Re: [scrumdevelopment] Large research projects
Am 05.01.2005 um 07:33 schrieb Charles Prakash Dasari:
>The idea of SCRUM is to have a "mini alpha version" after each sprint.
> Hi All,
> I am a lurker in this group for sometime now, and have been fascinated
> by Scrum as a project management methodology. Following the
> conversations in this forum makes me more confident of applying Scrum
> to virtually any [software] project.
> I am trying to understand how best we can apply Scrum in the following
> There is a new product idea, and we believe there is a need for this
> product in a certain market, and possibly this product and a host of
> associated products (to be envisaged and developed) could create a
> market for themselves. We need to develop this idea so we can demo to
> anybody to explain what we are trying to do. Our rough estimate is
> like 2 yrs even before we get to the first alpha. We want to start a
> project for this.
>This limitation is only in your mind, IMHO. E.g. surely you do not need
> The product is much similar to development of a new programming
> language. Now you can easily imagine that there are quite a lot of
> infrastructure pieces in such projects that need to be developed,
> before it can be first used. And even before it can be put to any
> evaluation, we need the compiler+parser, code generator and a host of
> basic set of libraries. Designing these pieces with utmost care is
all thinkable container classes immediately.
As a unrelated hint I would suggest to found your new tool chain on top
of an existing platform like Java (libraries and byte code VMs
available, tools for code generation - byte code - are available etc.)
> key, as the success of the product could heavily depend on how theseProbably it makes sense to have two teams interlocked sprinting. One
> are designed upfront.
making the design and one making "prototypes". They should feedback
each other. The programming team can incorporate more and more of the
designs over time and the designers get feedback on how well their
> Moreover, in these initial stages, it is hard to find a committedSure. But I suggest to focus on a near term ROI goal. You have a grand
> customer, as we do not have anything more than a PPT to explain the
> product idea. The team composition is highly technical and of a size
> of about 6 - 7 people. There is only one (or max. two) top level
> executives committed to this project (who are also funding), others
> hardly know anything about this. But once we have the first alpha out,
> we are confident we can buy more committers to this project (both
> internal executives and a couple customers).
> Can I apply Scrum principles to this project? I do not have any doubts
vision of a new programming language and surrounding tools. Fine.
I'm sure there is "something" you can achieve in lets say 3 months, in
lets say 2 sprints of 6 weeks each, or 3 sprints of 4 weeks each.
What about a mini interpreter which is only capable of working in an
web application environment? Or an mini interpreter which is only
useable as builder for a Java Swing based GUI?
With mini interpreter I mean a "reduced to the max" very limited
language. E.g. no data types except strings ... no I/O except using one
single well known class of the underlying platform (Java or .NET)
Show the funding executives how the interim releases (sprint results)
will earn money.
> about the need for team work and the increase in communication amongStart with a backlog. Try to prioritize in a way that your very first
> them; this is absolutely key. I am more worried about how I can
> organize my sprints, what is the delivered value after each sprint,
> who should evaluate the stages of the product etc.
version after 3 months is a running showcase. The showcase should be
able to do/to demonstrate something which is at least of value for a CS
student or something similar.
Solve one problem you have in your organization with your new tools,
even if its only a small one.
>Hope this helps!
--------------------------- www.oomentor.de --------------------------
Angelo Schneider OOAD/UML Angelo.Schneider@...
Putlitzstr. 24 Patterns/FrameWorks Fon: +49 721 9812465
76137 Karlsruhe C++/JAVA Fax: +49 721 9812467