Surprise! 5 Year TCO for Computers in Schools: $2,700

   
   
   
   
   

While I am not always a fan of Microsoft's actions around technology in education, I do have to give them credit for hiring my friends Vital Wave Consulting to do a 5 year Total Cost of Ownership (TCO) study on introducing computers into Indian schools that focused on multiple different configurations, platforms, and even operating systems.

linux premium

Today, I listened to Vital Wave give a presentation about the report (you can download it here) and I was struck by the "Linux Premium".

Technologists in the developing world who know Linux command a 15-30% salary premium over their Microsoft-only peers. This should be the new rallying cry for the LUGs in every growing economy: Learn Linux, Make More.

Still, that wasn't the real bombshell. It was the overall per-seat TCO for computers in schools, no matter the configuration, form factor, or operating system. Let's have JamesU explain:

Brazilian OLPC
Now $2,700 per computer per child
For me, the huge, eye-opening takeaway from this work isn't that Windows and Linux cost about the same to put into school labs in poor countries, it's that the 5 year cost of ownership for doing so is about $2,700.

That's right, $2,700. At a time when the press likes to write about whether the $100 laptop costs $200 or $300, economists who live in the countries where these systems are being deployed went out, assessed actual computer implementations, and came back with an estimate that the actual 5 year ownership cost is about 10 times as much.

Or almost three times as much as OLPC News calculated two years ago using only the collective estimations of the OLPC community. Our "$1,000 laptop" estimation doesn't look so crazy anymore, eh?

Yet one laptop per child programs on a countrywide scale now look even more insane. At $2,700 per child, exactly which poor developing world country can afford to implement at a 1:1 computer-to-child ratio on a national scale?

Related Entries

58 Comments

From the paper, "The India TCO model assumes that 10 days of complimentary training will be provided to one teacher per school under Project Shiksa or other Microsoft-sponsored IT training initiatives.".

That would be a good assumption only if Microsoft guarantees it in writing.

Maybe an assumption of LUG support for 10-20% of the problems would offset the cost of Linux support. In addition, after a year or so, the cost differential between a Microsoft trained tech and a Linux guru will be more in line.

One other thing I had an issue with is the assumption that all ultra low cost machines would need to be replaced in 3 years. The OLPC would only need some fraction of the machines replaced, and that is just to supply parts that fail. They are designed to be repaired by individuals with minimal training and tools.

Mark

Mark,

In theory and photo ops, yes, the XO can be repaired by students, but the Nigeria pilot had a +30% failure rate on its machines - unrepairable failures. From what I've seen in the developing world, I do not think that's a wildly high fail rate for laptops, no matter the build. Desktops are much more hardy, if only because they're less often dropped.

Even if the Vital Wave TCO estimate is too high, it still supports a point I have made several times here at olpc news (and that no one has even attempted to refute).

This is that the real educational impact of olpc is not going to come through deployments within developing world school systems. It will come in a few years from now when the XO gets so cheap that villagers buy them on their own, and likewise on their own learn to use the little machines to run self-instructional software, starting with reading. writing and numeracy.

Come on, folks. If a problem can't be solved through standard methods, then you need to think outside the box.

Even if the Vital Wave TCO estimate is too high, it still supports a point I have made several times here at olpc news (and that no one has even attempted to refute).

This is that the real educational impact of olpc is not going to come through deployments within developing world school systems. It will come in a few years from now when the XO gets so cheap that villagers buy them on their own, and likewise on their own learn to use the little machines to run self-instructional software, starting with reading, writing and numeracy.

Come on, folks. If a problem can't be solved through standard methods, then you need to think outside the box.

Oops, sorry about the double post.

Yes, I'm surprised. I'm surprised that you used a Microsoft sponsored report[1] and then have a Microsoft employee 'explain' to us what the report means. I'm surprised because most people reading this blog are rather more computer literate then your average 'Education Official' and even after brief analysis of the 'report' it's obvious that their methodology as applied to a XO computer isn't of much value - take a few of their points as an example:

1) "two of the three largest costs – training and support – are remarkably consistent"

It's save to assume that training how to use 'Write' on XO would take fraction of time of time required on training how to use MS Word - same goes for other comparable applications. As far as support goes, anyone who actually had any experience with large deployments of computers using Windows will tell you that a big chunk of support is spent on ongoing virus prevention...

2) "replacement for [] ultra-low cost computers, whichhave a shorter lifecycle than mainstream or low-cost desktop computers"

It's a common knowledge that the main cause of a computer failing are its mechanical parts, i.e. hard disks and fans - since a XO has neither its longevity is greater and not the other way around,

3) "Electricity and connectivity are two other areas where dramatically higher local costs could have a large impact on TCO. This would put electricity costs into the top cost categories, being nearly equal to the initial purchase price of the 16 ultra low-cost computers"

Indeed - an XO uses on average 1/10 of electricity of your average laptop and 1/200 of your average desktop. And yet, somehow, their 'analysis' shows (see Table A on page 6) that this would be cheaper for desktop computers...??? The requirement for reliable and continues power supply, as needed for conventional computers, is also of a less importance where XOs are used.

All in all, at least where XOs are concerned, the 'report' is next to useless - bit like 'find the truth about Linux' MS-sponsored reports we have come across in the past...

I'd be curious to find out what Bryan's real experience in Nepal is on XO's TCO.

[1]Vital Wave Consulting:
A Total Cost of Ownership (TCO) Model for Education Officials
( http://download.microsoft.com/download/2/0/a/20ac945c-34d0-4a60-8245-f80e80fe954f/Vital_Wave_Consulting_Affordable_Computing_TCO11June08.pdf )

Ops. "1/200" above should be 1/20. I assume average 5W for an XO and 100W for a low energy desktop (including screen)- of course, XO's screen power optimization would put that difference even higher in favour of XO...

@Delphi:
"As far as support goes, anyone who actually had any experience with large deployments of computers using Windows will tell you that a big chunk of support is spent on ongoing virus prevention..."

