Besides the lack of security the reason why the worm  ILOVEYOU used Outlook on MS-Windows as a distribution channel is that both have strong (if not major) presence on ordinary user workstations everywhere. Other email clients could perhaps be challenged as well in similar way, b.t.w. PCs which used different mail clients but still were able to execute the *.vbs file could get corrupted themselves too, only protecting their friends by not sending the poisoned message to them.

Other operating systems and email/groupware systems perhaps utilize better level of security but still keep the above written paradox principle of programme access rights to the data at the same extent as the user has. One may after some effort to shield certain computing environment eg. JavaScript from the other PC environment but it also limits the usefulness of both systems when they cannot interchange data freely. How useful would be perfectly encapsulated e-mail client if user cannot process the attached file in any way on the PC?

I suppose it is the right time to throw away all currently existing operating systems and create the new really secure operating system (or more, in order  to compete).

If we can  digitally sign the data then it is also possible to digitally sign all the computer code,  all programmes incl. libraries and object classes and let the computer execute only those which are created by the trustable authors. Such functionality could be built into the very centre of the OS kernel and inherently supported also by all applications which utilize some level of automatization (macro and script languages). Thus OS kernel could prohibit the program execution to both administrators as well as the ordinary users when the code author is coming from ucertain sources.  At the same time allowing those automatization elements which provide necessary productivity of those created by known and trusted program authors.

PCs proved to be preferable choice by the users. The NetPC concept did not find enough interested users yet. Poeple like the freedom to have a complete computer power, incl. disks and certain ability to operate the programmes of their own choice. But current PC model was created around the year 1980 when nobody had the long term view of hundreds of millions PCs connected together nor economically acceptable still powerfull technology to build-in enough security.

So how about creating a secure PC now? The glue which keeps the user loyalty with their current OS is not the system itself but the data format of the most often used applications and the familiarity with the application control layout. One may easily switch to other car as long as it has the similar steering wheel and use the same gasoline and backed up with the same gas station infrastructure.

It is likely that the estimated damage $1,5 billion caused by ILOVEYOU should be considerable enough amount of money to create such basic, secure operating system for current PCs from scratch. Even the single states like USA, EU or China are rich enough to order and finance creation of such secure PC system from their public budgets. If not them the large corporations are able to create user syndicate and force computer vendors to create such an operating system. Even the current big computer hardware vendors could be interested to join and create a new SOS because their market share need not be seriously damaged by such a step. Other potential leading customer could be Pentagon. Similarilly to GPS, the side effect could be that it will be made available to public. Thanks to the current market fully accessible via Internet, all currently existing applications could be ported to the SOS for a reasonable price.

According to my memory the virus war started somewhere around 1985. The first lecture which I attended concluded  sceptically that the war shall be never ending with the virus pirates one track ahead. The advices were:
1. do not use software from uncertain sources
2. use up-to-date antivirus programmes, and
3. backup regularly
are still valid.

At present time,  the maximal defense architecture follows these rules (see picture 1). Classical firewalls protecting intranet computer systems are combined with the content checkers, among them the most often used are the antivirus scanners. Firewall content checkers may monitor all the traffic, both email as well as the web browsing. Virus scanners can be further installed on inner email servers as well as on every user workstation. Public ISPs usually don't scan email boxes for viruses and leave the  entire responsibility of their individual customers. Next, the modern virus scanner programmes are able to update their virus database from the central antivirus company archive on the daily basis or even more often because they are connected on-line over the internet.

Despite the current level of uneasy sophistication this architecture totally failed in the case of ILOVEYOU virus simply because the virus spread by one order more rapidly then the reaction of the antivirus tools did. The virus got wild at least 5-10 hours in Europe prior the central virus databases of major antivirus companies were updated to catch him! In the next hours these antivirus centres got congested. Considering the future it may get only worse. The increasing internet bandwith may cause that the next generation striking virus will get down to knees half of the computers prior the human dependable antivirus intelligence shall be able to analyse the virus and react apropriately within 1-3 hours.

The virus detection has another drawback -- the number of the viruses is increasing and internet may encourage more and more authors to let them fly. The virus pattern databases are growing, and the mutations need not be programmed by the author himself but by his "human" followers. Detection may cause false alarms too. In my point of view ILOVEYOU demonstrated that we are near the end of practical usability of virus detection approach. We shall continue to use them but the basic trust has been lost.

The data backup saves, but does not prevent the damages for the wasted time spent by restoring the systems and data.

The next possibility is to inhibit all active content coming from the internet which shall undermine its further development in all applicable areas and is undesirable by the entire computer industry.

Well this is why the remaing option is to concentrate on the origin of the program source to define its privileges independently from the typical user rights who uses them - and that is the essence of the above mentioned SOS.

Viruses are not the only reason why we have to concentrate on higher computer security from the very bottom. The complexity and computing power are ever increasing and causing more potential danger. See the Wired's article : "Why the future does not need us" describing how the future computing power increase seems to follow Moore law and in next 20-30 years may create our entire dependency on the computers. Combined with darwinian market forces it could lead to the extinction of the entire human beings if not all of the current planet's life. Due to these possible threats it is necessary to concentrate on new, more secure architecture of computers, to let them be more contrallable by poeple.

In order to ensure all public users that the computer can be fully controlled by them I suppose that the security kernel of the operating system should be an open code or GNU licence-like. For the same reasons of user loaylty mentioned above I believe that the main players in computer market shall not be influenced significantly by that. The end user is only attracted to their application software and hardware brands they provide. Open kernel will help to convince the customers that the computer on their desk is not the hidden enemy but the controllable tool.

The concept of SOS creation is quite real and it will not take more then 2 years plus one year to port all major applications.

The GNU concept of the SOS is interesting also from other security reasons. The creators of the commercial products often tend to develop intentionally complex APIs. Their learning and usage is acquirable only by  larger software vendors, which decreases the uninvited competetion entry. But complicated, quickly programmed and unsystematically appended APIs cause the software errors even within the software of the companies that created them. The innovative cycle requires to append more and more new, unnecessary features which increase the total size of the software based on the original APIs thus containing more security holes. In other words - the moves taken by strong players of market economy, ie. expensive entry ticket plus marketing diversification are going against the security demands and the user interest. From this point GNU rules bring less greedy concepts in software design and helps to improve security. The good engineering practice is to keep things as simple as possible because simple things are controlled easier. GNU even allows the parallel engineering by different teams, and use the best and simpliest solutions.

From the security point of view we should reconsider even the glorious Von Neumann computer architecture, sharing the RAM space together for program and data. In its - at his time - quite  revolutionary concept can be now abandoned  thanks to still falling memory chip prices in order to provide better security.

Unless we shall deal with the security issues seriously, we should not be surprised if one day the net is called  the World Wide Worm instead.