or providing input to the team. I agree with Leni that it's best not to have the technical expert be the SM, as that consolidates too much social power in one role and places further psychological blocks to true self-organization. (Having a full-time SM prevents real self-organization from the social psychology perspective, so I have members rotate the role.) But those experts can serve as advisors by coordinating bottom-up standards development across teams, providing input into user stories, sitting in on various teams' Planning Ceremonies and Demos, etc.
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 5 years ago.
Improve this question
We are acting Scrum in our department now. But the up level management structure is traditional, such as Project Manager(PM), Development Manager(DM), Team Leader(TL) and Test team leader(TTL).
Team Leader act as a Scrum master, he controls all the things in our team: communicated with PM/DM/TTL, development management... Our PO's responsibility is just maintaining PBL.
Our managers and team member are accustomed to the traditional management type, they do not care Scrum, and they said some Scrum rules are hidebound.
I act as another SM, I want to change the current status.
But I haven't any headship, just is an ordinary developer in our department. Does anyone has this kind of bother too?
Thanks in advance!
I heard a lovely saying once and can't remember who said it. "They want Agile but they don't know what it is - so we give them Agile but we don't know what they want."
It sounds as if this is happening in your company. Someone, somewhere wants the team to use Scrum, but it's not the team.
That must be a difficult job for an SM, especially if you're doing it unofficially! There are some things I can suggest for you. First, learn some basic coaching techniques: positive language, GROW framework and giving and receiving feedback. This will give you some additional tools which are outside of Scrum and support someone in a leadership rather than a management position (even an unofficial SM can become a leader).
Then, don't worry about the actual practices. If someone has mandated Scrum then the team will be forced to do this anyway. Instead, concentrate on the values and principles of Scrum - particularly collaboration, communication and transparency. Help the team to work with each other instead of being silo'd away. You will have to be an example for them. Don't mandate pair programming, but do go over and pair. Don't mandate stand-ups, but do have conversations first thing in the morning and draw in as many people on the team as you can. Look at the principle of "Continuous Improvement". Learn how to do root cause analysis and the 5 Why's so that the team can understand better why things are hard and take action themselves.
I also recommend Mary Lynn Manns and Linda Rising "Fearless Change". This will help you to work out who else could help you.
Finally, I will echo #sjt. Don't commit Scrum suicide. However, if it's something you really want and your company aren't doing it in the right way, don't be afraid to look elsewhere. Learn some of the fundamentals, practice TDD on your own and find a new job.
Whatever you do, good luck! The first step to change is desire.
If you don't have buy in from your other developers its not going to work. Period.
Scrum requires a heap of discipline, especially during the early adoption phase.
I wouldn't be bothered that management don't care for it. If you're free to do the work of developing the software, and all they care about is results, then it shouldn't matter if you happen to have a 10 minute stand up each morning, and plan small chunks of the work into manageable bits, as long as you're hitting the targets they want you to hit.
If you're team isn't on board though, you're going to have a really hard time getting it working, and it will probably fail and cause more impact that not having tried at all.
If you can try to start it in a small project, with a few developers who are on board with the idea, then you can report back to the rest of your development team on how you found it works, what were the benefits and what were the negatives (reflecting is after all an important part of Scrum).
If you want to get your management on board, you might find that after doing a few projects this way you're much better at estimating the time it will take to develop the requirements you've been given by the PMs, hopefully being able to hit deadlines with more accuracy.
Remember, the PMs and BAs can still work in their normal way, once they've handed requirements to you, you're able to build them using Scrum. Its not ideal, but short of having the buy in of everyone, and the ability to speak directly to users and get them to help write user stories, it will be the best you've got.
When asked to estimate the time it will take to complete the project you can apply Scrum techniques. You can break the specifications down into smaller chunks, group them into sprints and develop them accordingly, hopefully yielding better results.
"I act as another SM, I want to change the current status"
Well, that's a good start right there, wanting to change the situation. Although I must say that without the management buy in, it will be tough. Try and arrange an experienced Scrum Speaker or Agile Coach come and do a presentation or workshop at your company which involves all the upper management. Once you have the management believing in Scrum, it will be all downhill from there.