Actually, after malware is contained, the most of support goes into preventing and repairing users to mess up their computer installation. Which is solved in MS deployments by centralization of the storage. After that, support goes into keeping the central storage on-line. I have heard about the most idiotic configuration mishaps.

As the worst that can happen to a XO is a zap to its initial state and all backup is automatic, that is one other problem less for support.

So NO viruses and hardly any configuration troubles.

Winter

Obviously, another "Gut the Facts" campaign. The earlier "facts" were all debunked (remember the outage at the LSE?).

One point is the question WHY Linux gurus are more expensive. Probably because they are more productive. Which means that, on the whole, Linux is easier to administrate. Each operator can handle more installations.

Here is another discussion of this report.

Microsoft study overlooks Windows biggest cost
by Matt Asay
http://news.cnet.com/8301-13505_3-10050777-16.html

>>>>
In short, I believe Microsoft's sponsored TCO study may have many flaws, but only one that really matters: it overlooks the cost of buying a one-way ticket into Microsoft's walled garden. The cost of entry may in fact be quiet low. But what's the price of exit? Open source makes the cost of exit as close to free as the cost of entry is. It's a software development methodology and a CIO risk mitigation strategy* all rolled into one.
<<<<<<<<<<

* "What CIOs can learn from Honda -- Wednesday, Sep 24, 2008" http://news.cnet.com/8301-13505_3-10049610-16.html

"Open source shares risk between vendor and CIO. It's a more equitable and intelligent way of deploying software, because it allows the CIO-- Honda manufacturing style--to quickly adapt to changing conditions and custom-fit software for a CIO's particular requirements."

Winter

The XO consumes at least 10W when used indoors with wifi on (normal use). Your estimate for 5W is way too low. 5W average will happen only with sporadic use, with lot of shallow suspend and deep suspend. Anyone can check this on a G1G1 XO (8.2-759) with a simple power meter.

This is quite good, but only 1/3 - 1/4 of my powerful Toshiba m200 laptop, not 1/20 or 1/100.

@sola,

From ( http://wiki.laptop.org/go/XO_power_draw ):

"Summary of gnu results: Normal power consumption of about 6 watts. By dimming the screen, you can save up to a watt. By turning the screen all the way off, you can save another 0.5 watt (total 1.5 watts for screen). By turning the WiFi chip off, you can save up to a watt. By using a WiFi access point rather than a mesh, you can save up to 0.2 watts. None of these techniques involve turning off power to the CPU (the "Automatic Power Management" option in the Sugar control panel). "

Hence my estimate for an average usage of a XO being 5W.

In my previous post I estimate an average power requirement for a laptop to be around 50W (ie. 10 times more) and for an average desktop around 100W (ie. 20 times more). And while my figure for recent laptops is probably too high, the figure for a average desktop is probably far too low...

Either way, given, as the report itself emphasizes, high cost of electricity component in TCO, XO is, by far, the cheapest machine to run for everyday needs of a primary student (and that would include extended periods of 'book reading' activity when the powers usage of an XO is very small indeed)...

@Winter,

"Obviously, another "Gut the Facts" campaign. The earlier "facts" were all debunked (remember the outage at the LSE?)."

Obviously. But I would have never expected OLPCNews taking part in it...

i cannot believe this is posted here. this is terribly irresponsible for OLPC news and is just furthering my belief that this site is actually an anti-OLPC slanted site.

look, the study was funded by microsoft. MICROSOFT. so of course it says they are so much better than linux.

what is the cost of removing viruses? spyware? malware? what is the cost per laptop of the 'services' you pay for to prevent these? how much is the license per seat going to cost? what about security updates?

sure the 'guru's may be more expensive, but you don't need them as often. AND, please remember there is tons of free linux help on the internet.

this is FUD, pure and simple. this is MS say that the OLPC's should be running Windows and here is why. seriously OLPC News? seriously? think of what good could have come if MS had put the money they paid for this study into something, ANYTHING, worthwhile.

and it upsets me that OLPC News is a part of the FUD.

FUD, FUD and FUD.

Children can maintain the XO laptops for free. Jeez. That was the whole point. Pick the smartest of the bunch and make him/her the admin.

I've done the power meter thing on my XO, it is closer to 10 W than 5 W -- even with the battery removed to preclude energy use from charging. Personally, I would rather trust my own measurements over stuff posted to the OLPC wiki.

The other comment that I have is that TCO cannot be calculated for the general case. There are far too many variables to deal with, some of which are implementation dependent and some of which are region dependent. It will cost more to have IT staff maintaining the stuff, but you may get more uptime. If you have a reliable server, then you can just reflash an XO in minutes, rather than dealing with data recovery. Electricity costs will be highly variable. I can quite easily imagine a desktop being cheaper to run if the desktop is on the grid, but an XO is using solar. Studies and decisions based on TCO have to be made at the level of the buying institutions, even then they run the risk of being wildly wrong.

Stewie: That is a lot easier said than implemented or institutionalized.

Beyond that, though; let's presume a Microsoft bias in this report -- this means that the actual numbers will be probably larger, due to things other commenters have noted (anti-virus/malware/recovery support, "free" training, increased training costs due to more complex software).

The original OLPC lifespan was supposed to be 5 years. I look forward to seeing a not-very-biased report on how that's going. Circumstantial evidence from G1G1 and Nigeria would not indicate that the average XO will last 5 years.

Making a big deal over the MS vs Linux costs is only addressing one part of the cost survey. Linux administration (supposedly) costs more; but also you're more likely to find a LUG that will assist for cheap or pro-bono, not to mention hordes of online volunteers.

More important is the overall gist of the report - the cost of the hardware is not the largest cost of a deployment. Regardless of the software platform, you'll need teacher training, curricula and content creation, hopefully some M&E metrics, an Internet connection for each school, reliable electricity of some form (for the School Server, Internet router/modem/sat dish/etc., and to charge the XOs if households don't normally have power), replacement parts and some trainer to at least guide the first round of volunteer techies in the more complex hardware repair tasks...

The per-computer costs of all of these very important pieces to a successful deployment will dwarf the raw per-computer cost in any TCO.

