If you poke around enough on the Laptop.org wiki, you find a few interesting corners. Linked from their work in creating a training and reference document for OLPCCorps, a link to an Excel spreadsheet to calculate OLPC-specific costs for a deployment, which has been created and maintained by OLPC's John Watlington
OLPC XO math on costs
Better late than never?
I've been beating the drum since November 2006 on the importance of including the costs for the full environment that the XO laptop needs to thrive.
Implementation planning has to move beyond the per-unit cost, it must include the solution for charging the laptops, host the more power-hungry XS server, extend the mesh network to the community, and connect to the Internet at least occasionally. And that doesn't even touch the costs of curricula development, teacher training, and institutional changes for distribution, maintenance, even testing.
My 2006 TCO
I got into a good amount of hot water by proposing that the cost of OLPC projects went beyond (number of students)×($100). The magic number was $972/laptop for a 5-year project. Widely decried, a few much more in-depth explorations revealed that my back-of-the-envelope TCO calculation was actually pretty close.
GeSCI 1:1 TCO
Roxana Bassi and I worked on a tool which any implementing group could plug in local numbers while being led through increasingly complex worksheets which provided calculations for every level of the school system, whether you were concerned about per-school costs or system-wide expenses.
You can read more GeSCI studies or go straight to the TCO Guide and Worksheet.
Vital Wave TCO
At the same time, VitalWave also came out with a TCO calculation comparing different approaches (1:1 computing, computer labs, etc.) with special attention to the OLPC in the 1:1 model. The VitalWave paper was covered in a World Bank e-Development webinar last November.
And now ... the OLPC TCO?
The Laptop.org tool in the wiki seems to be a relative of the GeSCI tool in its approach, but cuts out some 1:1 computing calculations which are irrelevant to the XO laptop, and adds in other calculations specific to the XO to help plan for the maximum power draw while charging.
This is a great tool, and it's unfortunate that it wasn't part of the OLPC discussion from the beginning; joining the open source ideals of their design process with the rather closed business process.
It also reveals some of the presumptions being made in the design and implementation processes - the list of Internet connectivity options range from DSL and VSAT to "none", and call out specific cellular connectivity options, but don't even mention plain dial-up (there is an "other" category).
The following blog entries will present these various calculations head-to-head, first outlining what we will compare (and why) and then running the numbers to see both what the current cost of OLPC deployments are, and to reveal differences between the various tools.
Hi Jon,
this worksheet has been in the works for a long time, and the numbers there are based on our experience on actual deployments.
AIUI, it was shared in draft form with people working in early deployments. (As you'd expect from a draft with lots of "we're not yet sure about this" sections.)
The workbook was published Nov 2008, and it has seen a few revisions since (as it is obvious from the Wiki).
Of course, in some places VSAT costs are huge, so internet connectivity does take a backseat (and there are XS-based efforts to provide some content, and things like Wizzy Digital Courier). Not the best, but IRL you can't win every battle.
POTS dial up? Hmmm. We haven't heard requests for that, but the local telco surely will provide the relevant numbers.
If you have more feedback on this, and you want to help OLPC, maybe engaging with Wad on this topic is the productive path. Measured suggestions based on fact and actual deployment data I am sure will be welcome.
Wad may prefer the conversation to happen over a list, I don't know about that. Check w him.
Martin, thanks for your reply--I have a few more posts in the wing along this line, moving towards a massive comparison of the various models now out. Maybe we can direct those comparisons toward specific gaps or confusing/complex calculations? Have any deployments used GPRS or other cellular data networks? Is there a db (or collection of spreadsheets ) with deployment budgets we could open up?