Wayan,

Stop linking to the flawed TCO report. Cooking up its page rank will not make it better. Have you actually done any constructive work for OLPC? Any lines of code? Any debugging?

Grow up, please!

Its interesting that you feel coding and debugging are the only constructive work for OLPC.

Personally, I feel that clock-stopping technology like the XO laptop is great, but without a strong implementation plan, of which understanding costs is integral, any project, no matter its grandeur or promise, is doomed to failure. And this belief drives my input into OLPC News, OLPC News Forum, OLPC Learning Club DC, and all the many OLPC-related activities and XO user groups I've helped start and grow over the past 3 years.

I only hope you can join us Thursday night so you too can add to the constructive conversation around One Laptop Per Child.

I think linking to that post is just fine, as long as people also read the comments. If they do so they will learn the %2,500 figure is completely bogus, and the real number is more like $450.

A note to people who are new here: Wayan is a splendid fellow who has made many important contributions to the olpc discussion. He does, however, have a flaw, which is that he often won't admit it when he is shown wrong.

Eduardo,

Funny enough, I've often admitted error when, in fact, I am wrong. However, I do feel that the VWC report is much more accurate in its TCO calculations than Tony's, mainly because Tony does not take into consideration major costs that while dismissed by OLPC, are very real in OLPC deployments. Still, I have asked Tony to put his thoughts into a post for OLPC News to further the discussion around TCO.

Personally, I'd love to be proven wrong, that TCO is much lower. Sadly, from my experience, and work in the field, I think that even $2,700 off - less than the real TCO for XO or any other computer deployment.

Doesn't wash, Wayan.

For one thing, most of points that were made about mistaken assumptions in the study you never replied to. I am assuming that was because you knew the criticisms were correct but didn't want to admit it. Instead you made the lame argument that the authors were trustworthy people and so we should accept whatever they claimed.

As to this point you are making now about Tony allegedly leaving out costs, you didn't say that in the original discussion, but only when I forced your hand by bringing it up in a new post in a way ("note to new people") that you couldn't let pass.

You say that the VWC report underestimates costs, and your solution will be to publish a post by Tony that underestimates them even more. Huh?

Look, you have a good point, namely that we need to take into account all the expenses involved in deployment. The right way for you to go about making that point would have been to do an analysis with correct assumptions. Instead you trumpheted a report full of mistaken assumptions, and then refused to admit they were mistaken. The right thing for you to do at this point is to go back and re-do the analysis with correct assumptions. That would be a much more productive basis for discussion and debate.

@Eduard Montez:
"The right way for you to go about making that point would have been to do an analysis with correct assumptions. Instead you trumpheted a report full of mistaken assumptions, and then refused to admit they were mistaken. The right thing for you to do at this point is to go back and re-do the analysis with correct assumptions. That would be a much more productive basis for discussion and debate."

I agree completely with the both of you. However odd this may sound, this is possible.

The point is, the original TCO study, just as the one a year ago, and every other TCO study ever contracted by MS, is not corrupted, faulty, or wrong, or whatever. What happens is that MS carefully designed a use case, context, and time window that would put their own products in a favorable light, and the competition in a bad light.

But there are no *correct* assumptions, just assumptions. After the fact, you can determine the costs of a specific deployment in a specific situation. But if you want to study "average" deployments in "average" situations, you must perform a sample of deployments in a range of situations. But that is not a TCO study. And by the time you are finished, technology has moved on and everything you measured has become obsolete.

A real-world TCO study is a modeling effort where you measure some of the free parameters of your model.

Wayan is right when he states that deployment is ALWAYS more expensive than simply shoving in the hardware. And any serious deployment study gives you data you need to estimate the future costs of deployment.

I see Wayan trying to tease the OLPC to formulate a realistic deployment model with realistic costs. This MS designed TCO study is basically wrong for XO deployments in every possible way. But it contains realistic numbers for some cost factors.

You are right that the costs of XO deployments are way different, and possibly cheaper than this crazy high school computer lab plan.

But, Eduard, if you have suggestions about what costs should and should not be used in such a calculation, post them. I am sure the challenge will be picked up and someone will combine the whole lot.

Winter

Winter,

"What happens is that MS carefully designed a use case, context, and time window that would put their own products in a favorable light, and the competition in a bad light."

I'm afraid this report goes beyond of what you described above - see again my relevant post showing, for example, where they took teachers' salaries from UNESCO source and substituted PPP$ with US$ (i.e. assuming the PPP$=US$), which, in case of developing countries, is completely nonsensical. Other figures' sources are less transparent and, therefore, more difficult to verify but I suspect similar disregard for correctness of information...

@Delphi:
"for example, where they took teachers' salaries from UNESCO source and substituted PPP$ with US$ (i.e. assuming the PPP$=US$), which, in case of developing countries, is completely nonsensical."

Obviously, when someone is changing the data, that can be an "error" or even fraud. But I live by the maxim:
"Never attribute to malice that which can be adequately explained by stupidity, ... but keep your eyes open"

On the other hand, I see no people left standing that trusted Microsoft.

Winter

Eduardo, Allen, Winter, et all - I do hope we can meet tonight in San Francisco and continue this conversation in person. I would be good to raise a toast to our differences and agreements that both make the OLPC initiative so interesting and compelling.

And no matter what comes of it, we'll have a more lively debate than anything on TV.