I love the finger-pointing and the recriminations. All kinds of FUD towards the report itself. Why? Because it lays bare the real issue with computers in schools.

Its not if it runs XP or if kids use XO's - its that per seat costs average $2,700. Give or take a few hundred and you are still waaaay outside the budget of any developing country. Again, I ask:

Exactly which poor developing world country can afford to implement at a 1:1 computer-to-child ratio on a national scale?

I want to add some to my comment above. People are looking at the problem of education in the developing world in terms of technology and education methods. That is appropriate, but you have to understand that the heart of the problem is political.

Compare the literacy rates in China and India. For China it is %90 but for India only %61. The difference is not due to money, as China was until recently as poor as India. The great disparity is because the Chinese government has long had a real committment to education, whereas the Indian government hasn't.

A lot of this has to do with governmental corruption. In India many school teachers get paid but never show up in the classroom because they pay off education ministry officials. And much money allocated for education disappears into the pockets of these same officials. China has a great deal corruption, but due to its committment to education it has kept corruption from totally undermining primary and secondary education.

There is no way, at least for the foreseeable future, to reform most of the many developing world governments that neglect education. That means that great ideas like olpc will simply never be implimented, at least on a large enough scale to significancly raise literacy rates.

The only practical solution is to go around the government with new technology. That is what happened with phone usage. In developing countries the phone systems were long controled by governmental organizations that set the prices so high that very few people could afford phone service. But then came in cell phone networks with $20 phones, and the result has been an explosion of phone usage that in turn has helped lift many people out of poverty.

Something similar is needed for education, and here the XO, once it gets cheap enough, seems to be the answer. Let the people take over education, at least for those who are not at present being served by the government. And this has the added benefit that the families that buy XO's will have many other uses for them, including telecommunication.

Like I said, if the standard solutions don't work, you need to think outside the box.

Wasn't one of the original ideas behind the XO to do an end-run around the teachers, ministries of education, and (to a degree) governments by placing the education tools in the hands of children.

Now I'm a strong proponent of giving teachers good tools, but if you're claiming that the teachers themselves are corrupt (by not teaching the children) they you need good tools for the children to use on their own, perhaps with some guidance from their parents. Now in a country with a high literacy rate like China, that should be easy enough to do because computers are designed to serve a literate audience. But, as eduardo pointed out, they don't need it because they already have an effective education system.

So that leaves countries like India, that have much lower literacy rates and a less effective education system. In that case, we cannot expect the child to know how to use software that requires a certain degree of literacy nor can we expect the care-giver to have the necessary levels of literacy to guide the child. Perhaps the key in a situation like that is to develop software that can use the spoken language. Perhaps the software can also take initiatives such as improving adult literacy.

As for this TCO bit, and the 1:1 computing initiatives, maybe Wayan is right about asking which poor countries can afford these initiatives if the TCO is in the thousands of dollars for 3 years. It certainly makes these initiatives more affordable in the developed world, but it is still out of reach of the developing world.

Maybe the answer to that is to go to these countries, figure out the actual TCO (thus far), and figure out what the leading costs are in that TCO. The OLPC foundation should then start looking at ways to drive down that TCO so that this technology does become accessible.

But I think that utility is an important factor here. I've seen quite a few activities developed, things that range from music to physics. There are quite a few games, and I'm sure that individual countries are developing stuff to target literacy. But the question is, do they need that sort of thing? If they are an argrarian society, they may need to know about the best practices for farming. They may need to know ways to improve public health, and address individual health issues. But most of all, they need this information delivered in such a way that they will accept this information. The "white man's" guide to growing crops without fertilizers, herbicides, or insecticides won't necessarily help them if they distrust what they are told. When all is said and done, a computer may be the best way to deliver this information, but the XO may not be the best computer. Something like a cellphone with more memory may be more appropriate. Let's face it, the XO is essentially a tool for the literate: that's why you need a big and high resolution screen (compared to a cell phone, not compared to our computers).

I think Jon's comment hits the nail on the head: "More important is the overall gist of the report - the cost of the hardware is not the largest cost of a deployment." 'nuff said.

Wayan,

"I love the finger-pointing and the recriminations. All kinds of FUD towards the report itself."

Methinks you doth protest too much ;)

I've already pointed out why this report's findings [1] don't make much sense as applied to XO machines. And, since the details of how the figures were calculated are not available it's, in a effect, pretty much useless at least as far as the deployment of XOs goes. And we don't have to speculate. Bryan Berry (and one of OLPCNews editors, after all) has a real experience of XO deployment costs in Nepal - hopefully he could make time and post on his own experience on the subject...

Now, in your "Let's have JamesU explain:..." you quote James Utzschneider who is the general manager of Dynamics Marketing in the Microsoft Business Solutions, of which, of course you are aware of as it's the same guy who, just not long ago, you accused of faking a video [2]. I don't believe James to be a 'villain' you portrayed in that post nor I believe he's an 'expert' we should rely on in this post. He's merely doing his job as a PR guy for MS and "fake" video or this report are part of the same effort - a PR effort with an obvious bias and, as Winter already pointed out, very similar to MS's "Get the facts" campaign in the past...


[1]Vital Wave Consulting:
A Total Cost of Ownership (TCO) Model for Education Officials
( http://download.microsoft.com/download/2/0/a/20ac945c-34d0-4a60-8245-f80e80fe954f/Vital_Wave_Consulting_Affordable_Computing_TCO11June08.pdf )

[2]Did Microsoft Fake XP on XO Press Media?!
( http://www.olpcnews.com/sales_talk/microsoft/microsoft_fake_press_media.html )

Jordan,

Yes, you would need talking software. I was assuming that, but should have made it clear. I am glad you included adult literacy, since that is an extremely important problem but gets missed in the elementary school classroom that olpc is focused on at present. And I like your ideas about subjects that would be useful to people in agrarian societies.

As far as corrupt teachers go, it isn't just that some teachers are corrupt but that the ministry of education and the whole government are.

