User:Mokurai/Evaluation

Nicholas Negroponte hates the idea of having trials of competing laptops and software, and evaluating the results, but it isn't his choice. The question is not whether to compare them, but how. Depending on the questions you ask, or fail to ask, you will get wildly different results. So if your evaluation criteria include those that some government officials have stuck in their minds, like

• Memory size
• Hard drive size
• Processor speed
• Graphics acceleration
• "Industry-standard" software (Windows)
• Teaching to the official tests
• Corporate support

you get one kind of result (not the XO, if you were wondering), and if you ask about what matters to village schoolchildren for getting out of poverty,

• Lowest cost
• Lowest power consumption
• Ruggedness
• Security
• Screen resolution
• Readability in daylight
• Book reader mode (screen rotation)
• Software Freedom
• Mesh networking
• Software for collaboration
• Software designed for education
• Powerful programming tools designed for and tested with children
• UI design for children
• Teaching understanding
• Support by the community, including the children

why then none of the supposedly competing solutions gets a look in, it's XOs all the way. This year. We can look forward to a multitude of other computers emulating some of the XO's advantages, or even going further, such as OLPC's XOXO (promised for 2010 @ $75), the Pixel Qi laptop (promised for 2010 @ $75), and perhaps the promised $10 laptop from India.

What we have to do, then, is to answer all of the questions, and then demonstrate which answers correlate with the effectiveness of the education delivered and the cost to do it. And, as is required in scientific investigations, provide not just one set of answers from one set of circumstances, but a sufficient variety to convince most of the doubers (though probably not the nay-sayers), and more importantly to plan the next round of improvements to the program. We cannot expect the same kinds of gains in the poorest villages and in urban schools in the US, or among the learning-disabled, the normal, and the over-achievers. We have to allow for children recently traumatized or even injured in war or famly violence. There are dozens of such possibilities in endless combinations.

We already have anecdotal evidence from village deployments for important effects:

• Increased school enrollments
• Increased math and reading scores
• Increased sharing outside of class
• Children taking on responsibilities that were considered to be far beyond them, such as the Children's Hospital in Nigeria
• Individual children becoming the class or school expert on XOs as a whole or on some facet of the hardware or software
• Shared learning
• Ability of teachers to improve lessons through Internet access
• Acceptance by subsistence farmers of loss of child labor now that the children are getting real education

Each of these effects warrants its own investigations.

We also have evidence from decades of experience with Smalltalk, Logo, and other software for children's education that education can be accelerated and improved, but that this only works when teachers are trained, new teaching materials created, and curricula revised to support it.

There are a few formal studies showing advantages from providing a computer to each child. Saul Rockman has published some of them. There is almost nothing on results specific to Constructionism.