"Team Leader act as a Scrum master, he controls all the things is our team"
This goes against the Self Organized and Self empowering Teams principle in Scrum. A good Scrum Master would empower the Team in a disciplined fashion within the Scrum Rules, to that appropriate level that, the Team should be able to run on it's own. One suggestion is that the Team Leads need to have a different mindset when working as a SM and different one while working as a Senior Developer, there are no Team Leads in a Scrum Team, only Scrum Team members. You cannot assign true leadership, that is a mutual role which can be earned by creating a reputation of helping others and mentoring others. Have them spit time between SM and development duties 30%, 70% or 50-50 or whatever you find appropriate. Command and control could be counter productive for the Team.
Our managers and team member are accustomed to the traditional management type, they do not care Scrum
A Scrum Trainer had told once told me, "Do not commit Scrum Suicide". If your managers do not care about Scrum, don't get fired trying to convince them. Whatever methodology you guys might follow or "not" follow, you have to realise that all this is a business. Your pay check is dependent on your boss's approval, if your boss or manager does not care about Scrum, then don;t do it. If they care about waterfall, Switch to it, do it like you care, but don't do Scrum halfway and call it scrum.
What has worked for me in the past is to identify and communicate pain-points. Certainly, you should never do something because Kent Beck told you to, especially something that will just get you fired. However, some smart people worked at figuring out a set of practices which is cohesive, and divergence from these practices almost always leads to pain points.
As just one example: if you do Scrum where you have a requirements iteration, a design iteration, an implementation iteration, and a testing iteration, this in theory could work but in practice never does. (When it does, it ends up being Waterfall, and the "iteration" notion becomes meaningless.) Pointing out to your boss that you learned something about the requirements while QA was testing might help him realize there's value in getting QA involved in requirements. Or finding risks in the software design by doing a small prototype may help to show why it might help to collapse the design iteration.
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
This question does not appear to be about programming within the scope defined in the help center.
Closed 5 years ago.
Improve this question
We use Scrum. We are experiencing problems during sprints when we find the user stories are not sufficiently granular to capture the effort required to complete the sprint.
In particular, we find that we are supplied with UI wireframes that contain much more complexity than the original stories would have implied (e.g. functionality duplicated for usability reasons). This leads to the burndown chart looking like everything is completed on the final day of the sprint.
We spend the Monday at the start of each 2-week sprint going over the stories as created by the project team, during which time we typically refine the stories a little and break them down into tasks, estimating the hours for each to create the burndown chart. During this day it doesn't feel like we have time to meaningfully improve the quality of the stories.
How best to break the cycle of incomplete / insufficient stories for our sprints?
Is this a failure of the project team to nail down the stories sufficiently at the outset, or should we (i.e. the dev team) take some of the responsibility?
So are you saying that:
Customers/users talk to project team
Project team writes stories and creates wireframes
Development team breaks down stories into tasks and estimates
Is there a possibility of the development team actually talking to the customers/users? User stories are sometimes seen as a way to kick off a conversation, as opposed to requirement documents/specifications.
EDIT: Some links:
A user story is to a use case as a gazelle is to a gazebo
Six Features of a Good User Story - INVEST Model
The Customer is Always Available
EDIT: Martin Fowler made a blog post yesterday on ConversationalStories that covers this far better than I did.
Are you running sprint retrospectives? At the end of the retrospective you should have high priority actionable items to improve on what happened in the previous sprint. The same things shouldnt be going wrong repeatedly.
Is your product owner accessible during a sprint? If not you may need to add extra to any estimation as the detail of a user story is incomplete.
#Pascal suggestion to dedicate 5% of your sprint to product backlog grooming is a good one. This should enable the user stories to be in a more detailed place before the start of your sprint.
We spend the Monday at the start of
each 2-week sprint going over the
stories as created by the project
team, during which time we typically
refine the stories a little and break
them down into tasks, estimating the
hours for each to create the burndown
chart. During this day it doesn't feel
like we have time to meaningfully
improve the quality of the stories.