Wayan,

You know, a few months ago when Pamela Jones was falsely attacking you for being an anti-olpc FUDmeister, I put up some comments at Groklaw defending you. And now you pass on a royal piece of Microsoft FUD. I understand you have some very valid concerns about TCO, but couldn't you find a better source of information on it?

Jon,

"More important is the overall gist of the report - the cost of the hardware is not the largest cost of a deployment."

Perhaps. But it's the conclusion that the amount ($2,700 !) is the same no matter what hardware (so you might as well buy more expensive computers) or software (let MS take care of you) you're going to use, which the report, MS's James Utzschneider and, finally, Wayan is emphasizing.

Which is, of course, nonsensical - a bicycle (XO), a motorbike (ClassmatePC and the like) or a SUV (conventional desktop) are not going to have the same TCO just because they all can be called 'transportation vehicles'...

Delphi and Eduardo,

You may have missed it in the opening sentence, but this report was developed by friends of mine, the people of Vital Wave Consulting. Had it been anyone else, I may have suspected major Microsoft bias, but with them, folks I know and trust, I can rest assured that they did their best to be objective in a very subjective subject.

Is the report flawless? No. Is it totally comprehensive? Nothing can be. But if you look at their sources, and study their methodology and assumptions, its the best TCO analysis of computers in Indian schools that we've had to date.

And again, let's say they are off by $500 even - either way. That still gives us at least a $2,000+ per seat TCO. This is the shock we should be digesting.

Shouldn't the cost of teacher training be amortized over the career of the teacher. They seem to assume either a very short career for the teacher or that the teacher will need continual training. They also assume that a linux tech is only as productive as a MS tech. Maybe once diploma mills start pushing them out this might happen, in my office it take 3 MS techs to support the same number of machine as the linux techs.

You can't really amortize the cost of teacher training over the career of the teacher since a lot of people do not adapt to new technology very well. When new software is introduced or, to some degree, new hardware is introduced, you will have to retrain many of the people who use this technology. Otherwise you run the risk of them using new technology in the old way, or even not at all.

Think of it this way: a lot of western teachers were learned how to use stuff like word processors and spreadsheets, the web and email, or certain education applications that are best described as skill and drill (Number Munchers, Oregon Trail, etc.). They know how to integrate this stuff into a classroom, or even their own workflow. Yet if you go into a classroom that has access to tools that don't fit those moulds, you won't see the software used all that often (I'm thinking of tools like Fathom and Geometer's Sketchpad). Or worse, I've seen some teachers who were trained with Fathom say, "why the heck would I want to use that when I could just have my students use Excel." (The connection between Fathom and Excel is a lot like the connection between the Write activity and Word. One is geared towards education through the inclusion of special features and simplification, the other is not.) Of course, you would also have those teachers who will pick up and appreciate the new tools without training.

I would imagine that the same sort of dynamics would exist among the teachers in various nations: some don't care about learning new technology, some care but are inflexible in their ways, and some both care and are flexible. The last group you don't have to worry about too much, but the first two groups will incur on-going retraining expenses.

"The XO consumes at least 10W when used indoors with wifi on (normal use)."

No. I't 7W to power the laptop itself (if power management never suspends it). You will get more current if you are also charging the battery, however that means that laptop was working with no external power, so averge is still 7W.

"Is the report flawless? No. Is it totally comprehensive? Nothing can be. But if you look at their sources, and study their methodology and assumptions, its the best TCO analysis of computers in Indian schools that we've had to date."

This is like defending the result of posting KKK-sponsored paper on sickle cell anemia. The problem is not as much with particular errors with report or people's unwillingness to seriously discuss it but with why the Hell would anyone assume a Microsoft-sponsored paper to contain anything but Microsoft marketing in the first place. It's not stuff that comes from Microsoft Research, where Microsoft keeps people just so they won't work for anyone else, it's marketing-driven paper not unlike their countless performance and TCO studies.


> It 7W to power the laptop itself (if power management never suspends
> it).

I'm sorry, but no matter how often I do the measurement, I'm not coming up with a power consumption of 7 W unless the machine is suspended. Measurements of the DC current entering the computer are about 12 W at 18.5 V and 23 W at 12.2 V. This is a measurement of the DC current after it leaves the power supply. It does not even include the inefficiencies involved in converting a 120 V AC wall current into 12 or 18 V DC. Incidentally, the battery was removed in these two cases. I don't place much value in the power consumption, since the battery has to be charged through the DC socket anyway. (At least in most cases.)

> You will get more current if you are also charging the battery,
> however that means that laptop was working with no external power, so > averge is still 7W.

Uh, no. You're breaking the laws of physics there since no process is 100% efficient.

As for comparing this TCO study to accepting information from the KKK, are you nuts?

> I don't place much value in the power consumption, since

That should read:

> I don't place much value in the power consumption as measured from the battery, since

Wayan,

This discussion sees to miss the key point. This study assumes a model in which computer labs and computing represent an added course (ICT) in the curriculum. In this context of setting up a computer lab to support teaching of computing (Word, Excel, and Powerpoint), there may not be much difference between a desktop and the XO.

What sold me on the OLPC, is that it represents a tool for teaching the entire curriculum. The students are not learning to use a computer, they are using a computer to learn about the world they have been born into. In a South African school I visited last year, a child got 30 minutes access to a computer per week. How do you take a desktop home? With OLPC, the child can use the computer at any time.

This discussion reminds of the early days of the railroad. I am sure a TCO analysis at the time would have concluded 'get a horse'.

Teapot,

Of all the folks skeptical of Microsoft's intentions in the education space, I would hope to be damn near the head of the line. That's why I start off this post with my stated bias against them.

That said, they contracted good friends of mine, those that I would trust to be honest with their reporting, even if it was detrimental to their client. So in this case, I'll take exception to you comparing them to anything but the learned practitioners that they are.

In fact, if you look at the VWC pedigree, you'll note that individually, they were in the ICT4D space before most - maybe even before you and definitely before me.

