Michael Trucano, a recognized expert on deploying ICT in education, recently blogged about Worst practice in ICT use in education:

Here's a list of some of what I consider to be the preeminent 'worst practices' related to the large scale use of ICTs in education in developing countries, based on first hand observation over the past dozen or so years. I have omitted names (please feel free to fill them in yourself). The criterion I used for selection was simple: The given worst practice was easily observable in multiple prominent initiatives, with (one fears) a high likelihood of re-occurrence, in the same or other places.

As one of his humble understudies in ICT4E, I found myself in agreement - these are worst practices that are repeated again, and again, and again, when technology and schools are mixed. I've also seen these practices in multiple OLPC deployments.

Walter Bender sees OLPC succes

In fact, I started OLPC News exactly four years ago because I saw OLPC itself espousing many of these worst practices as core aspects of its deployment methodology.

Yet, not all OLPC deployments have committed these sins against common sense. To prove it, Walter Bender has rebutted each worst practice per the Plan Ceibal and Paraguay Educa deployments:

1. Dump hardware in schools, hope for magic to happen.
   Far from it, there have been extensive support mechanisms in place in .ur and .py from Day One
2. Design for OECD learning environments, implement elsewhere.
   While there is some sharing of content and best practice, it is the local pedagogical team that calls the shots in both deployments.
3. Think about educational content only after you have rolled out your hardware
   Again, pedagogy has driven the pace of deployment. At the same time, the entire deployment has been thought of within the context of a learning platform, which includes laptops, connectivity, servers, training, content development, documentation, support, community outreach, etc.
4. Assume you can just import content from somewhere else
   The key here is "just". Both .uy and .py think deeply about content, but they are also opportunistic - taking advantage of great content developed elsewhere, for example, by the Etoys community.
5. Don't monitor, don't evaluate
   At Ceibal, they have an extensive operation for monitoring the state of the network, servers, and laptops within their deployment. There are numerous ongoing evaluations of the program, both internal and external. Paraguay Educa was the subject of an external evaluation by the IADB, which issued a very positive report.
6. Make a big bet on an unproven technology (especially one based on a closed/proprietary standard) or single vendor, don't plan for how to avoid 'lock-in
   Both programs have used a open bidding process and have some percentage of hardware from multiple vendors. Both programs use Free Software.
7. Don't think about (or acknowledge) total cost of ownership/operation issues or calculations
   .uy has been diligent in publishing their total-cost-of-ownership numbers - these numbers, based upon the costs measured in the field happen to be much less than the inflated numbers fabricated by naysayers.
8. Assume away equity issues
   While no one is claiming that equity issues are no longer a concern, the fact that the per-household penetration of computing in .uy is inversely proportional to household income says a lot. And in every one of these households, the children have free Internet access. Wow.
9. Don't train your teachers (nor your school headmasters, for that matter)
   The biggest investment in the .py program has been in teacher training. As the project scales, finding ways to make this process more efficient will be key. But no one has every suggested that it was not a vital part of the process.