It sounds like this is your sprint planning session, do you have control over what user stories you are commiting to complete during the sprint? How can you commit if you dont have sufficient detail?
This takes you back to having a good retrospective and solving the issues raised.
How best to break the cycle of
incomplete / insufficient stories for
our sprints?
Retrospectives, planning, backlog grooming.
Is this a failure of the project team
to nail down the stories sufficiently
at the outset, or should we (i.e. the
dev team) take some of the
responsibility?
Its the responsibility of the team as a whole. Finding blame isnt going to give value, but everyone taking responsibility will give everyone a chance of completing the project succesfully.
Maybe during those Monday morning planning sessions you can involve whoever is writing the user stories / wireframes and work together to find out what detail is missing from them, what detail would make your estimations easier and more accurate. Maybe a template of what they should include could be drawn up.
We had (and continue to in some respects) this same problem. I think this problem is a lacking of upfront analysis and a lack of devs spending enough time estimating a user story.
You might start with a story like:
As an administrative user I can create a new widget.
OK, what does that mean? After some analysis, it might mean:
As an administrative user I can create a new widget in created status with complex data validation errors.
So after that a listing of fields, how big, and what the required fields are for saving to the database. A basic UI mock up would be nice as well.
Another user story for the next sprint might be:
As an administrative user I can edit a created widget and correct the complex data validation issue to move the widget to completed status.
Then list of the complex validation rules.
We spend the Monday at the start of each 2-week sprint going over the stories as created by the project team, during which time we typically refine the stories a little.
At the start of the sprint, the stories should be READY. If you need to refine them a bit, I think that you (the dev team, the ScrumMaster, the project team) should do that a bit ahead, during the previous sprint.
How best to break the cycle of incomplete / insufficient stories for our sprints?
You are maybe underestimating stories or they are too big and too vague. In both case, this sounds like an estimation problem and a good way to improve is to reduce the size of stories. To work on this issue, you could dedicate some time (e.g. 5% of every sprint) to Product Backlog Grooming in order to prepare the most important stories and reduce their size by putting them on diet if required before the next sprint. And this will actually make the sprint planning meeting smoother.
Is this a failure of the project team to nail down the stories sufficiently at the outset, or should we (i.e. the dev team) take some of the responsibility?
The responsibility isn't that important IMHO (except for political reasons maybe but they do not produce much value anyway), both the dev team and the project team are working together and "failing" together. What is important here is to inspect and adapt to remove the obstacle. So, it's the dev team responsibility to make this problem visible (it is an impediment). And it's the ScrumMaster responsibility to work on this impediment. The failure would be to not work on it. Backlog Grooming sessions are one way to do it. And at the end, I'm sure the project team will improve and get a better understanding of what the dev team is expecting. And you will both produce better results.
Lots of good ideas here already on the scrum aspects of your problem. Based on your comment:
particular, we find that we are supplied with UI wireframes that contain much more complexity than the original stories would have implied (e.g. functionality duplicated for usability reasons).
I also have a concern that you might need to work on you development process as well though. Accessing functionality from multiple locations in a UI should be a simple addition that consumes almost no time at all. If you are finding this to be a common problem then your functionality is too tightly coupled to particular UI elements. Your team might need to improve their design skills (eg: pattern usage).
This is interesting. It would appear that you are doing the sprint planning in the sprint? And that the Sprint Backlog is committed before the Sprint Planning? If so, how are the team commiting to the sprint backlog without discussing the details of the stories?
An alternative approach could be to make the Product Owner aware that certain stories cannot be added to the sprint backlog due to lack of clarity. In particular that the acceptance criteria are not fully understood. This could provoke the necessary conversation with the Product Owner. Ideally it shouldn't come to this. It should be discussed and resolved in the retrospective.
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 7 years ago.
Improve this question
My organization has been experimenting with the introduction of more "Agile" methods. We've been trying the Scrum approach for a short while, and most of the team has, more or less, adapted to it. I like it as a whole, but I'm concerned about one potentially severe impact of the methodology: as teams are consistently focused on features and backlog items, and testers are more integrated with the overall development process, it seems like skill sets are becoming blurred, and people are sensing less respect for their individual abilities.