Tony,

Good point, ICT should not be an additional course, but a basic way to educate. Either way, the introduction of ICTs into schools does require a basic level of expense. In addition, the learning model that has ICTs blended seamlessly into the classroom would, theoretically, cost more than ICT's as an additional course as teachers would require further training on how to do just that.

"Measurements of the DC current entering the computer are about 12 W at 18.5 V and 23 W at 12.2 V."

If it was true, 12.2V 23W would require 1.88A current. According to the label, XO power supply is rated only up to 1.42A, what would be 17W. Maximum current should include battery charging while running the computer at full load -- otherwise it would never pass safety testing.

"Uh, no. You're breaking the laws of physics there since no process is 100% efficient."

Computers are pretty close to 100% efficiency of converting electricity into heat (formally it will be 100% only if placed in a completely shielded room) -- the question is merely how. So whatever is not eventually powers the computer, is already spent on heating up the power supply, battery and voltage converters on XO board.

Switching AC to DC converter (what all modern power adapters are) is 60-80% efficient -- that is, a laptop advertised as consuming 50W, takes 60-80W from AC outlet, with "missing" 10-30W heating up its power adapter while 50W mostly end up heating up the air blowing through its vents.

"As for comparing this TCO study to accepting information from the KKK, are you nuts?"

No. Both organizations have approximately the same (zero) credibility when it comes to anything specific to the targets of their hatred.

"That said, they contracted good friends of mine, those that I would trust to be honest with their reporting, even if it was detrimental to their client. So in this case, I'll take exception to you comparing them to anything but the learned practitioners that they are."

Unfortunately for anyone who is not familiar with those people before (what would be everyone involved in this discussion except you), this is merely a fancy way of saying "trust them because you trust me". In my opinion, anyone who accepts money from Microsoft to produce a paper used in marketing against Microsoft's enemies, can not possibly be honest.

This -- a thinly veiled convoluted justification made to fit the predefined conclusion -- is the only thing that happened in the past whenever a Microsoft-sponsored "TCO" or "performance" study ever was released, and no one in his right mind would expect it to change now, when Microsoft is actually losing ground in consumer products, something that never happened in Microsoft history before. I would say that I am sorry for your friends selling out, but having to be honest I must admit that I don't know and don't care who they are and what is so special about them in the first place.

> If it was true, 12.2V 23W would require 1.88A current. According to the label, XO power supply is rated only up to 1.42A, what would be 17W.

I just butchered the XO's old PSU and tested against that, and it was drawing 1.2 A of current, or 14.7 W. I'm not sure why there is a discreprancy. I'm a physicist by training, not an EE. :/

> Computers are pretty close to 100% efficiency of converting electricity into heat

The XO makes a very poor space heater, and it's not intended to be one. Besides which, that doesn't change the bit about considering the machine's total energy budget.

"I just butchered the XO's old PSU and tested against that, and it was drawing 1.2 A of current, or 14.7 W. I'm not sure why there is a discreprancy. I'm a physicist by training, not an EE. :/"

With batery full or charging? If it's with full battery, power supply still runs out of specs when the battery is charging, otherwise what you see is both charge and actual power consumption of the laptop.

"The XO makes a very poor space heater, and it's not intended to be one. Besides which, that doesn't change the bit about considering the machine's total energy budget."

Actually all electronic devices are [nearly, see my previous response] perfect as far as heat production is concerned -- if you place a computer into a room that has an electric heater with a thermostat, the effect will be that heater will consume less energy by the amount consumed by the computer, and air will remain at the same temperature. The effect is small, may be dwarfed by other energy sources and sinks (such as you walking into the room, producing your own heat), and it may be counterproductive because of uneven heat distribution (such as modern mega-laptop burning your hands and not doing much for the air across the room), however neither XO, nor a typical desktop have problems dissipating the heat, so it is evenly distributed as a result.

Yes, it's true -- all energy-saving measures, from energy-efficient lightbulbs (unless you have high ceiling, so hot air from ceiling light fixtures dissipates heat through the walls and ceiling far above you and the thermostat) to power-saving modes on desktops and laptops, do nearly nothing in a room with electric heater that is always kept at the constant temperature higher than outside air when other electric devices are in use. This is why thermostats for office and industrial buildings have timers, and why thermal insulation is more important than the type of appliance you have at home.

Wayan: "Had it been anyone else, I may have suspected major Microsoft bias, but with them, folks I know and trust, I can rest assured that they did their best to be objective in a very subjective subject."

Perhaps you have misjudged the moral character of these friends of yours. That does happen, you know, even to the best of us.

Have you cross-examined your friends about why the report includes so many biased assumptions? Or do you believe that, because they are your friends, they should not be subjected to such indignities?

Your basic argument seems to be that, even though the report clearly has many biased assumptions, we should nonetheless pretend it is unbiased because the people who produced it are believed by you to be unbiased. Can you see the logical flaw in such an argument?

I would say that the correct thing for you to do at this point is forget about your friends, and re-do the analysis of the report with correct assumptions, and then see what TCO numbers you come up with. That would be useful.

I tend to check the power consumption with the battery removed, so no the battery could not be charging.

And yes, I am painfully aware that computers can be used as space heaters. This is the end of summer after all, and I refuse to get an AC, so the desktop was getting less use and the XO more use. And yes, I do use the desktop computer to warm things up in the winter.

Still, computers aren't intended to be space heaters. And while turning them off, switching to CF bulbs, etc., during certain seasons may not make a huge difference in terms of overall energy use, they do make a difference during other seasons. (Whether having more energy efficient computers, light bulbs, etc. makes much of a difference to the environment when there are far heavier polluters is a different issue.)

Wayan,

"this report was developed by friends of mine.. folks I know and trust"

It doesn't matter whether you source your information from your friends or otherwise (after all, it didn't stop you use MS's 'fake video' PR guy James Utzschneider as your source as well, as long as it supported the line of argument you wanted to put forward...).

What matters for readers of your blog, like myself, is whether the info is true/accurate and, as always, the deciding factor is the transparency of the original info so that the basic facts can be verified. This is not the case with this report (e.g. where are the supporting school case studies ?).

