An alternative anti-theft mechanism for OLPC

FAQ addition:

Q: Isn't observing the user spying?
A: Not if the observations aren't sent to anyone. In my humble opinion, contacting an authority every day a laptop is used is closer to spying.

Q: What if a child wants to let its parents use the laptop for a while?
A: Because we're not trusting the biometrics and asking other laptops for their opinions, and only shut down if the friend laptops aren't seen for a long time, this will only be a problem when [b]most[/b] children decide to let their parents use their laptops, all at the same time!

... said the guy whose mobile is always switched off

Hi,

You talk about how the laptop requests permission to work (which is true) and will shut down if it doesn't get it, and then in the next paragraph you talk about the problems of humans filing theft reports.

Why does a human need to file a theft report? The laptop will shut down after a few weeks of not receiving a new lease from the school server; there's no theft report involved in that, and certainly no OLPC inspections.

A theft report is optional if you want the laptop to shut down immediately rather than when the lease runs out.

- Chris.

Hi Chris,

You're right theft reports are optional. Not having the possibility of theft reports would close quite a can of worms, and it would be odd theft reports (especially lots of laptops) that would cause OLPC inspections.

This suggests that "get rid of theft reports" solves the problems. Not so.

If a theft isn't reported, one could in principle steal a laptop and let it continue functioning by taking it near the school every three weeks. One could even give it internet access through some other access point than the school, but maybe OLPC can detect it's connecting from the wrong place. On the other hand, it's possible some children who live far away from their schools will make legitimate use of an "unofficial" access point. Their laptops shouldn't be disabled.

It's possible a certain school won't have an internet connection for more than three weeks (or whatever period is chosen). To prevent the laptops from disabling themselves because of that, one would need to give the teacher some kind of password or device that resets the anti-theft mechanism. Both passwords and devices can be stolen... Let's try to fix that.

Maybe the passwords should be usable only one time? Then the problem reappears after six weeks. How about letting the teacher make a phone call to get the next password? ... but if you can make phone calls, you can have (slow) internet! Send the password by mail? Do we trust the mailman?

What about a secured device that can only be used by the teacher (for example it may contain a fingerprint reader)? Then what if the teacher gets replaced by another? And how difficult is it to modify the security mechanism of the unlocking device? Make the device password-protected? Oh wait... never mind.

Thank you,

TB

... said the guy whose mobile is always switched off

Very complex situation! It is due a more deep understanding/reading (by me) to get the whole thing. But just like a rapid sugestion I could mention one possibility: Make use of the ever present - in poor countries - AM (amplitude modulated) radios operating on the 550 khz to 1600 khz band. To implement a micro receiver to AM/medium waves is trivial. Generally, governement has statal radios que could broadcasting (hiden into the radio modulation) codes that could turn on/off one specific OLPC.

Just my two cents.

I have some experience on long distance traveling in
Southeast Asia somewhere with trucks -> problem of
pirates, robbers, and the like. A few, hopefully useful,
analogies might be made.

** Truck's drivers don't travel at night in given areas,
they stop at restaurants they know--which means: they
don't drive beyond restaurant[i] past some time in the
afternoon/evening, when the next known restaurant[i+1]
on the street is too far.

Laptops for children do not work past some given time
at night (with exceptions--see next).

** Under circumstances, not as method and in a subset
of the above mentioned areas, drivers may take the
risk and drive at night in convoys of two (a bit
dangerous still), three, four trucks--every truck
driver knowing well most of the others (two drivers x
truck in general).

A similar model can be made to use at night the laptops
for the children, exceptionally.

** Too odd patterns. Debris (rocks, branches, etc.) are
put on the asphalt, so to force a truck (20-35 tonnes)
to slow down as needed, and let it begin zigzagging;
the truck passes under a selected tree, or under a not
too tall bridge; the robbers jump on the truck from
the branch of the chosen tree, or from the bridge;
cut the tarpaulin; if there is no metal net under the
tarpaulin, they start grabbing merchandise and throwing
it from the top on the side of the street (forest,
given spots), for maybe 5-10-15 kilometers, till a
complementary zigzagging is met, for leaving the victim
truck; a pick-up follows, minutes later, to collect the
items. The same can be achieved via plenty of others
means and tricks. And a partial solution is: convoys
again, plus frequent rotation, so to *protect* the last
truck in the convoy [you can typically see nothing, or
very little, of what happens on your truck behind];

A similar model can be made for the laptops of the
children: behaviors which deviate too much, and for
too long [hours], from some typical children use, need
peer's [laptop for children] validation/input.

** And so on.

Anyway, too banal anti-theft schemes would lead to denials
of service quite likely, deliberate and involuntary, till
the every given mechanism becomes disabled eventually. The
article intelligently underlines or hints at it. Single
point of failure are generally bad from the perspective of
*securing* something.

Beyond the fact that reporting thefts is definitively NOT
peanuts [and that for m reasons] in developing countries,
calling teachers at just every glitch, doesn't take the
problem of *corruption* under many forms (I could tell
enough *real-world* [developing country] examples),
which it exists at a _different_ level than in developed
countries, and be it only as a game of power to become
more influential fast and make a better *career* at
the school, into account; nor does it considers the
administrative burden (time, e.g.) and the ambient
situation.

I slept with this problem and this is my dream :)

SXO - propably stolen laptop (Stolen XO)
NXO - laptops from old neighbourhood(Neighbourhood XO)
UXO - XO user
XO_ID - identification number of every XO

After 3 weeks out of known neighbourhood (long holidays?) send to old neighbourhood (NXO): Am_I_Stolen(SXO_ID, UXO_Name, UXO_Surname, UXO_Nick)
{
if answers<5 then wait 1 day and ask again (7 times?, then go to eternal sleep)
else
{
if Stolen > NotStolen then
{
I'll shot my head
Buum!
Dead
}
else
{
I'm not stolen
Start learning new neighbourhood
}
}
}

Am_I_Stolen(SXO_ID, UserName, USerSurname, UserNick)
{
if my neighbourhood is OK then
{
ask my UXO: do you know: SXO_ID, UserName, USerSurname, UserNick?
if YES then
{
if (SXO was stolen?)=true then
{
send to SXO: YES, YOU ARE STOLEN
}
else
{
send to SXO: NO, YOU ARE NOT STOLEN
}
}
else
{
do nothing
}
}
else
{
do nothing
}
}

Of course it's only idea.