Reply to Message

Hi, what can of worms You opened!
It is something that fell pey to Murphy's Law: If there is 11 ways to do something and only one way would screw it, then it would be done this way for sure......
First total synchronization is not practical nor possible. It would waste too much bandwith to even try to sync movies, and surely all devices have different size of storage, so while You can download terabytes of movies to home based PC, it would not be practicaly possible for smartphones, for instance. Also, bandwidth cost would be huge, so forget it, at least for now....
In regard to security of the clouds, that is easy to implement with three way network, where even distributed databases could be safely stored. Simply, each new data is written three ways, and that include updates of existing data. For reading data, it could be distributed and only one out of 3 groups would receive read requests, for data that are primarily stored there. That take care of hardware failure as copy of data would be on neigbouring computers in the net, and could be used to restore original HD content on new drive.
Security from malware is as simple to implement, but then all Antivirus industry would lose jobs, which does not matter since mayority of them actually produce viruses and let them spread to stay in business.
It could be done either by precluding viruses to leave their point of origin, or simply preventing execution of any program without permision of user who installed it, at least first time, and checking the program for bad intentions. Programs should be able to acess only files they create or those choosen by user, they should not be allowed to change directly anything, and that could be done using OS services which could control what program is allowed to do and what it is not.
Finish problem!
Now many compare Cloud computing with mainframes, and as one who worked on mainframes, I agree that it is similar. It has its good and bad aspects, where good would be that it would be possible to have at disposal all latest upgrades of program without need to pay for it, and it would be more profitable even for software wendor and program users to have "pay per use" system, since if it would be cheap enough then there would be no pirate copies and unauthorized use, and for those who need expensive aplications that in turn require expensive hardware to run it, that way it would cost less. Same would be with movies, as nobody would mind paying few cwnts per movie and not have to hoard them on their computers and worry about losing them because of hardware failure. That way movies would have the worldwide audience, and so would songs and books. Who would like to look, listen or read same thing again and again, or to have hardcopy at hand, they still would be able to download what they like and pay for downloading that to their devices.
Applications would have to be assigned to devices where they are used, so there would be no need for special hardware for regular users.
Basicaly computation intensive work should be done in paralel using Cloud computing same way as formerly mainframes were having Batch processing, as in the end user is preparing and sending the data and want to see the results, and both can be encrypted so nobody would intercept and steal them. As acess to user owned devices would be in principle restricted to owner, and all files could be encrypted as well, nobody but owner would be able to acess them even if they are stolen.
Even stealing of personaly carried devices could be precluded with addition of mobile phone characteristics to each of computing devices, as they can automatically report that they are unauthorizely moved or just report their position if owner report them stolen, but since three way security for the data no data would be lost, and stolen device could refuse to work for one that have stolen it, or even self destruct (but that would be dangerous, so better we should have the "passive resistance"...) or at least report its position to nearest police station.

On Clouds would have to reside all public information systems that only provide read acess to their data, and same should go with movies, songs and books that could be downloaded for one time use (cheap) or for many times use (not so cheap but not much expensive also, since then nobody would find it profitable to send such movies to their friends, specially if they would be coded for use by this one user and nobody else. ) That way user who has paid for permanent use of movie, song or book can make as many copies for own use, and only complication would be in case of books that could be printed out or if buying printed version is possible instead of being able to print it from digital form purchased, it could still be photocopied (unless this is also precluded by special paper that prevent photocopying) so author's rights would be broken..........

So this is how everybody would profit from Clouds use.
Undecripted synchronizing makes data vulnerable to cyber stealing, therefore data could be kept for backup purpose only heavily encripted, and things user has purchased would not need to be backed up as there would be records of purchases and in case downloaded copy is lost, user would be able to download same movie, song or book as many times as necesary, but only after data loss is reported. Program can check such statement also if buyer configuration data is also recorded, as it would be simple to confirm that HD to which purchased intelectual property is downloaded is no longer there while the rest of configuration is. But with three way storage of all data this would not be necesary as data would be recovered internaly.
So in the end Google way is half right as it would never be necesary to keep all data in the Clouds and people would like to keep their privacy.
Same Way it would never be desirable to synchronize data files for all devices used, so there must be a selection by user which date should and which need not be synchronized.

As an old hand in programming of computers, I am worried how much things computers do unnecesary, like for instance copying files that were not changed during backup, reading again WEB pages that were once read (at least same day) since they could not have changed in the meantime, and contents were downloaded into temporary folders anyhow. It was fascinanting also to observe how HD capacity was growing and was able to store barely same data because cluster size was growing and file that was smaller than one cluster would allways spend whole cluster no matter which size it was or is.
As one that started to work on "mainframe" with 32Kby of RAM, I am horified to see bloatware which is created those days, starting with OS-es. I am having over 4 million files on my computer, half of which is backup on separate HDs, but this is entirely too much and I would say 9/10ths useless.
It also slow down computer use, specially if application is larger than RAM available, but we were using our 32 Kby effectively by swapping program segments from disc if necesary, and there are allways parts that are used just once in any application, there are functions that are not used together nor could be used together, so application should configure itself to work user intend to do, not loading all and sundry and then also swapping it all with remainder of program regardless if this part is needed or not.
Same way I am surprised at Cloud technology and its huge capacity that is generaly unused, save in special cases. I have ideas that would solve problems in paralel processing, such as implementing "virtual time" during program compilation and deciding which computing operations have to be done in same virtual time period, that would not depend on each other results. Then all that is left is to spread that "virtual" time slice to as many cores and computers as is necesary to execute them in same real time clock period in paralel. I could make algorythms that would do graphic operations 1000 times faster or even more, because there is more and more "bruta forca" style of computing instead of inteligent use of resources available.
But then, who would need Cloud computing if just a single workstation could do what is required?
One thing is certain: Clouds have enormous potential for alll kinds of data processing, but so have system of distributed processing that use some scientific projects like the "SETI at Home", tapping into wast unused resources on personal computers around the world. But, again someone screwed good idea making programs that were just "number crunchers" instead of searching for extraterestrial inteligence inteligently.
Well, maybe it is just me, same as at time when I solved old (and according to some, still unsolved ) problem of "Traveling Salesman". Everybody considered that number of solutions grew exponentially with increase in number of cities, where I concluded that for any given number of cities there is ONLY ONE solution of finding shortest path between them, as difference could be only which is starting/ending city of the tour. So once shortest path betveen cities is found and cities are ordered by sequence of traveling among them, all that is left is to move table sequentially in circle untill choosen starting/ending city is at top and this is exact solution required for such case.
Even on my ATARI 520 PC working on 16 MHz I was able to compute results for 100 cities in under 3 hours time. Official science claimed that it would take at least 100 years of non stop computing to get exact result :-)) That is when I invented "Fractal Adressing system" which assigned proper sequence to all points inside some coordination system, according to number of cities that would be used. All that was left to do was to sort coordinates of the Cities according to their sequence number and half of problem was solved, what was necesary next was to resolve any path crossing between cities that have occured because of natural convoluteness of fractal used.

Regards from Croatia, the homeland of one of greatest inventors from 19th and 20th century, Engineer Nikola Tesla!