There are many factors, as per my previous posts, why the report, and other ones like that, has little relevancy as far as the deployment of 'computers' like XOs are concerned. So forget about drawing any conclusions about XO's deployment costs from it. However, what caught my, and others' it seems, attention is the report's finding that the cost of MS-based computers and OSS-based is the same. This is just another re-hashing of very similar MS-sponsored findings in their very well known PR efforts.

Have a look at the report by, somewhat more established, consultancy group Becta from UK [1], based on school case studies - even for traditional desktops and the fact that in the last few years, since the release of their report, the range and the usability of the OSS software has increased dramatically, they still concluded:

"For OSS schools, cost per PC computers at primary school level was HALF [emphasis mine] that of non-OSS schools"

That's HALVING the cost just by using OSS software (in your friends/MS report cost is the same...) for a traditional desktops, even without further big cost savings from software being integrated like in the case of OX machines...

[1] Becta Research report: Open source software in schools:
A study of the spectrum of use and related ICT infrastructure costs
( http://publications.becta.org.uk/display.cfm?resID=25907&page=1835 )

Wayan:

"Had it been anyone else, I may have suspected major Microsoft bias, but with them, folks I know and trust, I can rest assured that they did their best to be objective in a very subjective subject."

PR 101: Get independent researchers to support your product/point
1 Research the question in house for many use case scenarios
2 Select the use case that gives your product an advantage
3 Select a small, well reputed outfit with honest people
4 Let them research your best use case, with the relevant limitations
5 Liberally quote the conclusions with bothering to qualify the real use case scenario

Here we see what they have done quite easily. The use case is a school computer lab. Why, because children are unable to manage their own XP laptop to any extend. Furthermore, they refrain from quantifying the time needed to support the computers, just noting the fact that Linux support personel is more expensive. But each of them can manage much more computers than a single MSCE. Also notice the 5 year window. That was not chosen by "accident".

In short, if your friends were completely honest, they would have been worthless to MS PR. But they are simply used.

Winter

Look, when you do a TCO study you have to make some basic assumptions. These assumptions will be based upon various criteria. In this case is was computer use in a lab environment. It is a realistic criteria because the $200 laptop is too costly for many places to implement a 1:1 program. In this case, it is even more realistic since they appear to be basing the cost estimate upon the cultural norm (of computer labs with lab managers). At this point, it has nothing to do with propaganda, or exploiting people, or any other such nonsense that a Microsoft basher puts forth.

Jordon,

But Wayan didn't say that the report was for a computer lab. Instead he said "Or almost three times as much as OLPC News calculated two years ago using only the collective estimations of the OLPC community." But those calculations, and olpc's in general were for 1:1 deployment.

Look, Wayan screwed up. He should have made clear that the report was about a different deployment model than the one we and olpc have been talking about.

Jordan,

"Look, when you do a TCO study you have to make some basic assumptions."

Sure, and the types of assumptions you make can completely change the actual results.

In the report [1] you'll see the largest proportion of TCO involves salaries (ie. services) to be paid to various people involved. In the report I could only find teachers' salaries - the report says [1 - page 17: Assumptions and Calculations]:

"According to a UNESCO study of 19 developing countries, teachers with 15 years experience earn a median salary of US$14 per hour in secondary and US$11 per hour in primary schools"

Now, if you actually look at that particular UNESCO study [2 - pages 126-127] you can see that UNESCO doesn't use US$ (as Vital's report misquotes) at all but so called relative PPP$[3] which reflects differences in the cost of living in different countries - i.e. prices of non-traded goods and services in case of developing countries can be many times lower.

The result: a complete distortion of the actual figures as specified in UNESCO study...

[1]Vital Wave Consulting: A Total Cost of Ownership (TCO) Model for Education Officials
( http://download.microsoft.com/download/2/0/a/20ac945c-34d0-4a60-8245-f80e80fe954f/Vital_Wave_Consulting_Affordable_Computing_TCO11June08.pdf )

[2]UNESCO(20707) Education Counts: Benchmarking Progress in 19 WEI Countries, World Education Indicators – 2007.
( http://www.uis.unesco.org/template/pdf/wei/2007/WEI2007report.pdf )

[3] World Bank - Relative prices and exchange rates 2005
( http://devdata.worldbank.org/wdi2005/Table5_7.htm )

@Jordan:
"At this point, it has nothing to do with propaganda, or exploiting people, or any other such nonsense that a Microsoft basher puts forth."

But that single assumption, "(of computer labs with lab managers)" with a five year window and ignoring support efficiency completely defines the (very high) cost. But it is not mentioned in the title, and not in the executive summary. It is not "A Total Cost of Ownership (TCO) Model for school Computer Labs". So this is putatively a TCO study on computer use in education, ignoring the 1:1 option completely.

They do quote:

"Ultra-low cost computers and Linux-based solutions are relatively equal in cost to traditional hardware and proprietary software solutions because they require higher labor and replacement costs over a five-year period. "

But these mini-notebook have not been designed for computer labs. I think a computer lab is the worst place to put a mini notebook. So why mentioning it?

You won't discuss bicycles in a TCO of transporting building materials. But bicycles are very good transport options for children.

So by selecting a use case that is completely unsuitable for "Ultra-low cost computers" and ignoring support productivity of "Linux-based solutions", MS have ensured that the costs of their software have been made invisible in the other very high structural costs.

Anyhow, why do you think this study drags in so many other costs to show that the licensing costs of Windows become insignificant? That way I can make a point that you should not care about the costs of pencils because the building and heating are so expensive. So please buy my ultra-expensive low quality pencils.

Winter

Forget about this study. Forget about Microsoft being or not being evil.

The bottom line is very simple: no responsible country will ever rush to implement a very expensive (ANY deployment will cost a lot of money per pupil) and UNTESTED proposition.

Common sense would indicate that IF (big IF) giving laptops to elementary school students were a great idea, rich countries would have done so a long time ago. The fact remains: poor countries like Peru or Uruguay, for reasons ranging from utter political corruption to sheer ignorance, are doing something very harmful to the future of their children's education. The USA can waste untold amounts of money on anything. Poor countries can't do the same. If they (the poor countries)are smart, they will wait until the positive results, if any, can be verified.