Some of our developers are excellent at server-side technologies and optimization of heavy-weight data provisioning. Others have invested a large amount of their careers learning GUI technologies and have developed a fundamental understanding of users and usability in an application. Neither skill set is better than the other, but they are certainly different.
Is this an inevitable result of the Scrum process? Since everyone on the team (as I understand it) contributes to satisfying the next feature/requirement, backlog item, or testing goal at hand, the underlying philosophy seems to be "anyone can do it." This is, in my experience, simply not true. Most engineers (developers, testers, etc.) have a particular skill set they have honed over the years, and the Scrum methodology, in my mind, tends to devalue those very abilities they were previously respected for.
Here's an example for clarification:
If a sudden change of technology occurs on the server-side data provisioning, and every item on the to-do list for the sprint is based on this new change, the GUI developers (who likely haven't had time to become acclimated with the new technology) might not be able to contribute to the sprint. At the very least, they will need to invest time to get ramped up, and then their code will be suspect because of their lack of experience.
I understand the need for rapid development to discourage "role silos" but doesn't this discount one fundamental reality: people develop skills in accordance to necessity, their interests, or their experiences. People seem to be less motivated when they perceive their position is one of "plug-ability" (e.g. we can "plug" anyone in to do this particular task). How does Scrum address this? If it doesn't, has anyone addressed this when adopting the Scrum methodology?
The short answer is an emphatic NO! Scrum does not blur or depreciate the skills required for specialization. Scrum does not promote generalization.
The long answer is that in Scrum, the most important thing is to get the work "Done". The team, as a team (as opposed to a collection of individual "stars") collaborate, as needed, in order to get the job done. Whatever it takes - however they want (Scrum is about self managing, self motivating teams, right?).
What this means is that a scrum team may be composed of several specialists, who primarily do what they specialized in (DBA, Graphic Design, even technical writers). The team, as a whole, should have all of the skills required to fulfill the requirements. This is not the same as saying that each team member has to have all of the skills aforementioned.
That being said, it is often desired - often by the members themselves - that members other than the specialists be at least adequate in skills different from their specialty. Another poster already mentioned Scott Ambler's "General Specialist". This helps the team when there's too much work of one kind, when the specialist is absent, and it helps the member when he really would like to gain experience outside his specialty.
Given that the team is self organizing, if for some reason a specialist finds himself in the middle of the sprint, without any work to do in his specialty, the best way to deal with it, is to simply ask the specialist what he wants to do. Let the team decide. The specialist can decide to help in his other areas of adequacy, do a POC for the next sprint, "shore-up" the defenses by fixing some long forgotten technical debt, or shine the shoes of the members who are working.
Yup. I don't know if this is the long answer. But it definitely was a long answer.
:-)
The point of Scrum is for the developers to self-organize. We use scrum where I am, and jobs get passively sorted by a person's focus. We don't do it on purpose with a chart and list, it just happens. We all know who's best at what, or what their main/secondary focuses are. If the 'main' person needs help, they get the person/people with a secondary focus in it to help. We do get plenty of tasks not necessarily in line with whatever our particular focus is, but you always know who to ask for help then.
For your example - I don't know that if you say had 3 server guys and 5 gui guys, that you'd expect to get all the work done in that sprint (if the server guys + some help from the others wasn't enough). The way the sprint is supposed to work is that from a prioritized list, the developers pick what they think they can get done in that 30-day timeframe. If that meant the GUI guys needed 2 days of server-side training in order to help, that's what it'd mean. Unless there were concurrent things also high up the list that they could do instead. The sprint tasks are not supposed to be dictated by management as a psuedo-deadline.
If you have a Safari account, there's an interesting mostly case-study book by one of the guy/s who invented scrum.
I've been working as a ScrumMaster for about 18 months and have worked with two different teams. I initially expected to experience the potential issues you raise but this has not been the case. What I generally observe is that the team evolves into a mixture of specialists and generalists as people find the appropriate role for themselves - one that they can enjoy and be successful at. This is self-organisation at work. I have never had a case where our specialists were sitting idle.
