Schneier on Security

Clive Robinson • December 1, 2004 3:42 AM

Just to echo Dan’s point, the underlying problem is the operating system. When Microsoft started on New Technology, the lead designer wanted to make “A better Unix than Unix”.

A nice objective to go for, however MS had too much invested in “MS-DOS” and “16 bit Windows” so the OS had to be able to support 99% of the legacy software without breaking it. Due to the complexity of this and the limited resources of the time NT started to develop an “all in one basket” approach which might originally have had clear interfaces within it but they quickly became blured and out of sight.

Another problem was the “single user” attitude, that the Personal Computer engendered that got carried forward into NT (remember the old joke about NT being C2 secure provided there was no floppy and network connected).

Untill recently MS had no incentive to sort out the mess NT had become, it was full steam ahead with bells and whistles. The result was that security was all but ignored so much so that in versions of NT it was not possible to know that a process was actually running on the machine if the process decided not to broadcast it was there.

Was this MS’s fault (partly) but you need to remeber that companies who put the effort into security usually went out of business as their products did not develope at the speed users expected them to. The message was clear to most organisations “security does not pay in the market place” so they activly avoided it.

The result we have an OS that still tries to support 20 year old applications that where designed for an insecure, resource limited, single user, single tasking, 16Bit OS…

Sorting it out is a bit like the old joke about the farmer being asked the way to somewhere, after some carefull thought he replies “If I was you I wouldn’t start from here”. Unfortunatly MS does not have this option the market place won’t let it start from scratch.

It has becom fairly clear that Business is happy with XP and 2000/2003 and has no intention of upgrading any time within the next five years. MS needs a very large pile of cash going through the door just to keep the lights on.

MS have a real problem (and so do we all) put simply,

1. It’s OS’s have gone as far as it’s main
   custommers want,
2. It gets a lot of bad publicity about the
   OS’s and their security,
3. The market says security does not pay,
4. There are several non MS OS’s available
   for nothing, some of which have a lot less
   in the way of security issues,
5. The functionality of non MS application
   software (some of it free) has the features
   most people want and used only to get from MS,
6. MS’s need for cash each year exceeds the GDP
   of a hundred or so countries.

This is a significant problem for MS and us as consumers in the market place. I am not sure where MS is going to go but I would expect to see,

A. More FUD
B. More litigation by proxie
C. More direct litigation
D. More preasure on law makers world wide
for new laws for things like patents,
copyright, reverse engineering etc.
E. Other “protective” measures such as the
“Trusted Computing” initiatives.

None of which are good for the market place or inovation.

Mad as it might be to some people we need to support MS to get security into their OS’s and we need to convince the market place that security does need to be there and payed for.

In many ways security is like the roads we drive on, if roads are either not there or of poor quality we go nowhere slowly and are horizons are very limited.

We generally pay for roads through the “commons” ie central and regional taxation, unfortunatly there is no mechanisum like this in place for software security.