In short: even if the TCO is $200, the money should not be spent until there is proof the kids will benefit form the investment.

"In short: even if the TCO is $200, the money should not be spent until there is proof the kids will benefit form the investment."

Go, troll Wikipedia, or something. This entry exceeded the troll limit long before you have noticed it.

Why all the anger, teapot?

:-)

Surprise! TCO for XOs in schools is $437.50 (4.5 cents per hour)

I have attempted to recalculate the TCO based on a more realistic description of the OLPC model, but using the numbers in the report.

The TCO model in the report is based on teaching ICT in a public secondary school. The purpose of this training is to prepare university-bound students to enter a course of study leading to a position in the computer field. Assume that the school enrollment is 160 students. The study assumes 3.5 hours lab time per day, 5 days per week, for 39 weeks per year (10920 available hours per year). Shared among 160 students for an average of 68.25 hours per year per student or 1.75 computer access hours per week.

The OLPC model is based on each child having an XO throughout the class day and at home on evenings, weekends and holidays. It is also focused on primary school education. Assuming a school of 160 students, at least 160 XOs will be needed. Assume a steady-state situation in which each child entering the school in the second grade gets an XO. Each child keeps his XO throughout the primary years (grades 2 thru 5 as assumed in the study). The school purchases an additional 5 XOs per class to provide for loss, damage, repair, incoming transfers, and the teacher. In addition the OLPC model assumes a dedicated computer (XS) per school which provides internet access, provides backup for the XOs, caches instructional materials, and supports a CMS (Moodle). Each child has a computer which could be used each day for six hours in class and two hours at home (plus ten hours on weekends). This is approximately 1800 hours per year - let's assume 1000 hours.

First, look at the direct computer cost, excluding initial setup, training, connectivity to the internet, electricity, and Tier 1 (software) support.

Case 1: Public Secondary School (grades 9-12) sets up a computer lab with 16 desktops (one is also a server) to teach ICT. The total cost is $8000 with a per seat cost of $125. This is a approximately $0.40 per hour for computer access ($125/312).

Case 2: Public Primary School (2-5) provides each child and four teachers with an XO. The initial purchase for this school is 160 + 4 (teachers) + 4 (reserve) for a rounded-up total of 170 at $200 each ($34000). The annual buy is 45 laptops at $200 per student ($9000). A dedicated server per school costs $500 (headless mainstream desktop). This is a total cost of $34500 ($216 per student). If we assume utilization of 1000 hours per year, the cost is($216/4000) or 5 cents per hour. Note: if leaving students keep their XO, the cost in the first four years is $34000 + $27000 (3 * $9000).

Other costs:

Case 1 the computer lab pays $33886 total for initial setup ($5351), training ($10620), software support ($10920), hardware support ($2432), electricity and internet connectivity ($5351), and damage and theft ($2640). This increases the computer lab cost to $2500 (41886/16) per seat and approximately $2 per computer access hour ($41886/16/4/312).

In case 2, the costs of Tier 2 and 3 (hardware maintenance) and the costs of damage and repair are handled by purchasing extra XOs. In the OLPC case the extra cost is $28314 (excluding $5072). The OLPC total cost is $70000 ($41500 + $28316). This is $437.50 per student or 4.5 cents per hour.

Wayan, how about going through Tony's analysis in detail and explaining where you think it is right or wrong?

@Tony,

Great calculation.

Note that there have been other TCO calculations discussed on OLPCnews. I will look up the link later. I think teacher training can be organized cheaper in a large roll out.

Winter

Eduardo, there is no sensational, headline grabbing, OLPC bashing reason that Wayan would want to do that.

Working out Total Cost of Ownership for implementing computers in schools is a little like working out the cost of saving the environment.

@Tony,

I looked up the link, it is:

http://www.olpcnews.com/sales_talk/countries/indonesia_laptop_payment_plan.html

Down in the comments, there is a lot of discussion about TCO. It shows, again, that those who do not learn from history are doomed to repeat it. Reading that discussion gives you the Deja Vue of the the current one.

Anyhow, I did there more or less the same calculations as you did and I came down to $465 per XO plus internet connectivity. But I did not have the above report to get the "real" numbers. I think I overestimated the teacher training and hardware replacement costs.

@Robert Arrowsmith,

Wayan is not OLPC bashing. He actually supports the 4P/OLPC mission, and with real effort (this site is evidence for that).

But Wayan has consistently made the point that the OLPC was seriously underestimating the cost of XO deployments. I suspect that he thinks the OLPC is leaving out the real numbers on purpose. And I agree with him that the real costs (and benefits) still have not been published (or are even known). I see his comments on the report as a way to get these costs visible.

This TCO report is completely bogus (as every comparison paid for by MS has been), however, it does give you an idea what to use in your real deployment calculations. Tony's calculations are a very good illustration of what can be done by applying these numbers to the XO.

Winter

Below a repost of my calculations on
http://www.olpcnews.com/sales_talk/countries/indonesia_laptop_payment_plan.html

Note that Irvin called himself Robert then. See how little things have changed in 10 months (it looks decades ago). :-)

===============================================

The TCO question comes up time and again.

Below I will give a "plausible" TCO based on pure guesswork. To understand why these TCOs are so way off you need to understand the cost difference between wiring up a single school and training a dozen teachers by US personell, and wiring up 10,000 schools and training 100,000 teachers by local people.

An Indonesian teacher earns up to $1000 a month. A local trainer will be rich if she earns $2000 a month. Applying a US ex-pat might cost $10,000/month and more. Setting up 60 computers for the first time is hard work, setting up a million imaged XOs is a routine chore.

Below I gues that to come up to a TCO of $1000, the internet connection must cost around $5000/year. For one school, this is mostly correct. For 10,000 schools you definitely can get a better deal. Say, by a special deal with a mobile phone provider.