If this did occur, I would expect it to be raised as an issue in Sprint Retrospective and the team would discuss how to improve the situation. The most obvious (and brutal) conclusion would be to change the team composition.
I am not sure why skill set will get blurred. There is a fair amount of confusion in the agile world. Scrum is a project management process and not a software development process and should not be seen as one. The engineers have to follow their own methodologies like TDD or extreme programming to add their own part to being agile.
Nothing goes away in scrum.
PM's still document as they go
Architects still architect their components. The only thing is they just delay some major decisions to more responsible point in time.
Developers should still follow best practices such as SOLID principles to enable for refactoring in a consistent manner as features change.
I think Scott Ambler addresses this issue very thoroughly in http://www.agilemodeling.com/essays/generalizingSpecialists.htm...
His concept of a Generalizing Specialist is exactly the thing Collective Ownership / Scrum Team calls for, and makes total sense to me.
Its hard to achieve in real life though ;-)
If you find for any reason ('sudden change of technology' or not) that the amount of work required for a system over a sprint is greater than the amount available then there's a problem with your scheduling.
One fix is that, as you suggest, you take programmers from other areas and throw them onto the mix. How well this works depends on the skills of that person and how different the problem domain is, but treating programmers as generic units that can be farmed out as needed is generally not a successful strategy for developing software.
This is still a scheduling problem though.
The best thing about Scrum is exactly the fact that skills do get a bit blurred! The point is to avoid silos at all costs by spreading specialist knowledge across the team and letting people work a bit outside their comfort zone.
Obviously this is not for everybody. Some developers are happy in their own narrow specialist field and such people are more of a hindrance in a Scrum process than an asset, whereas well-rounded and multi-talented people who are determined to get the job done, usually adapt very very well to it and are far more productive.
One of the key benefits of Scrum is to get the whole team actually involved and invested into the project instead of tackling their own special tasks and then riding off to the horizon. I'd claim that for most people, this is a far more rewarding way of working than the conveyor belt -approach of waterfall processes.
So I'd advise to boldly embrace the mixing of skills and having people come together to take down nasty problems instead of relying on specialist silos. The result of teams consisting of motivated people can be surprising.
Sounds like this would lead to more well-rounded developers, and also allow those who are experts in certain areas to continue to contribute their expertise.
I haven't used Scrum much myself (yet), but from your description, these types of teams would lead to a team/organization that is also more well-rounded as a whole - and shouldn't that be the goal of any team?
Handling sudden changes is part of Agile and this may mean that some people have to go off and learn new skills. Course this is more within the general Agile philosophy than anything Scrum-specific. There may be some extreme cases where the customer or business decides to change the world by bringing in something new and thus has to handle the subsequent pain of those people ramping up but if this is what they want and the developers are overruled, then there are only a couple of choices: (Take your lumps and try to handle the major changes) or (quit and get out of there).
While there can be some cases where someone that has specialized in something may be able to do things faster, this doesn't necessarily mean much if that is just one person on the team that is an expert and there is enough work in that area for 10 people for the whole sprint. Should those not an expert simply not do that work and let that one person attempt to get through as much as he or she can? I don't think so but there should be something to be said for those that aren't the best at something still trying to get done what they can get done.
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 8 years ago.
Improve this question
Yesterday I had a team leader of another team say that they took a while to figure out something I wrote on a wiki page because I referred to obtaining code from source control as "checking out" which apparently confused them. They said that they were use to Clear Case and had only heard of the term "joining a project" and said that they haven't really programmed much for a long time.
While this is fine, what it then made me think of is the different types of team leaders I've had over the years. I've had some that have been almost purely managerial and I've had those that are programmers that do managerial things at the same time.
