[YBA] Freescale i.MX27 project

[Shachar Shemesh]

sammy ominsky wrote:
> On 13/08/2009, at 09:04, Jonathan Ben Avraham wrote:
>
>> The customer is looking for someone who can a commit to doing the 
>> board bring-up process, write the BSP and set up a BusyBox 
>> distribution with Qt libraries within five weeks, with significant 
>> penalties for late delivery.
>
> Are there significant bonuses for early delivery?
Without knowing the details, or anything else about the project or the 
client, here is what my experience says will happen to anyone brave 
enough to submit a quote. This typifies my belief that some clients put 
up "money saving" requirements that end up shooting themselves in the foot.

The client is, likely, working under the following constraints:

    * They just received the devices from manufacturing, and need to
      have the board brought up as soon as possible
    * They are living under an external deadline. They cannot extend the
      five weeks deadline because it is a deadline to them. This is
      either a similar contract with their customer, or an investor who
      is impatient to see whether he wants to pull his money, or any
      such similar circumstances.
    * Engineer looks at the task to be done, says "well, 90% of it is
      just duplicating what has already been done for the Freescale
      development board, so it shouldn't be a problem".
    * Client then says "hey! let's outsource the risk instead of just
      the work".

Any reasonable contractor will perform the following calculation:

    * There are a huge number of unknowns about this work. You can never
      tell how much is "based closely", or what horrendous bugs you will
      find in the drivers once you start. I was once involved in a
      project (along with TkOS) where the board was "based closely" on
      the versatile platform. The project was a 9 months project that
      included a lot more than merely bringing up the board, but seven
      months into the project "board bringup" tasks were still being
      performed.
    * As a contractor, if I'm going to accept the client's risks, I need
      to be rewarded. In statistical terms, the expected value must be
      positive. If the project is at a loss if I am late, it must be
      really really profitable if I'm on time, or else there is no point
      in taking it up to begin with.

As a result, the quote is typically high. Very high. In addition, the 
contractor obviously states that all times are from the point where an 
order is issued.

The client is surprised. They usually don't understand that it was their 
penalty requirements that drove the price up. After all, this is 
supposed to be a simple project, merely performing adaptations to an 
already brought up platform, over in five weeks. As a result, it takes a 
few days, maybe even a week, to approve the quote (usually demanding 
that the price become lower). As far as the contractor is concerned, 
this week is not counted toward the delivery date, but since the client 
is constrained by external deadlines, as far as they are concerned, it 
does. The result is that the project is late, the client AND the 
contractor start disgruntled at the other side's "unreasonable 
behavior", and all sides lose.

Here is what could have been done to make things better. The client 
issues a request for a project at cost+, asking for a discount on the 
hourly rate in exchange for a significant bonus in case the project is 
delivered on time. Mathematically speaking, this offer is identical to 
the above offer, but as it is phrased in positive rather than negative 
terms, it is much easier to approve. This, of course, means that it can 
start much earlier, and have a better chance of succeeding.

Shachar

-- 
Shachar Shemesh
Lingnu Open Source Consulting Ltd.
http://www.lingnu.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cs.huji.ac.il/pipermail/linux-il/attachments/20090813/198c183e/attachment.html>