"TCO
$108.00 Setup (1-time fee)
$27.60 Yearly training (multiplied by the tco period)
$135.00 Yearly internet access
"

My own TCO guesstimates:
Millions of XOs, so fixed costs can be ignored. Only variable costs will add up.

6 year deployment of XOs
4 year average lifetime of XO (breakage and loss)
1 week training per teacher per year
30 pupils per teacher
1 server per 60 pupils, expected lifetime 3 years

1 School server $600 + $300 shipping + set up

Training:
1 Trainer earns $500/week, trains 25 teachers "locally"
1 Teacher earns $250/week
150% overhead for additional costs and on-line follow up
Training costs:
Training ($500/25 * 2.5) = $50
Teacher's salary = $250
----------------------------
added $300 * 6 years = $1800
Per pupil: $1800/30 = $60 TCO

Setup per XO: $5
2 servers: $1800 /60 pupils = $30

Together:
$200 1 XO
$100 0.5 XO replacement
$5 XO setup
$60 1 week training / year for 6 years
$30 School server
------------
$395 Computer infrastructure

Electricity:
30kWh/year per XO = 180kWh
10Watt * 10 hours/day * 300 days/year
(note that the charging power is higher than the consumption power, I guess in the order of 10W charging for average use)
The current production price of electricity on Java is around 200 Rp/kWh ~ $0.022/kWh
Thus, 180kWh/year * $0.022/kWh ~ $0.4/year
But let us take the cost at $10 a year on delivery for a total of:

$60 Electricity

Networking:
$10 6 mesh repeaters per 60 pupils in total $600/60
$X Internet connectivity, one connection per 60 pupils

Together:
$465 + $X TCO

To get at $1000, they need to pay around
60 pupils * $500 / 6 year $5000/year on internet connectivity for 60 pupils.


Winter

let's clear some things up here. The OLPC/XO is shooting for a 5 year life of the product. Yes, their first production model has had some problems but the goal is getting 5 years out of the device. On the other hand, Microsoft loves a 3 year cycle and their increasingly bloated OS cycles show that. The other things leading to a 3 year cycle for the "standard" PC are the moving parts and high heat generation of those computer parts. There are fans on the powersupply, CPU, and often the video GPU if not another case fan. Then there's the hard drives and CDROMs. So while the XO might not have 5 years, it is a target and the design is sound enough to expect them to get that.

So the 3 year cycle used in the TCO paper is too short for the XO.

Next is the high cost of support and training. The OLPC/XO software( Linux/Sugar ) is designed to be easy for children to learn and use in their classroom assignments( daily assignments with only a few days at most on one assignment ). The Windows user interface is not designed this way. Just look at how Microsoft keeps changing that user interface and each time they say they used experts and studies to proof the next new interface is the correct one. Sorry but they didn't spend time with children or else there would not be a half dozen ways to create and save files and to find where those files went to. Training on the XO is designed to be limited and communal as the 2nd year students can easily help the 1st year students and probably some teachers too. There will be kids who thrive in this part of the deployment and they'll become or build the support element. Windows and the PC hardware are way to varying, complex, and actually fragile for this to occur at the primary school education level. I know of software developers spending days fixing Windos XP systems which just stopped working because of a registry failure of some kind. Just powered on the next morning and stuff stopped working and all the embedding of software in the OS has made it fragile. Vista is worst but then again, though it is new, it can't even run on these low power devices.

So I see the TCO numbers inflated with regards to training and support when the OLPC/XO device is concerned.

I'm also surprised to see the $2,700 over 3 years mentioned here, and OLPC forum when so much of that is about Windows or at the very least full blown PC distributions( Windows and/or Linux ) and not a specifically targeted solution as the OLPC XO with Linux/Sugar.

And who does this kind of thing benefit anyways? I high TCO for the PC justifies the relatively high cost of Microsoft software and it is pretty obvious why this report was commissioned and by whom. We in the industry for more than 5 years have seen one company do finance this reports over and over again. And some of the anti-Linux ones were real doosays too with obvious manipulations in some areas to show an advantage for Microsoft.

Tony,Winter,

"Surprise! TCO for XOs in schools is $437.50 (4.5 cents per hour)"

Would be great to have your calculation comments combined and put as a separate post rather than buried in the comments section - and a great title too :)

As you have probably noticed, Wayan completely ignored the comments here showing the report to be bogus and still insists on using the 'rounded down' original figure in his newer posts.

MS's marketing PR guy James Utzschneider must be quite happy as Wayan's post is now spreading through the Internet (in the true FUD fashion), although for some reason James now removed from his blog a very agreeable comment from Wayan.

I still feel Bryan Berry, with his real life experience of XO deployment in Nepal, would be the best person to contribute on the subject...

What a bunch of pure bogus. If you can shout out $2700 as some kind of official Microsoft study, you should be able to list basic costs.

The only cost is the $100 laptop. Teachers if there are any, are paid anyways the same they always would be paid. Electricity would be free, considering you would need electricity to the school anyways. So install solar panels, wind or whatever you need, you will need to use that anyways. Otherwise the children charge their laptops with a $10 portable solar panel or the $10 yo-yo.

XO have very little quality control problems compared to your $1000 Dell and Microsoft reference design laptops. It will last much longer then normal laptops as we all know.

Google is sending up satellites that will provide nearly free wireless broadband to whole continents. Simply use one $20 solar powered WiMax modem per school until then. Internet connection is free, the government CONTROLS IT, it's free for educational purposes. Or less then $0.20 per student per month. It's mesh remember?

Content is done centrally by the government. It's free. Government spends money on making curriculums ANYWAYS.

Anyways, this post is complete bogus.

We can see where this Vital Wave report was targetted:

http://www.theregister.co.uk/2008/10/14/ncomputing_indian_schoolkids/

>>>>>>>>>>>
Andhra Pradesh will install 10,000 PCs in the state. Each one will run five virtual machines based on NComputing technology. Acer is supplying the desktops.

"NComputing is proud to have been chosen by Andhra Pradesh to fulfill its vision to improve learning and computer literacy throughout the state," said Dukker.
<<<<<<<<<<

Winter

Close