Do people have a preference as to what kind of team leader they have? How do you care if your team lead is active in the development of your product? I find team leaders who actually sit and code like the rest of the team more likely to understand things like (from my experience):
things aren't always as simple as they sound. Team leaders I've had who don't code or rarely code at all believe everything is a piece of cake and shouldn't take much time at all (which perhaps might be the case if you want to hack it together)
they are more understanding that developers don't always like sitting in long meetings and do their best to avoid getting their team into as many pointless meetings as possible
they understand what you say from a technical point of view. Those that might not have coded for a while might not be up to speed with a lot of the new technologies, techniques or lingo
I find it much more satisfying to have a team leader who has the mind of a developer and likes to get their hands dirty in the code as well. Perhaps there are some people out there that like team leads who distance themselves from the actual coding side of things and simply doles out the work, or perhaps another type of team leader that I haven't mentioned?
A team leader has to be a coder -- they can't lead the team unless the team respects them and where they're taking everyone.
A team manager, on the other hand, can either be a coder or someone who is just well organised and knows when to ask questions and interface to other management.
It is possible to find both a manager and a leader in the same person, but more often the roles (should be) separate and distinct.
You should read the book Managing Humans. I am of the opinion managers should keep their hands out of the code. They have more important responsibilities like keeping people away from developers, so they can do their job. Having them jump into development creates confusion as they aren't in it enough to know what's going on and have their time divided between that and other things, so it is difficult to count on them for major pieces of functionality. Plus, it really sucks when you have to tell your manager that something they just wrote needs to be changed, and you have to go back and redo it. Managers are really their to jump on the grenades for the rest of the team, so they can focus on accomplishing the task at hand.
That being said, should manager's know about software engineering? Yes of course they should, that's the field they are in. Should be know how to code in the latest and greatest whiz bang technology? That shouldn't really matter as long as they get how software development works.
I have no preference, I can't, I have to work with all of them, even though too many cooks spoil the broth. On a multi-developer typical project I have a technical lead, project manager and a non-technical customer. Of course, divisional and programme management will each stick their head in.
There are a number of types of leader, each have their own traits:
Non-technical customer: "The customer is always right." Often wants a moon-on-a-stick. Will call both the management and the technical bods and take the best answer as gospel.
Team manager/line manager: Somewhat pastoral role. Not particularly interested in the project I'm working on right now. Steps in when there is a decision to be made between project priorities. Probably really wants to be a coder, and delegates all the rest of his work that he can to his subordinates.
Project manager: Varying degrees of technical know-how. Is concerned only with timescales and costs. Does not understand, "I don't know how long its going to take, I need to play with it for a couple of days first to get a feel."
Team leader/technical lead: Just another developer, but with more experience. Responsible for technical decision making that will affect the whole project. Often fighting with the project manager to carry out good engineering practice, even though it will take longer in the short term.
Team leader/glorified secretary: Someone who is supposed to lead the team, but acts as more of a secretary. (Usually a grade above the team). Answers the phones, insulates customers from the technical bods. This works fine until they ask a technical question, where the glorified secretary tries to blag his/her way out of it, and eventually they work around the secretary and talk direct to the team.
We typically have a PM (non technical) who manages the project from an admin. viewpoint and a Tech Lead who manages the technical aspects and provides technical leadership to the team.
The Tech Lead will code parts of the project and will probably be the main (only) developer for the "Proof of Concept" stage.
On some smaller projects, they are the same person but it's a rare combination.
The absolute worst Software Leads/Chief Software Engineers that I've worked with were the ones that wanted to be intimately involved in the technical details. Too many important tasks were either missed or just not done. Managing a team is a full-time job. If the lead wants to get involved in the technical aspects it will certainly come at the expense of the managerial aspects.
I’ve only had 2 Software Leads/Chief Software Engineers out of dozens that I thought were worthwhile. While both were previously software engineers, those days were long gone for both of them. They knew it. They didn’t even try to pretend. Their job was now to manage. Their job was to make sure the developers had every chance to succeed. They did their best to remove all obstacles and make sure everyone was making progress.
I have a theory, but have never seen it in action, that the best software lead would be someone who is not, nor ever has been a software developer. They specialize in the true spirit of management, specifically that of being a facilitator. Unfortunately, most managers are more politically motivated or are just in the job because they've reached their pinnacle technically.