Re: [Full-disclosure] windows future

["lsi" <stuart@xxxxxxxxxxxxxx>]

Hi All,

Sorry for the delay, I had some urgent migration planning to attend 
to ... ;)  Stats below.  Short version: evacuate.  Long version:

- stats are in, exponential curve is real, see it for yourself here:

http://eval.symantec.com/mktginfo/enterprise/white_papers/b-whitepaper_internet_security_threat_report_xiv_04-2009.en-us.pdf

(page 10)

- I also added up the numbers at 
http://www.virusbtn.com/resources/malwareDirectory/prevalence/index.xml?year=2009
... exponential curve also visible, though I think their stats are 
dodgy, their website is already suffering from math limits - it is 
reporting current yearly stats as NaN% (Not A Number).

- average rate of change per year (annual growth rate), calculated 
from Symantec's chart: 243%

- approximate date when number of NEW threats reached 1 Million: 2008

- approximate date when number of NEW threats will reach 1 Billion: 
2015

- approximate date when number of NEW threats will reach 2 Billion: 
2016

- charts showing this: 
http://www.cyberdelix.net/files/malware_mutation_projection.pdf

- will the AV companies be able to classify 1 billion new threats per 
year? that is 2.739 MILLION new threats per DAY (over 1900 new 
threats per minute).

- will your computer cope with scanning every EXE, DLL, PIF etc 1 
billion times, every time you use them?

- aside from the theoretical limits imposed by hardware and software, 
there is one extra limit, imposed by users.  Users will not tolerate 
machines operating slowly, and will seek alternative platforms well 
before 100% CPU utilisation (either as a direct result of the size of 
the blacklist, or indirectly caused by swapping due to low RAM).  
This user limit might be lower than 20% CPU utilisation.  If users 
figure out that 20% of their time is being wasted, and rising fast, 
they will run for the exit.

- will you tolerate your machine constantly processing a list a 
billion items long?

- do you plan to, and can you afford to, upgrade your compute power 
by 243%, every year?

- will you do this, even though you know viable alternative platforms 
exist, at less total cost to yourself?

- if you're already irritated that AV is slowing down your machine, 
consider that malware levels will be 500 times higher in approx 5 
years (assuming growth rates continue at 243%). That means your AV 
will be running 500 times slower.  Unless you upgrade your machine by 
500 x current (eg. to an effective speed of approx 1000 GHz), your 
machine is going to slow down even more.  Given that chipmakers don't 
seem to be able to get much past 5GHz, without melting the die, that 
means you'll need 200 of today's processors, just for malware 
filtering, by 2015.

- Moore's Law says compute power doubles (200%) every 24 months.  
However, malware is growing at 243% every 12 months.  Thus it is 
already exceeding Moore's Law, by a massive margin.  I suspect this 
means this race is unwinnable, and we should give up now, and devote 
our resources to something sustainable.

- how AV writers will generate 2.7 million new threats/day: 

"Evolvable Malware":
http://www.genetic-programming.org/hc2009/3-Noreen/Noreen-Presentation.ppt

"A Field Guide to Genetic Programming":
http://www.gp-field-guide.org.uk/

Wiki:
http://en.wikipedia.org/wiki/Genetic_programming

- the insecurity of Windows creates a public space, of sorts, an area 
of common ground, with shared ownership - and this is thus 
susceptible to the tragedy of the commons ... 
http://en.wikipedia.org/wiki/Tragedy_of_the_commons ... so no, I 
don't think malware authors will slow down the mutation rate, so as 
to prolong the life of the platform, they do not work together.  As 
Messagelabs puts it, "there's no honour amongst thieves" ... 
http://www.messagelabs.com/mlireport/MLIReport_Annual_2008_FINAL.pdf

- the greenhouse emissions caused by billions of computers checking 
billions of items for billions of malware are likely to be 
measurable, and will increasingly erode the world's ability to meet 
environmental targets

- my own maths might be dodgy, please check it, spreadsheet: 
http://www.cyberdelix.net/files/malware_mutation_projection.ods

Stu

On 28 Aug 2009 at 15:32, lsi wrote:

From:                   "lsi" <stuart@xxxxxxxxxxxxxx>
To:                     full-disclosure@xxxxxxxxxxxxxxxxx
Date sent:              Fri, 28 Aug 2009 15:32:45 +0100     

> Thanks for the comments, indeed, the exponential issue arises due to 
> use the of blacklisting by current AV technologies, and a switch to 
> whitelisting could theoretically mitigate that, however, I'm not sure 
> that would work in practice, there are so many little bits of code 
> that execute, right down to tiny javascripts that check you've filled 
> in an online form correctly, and the user might be bombarded with 
> prompts.  Falling back on tweaks to user privileges and UAC prompts 
> is hardly fixing the problem.  The core problem is the platform is 
> inherently insecure, due to its development, licensing and marketing 
> models, and nothing is going to fix that.  Even if fixing it became 
> somehow possible, the same effort could be spent improving a 
> competing system, rather than fixing a broken one.
> 
> Just to complete the extrapolation, the below.
> 
> Assuming that mutation rates continue to increase exponentially, 
> infection rates will reach a maximum when the average computer 
> reaches 100% utilisation due to malware filtering.  Infection rates 
> will then decline as vulnerable hosts "die off" due to their 
> inability to filter.  These hosts will either be replaced with new, 
> more powerful Windows machines (before these themselves surcumb to 
> the exponential curve), OR, they will be re-deployed, running a 
> different, non-Windows platform.
> 
> Eventually, the majority of computer owners will get the idea that 
> they don't need to buy ever-more powerful gear, just to do the same 
> job they did yesterday (there may come a time when the fastest 
> machine available is unable to cope, there is every possibility that 
> mutation rates will exceed Moore's Law).  The number of vulnerable 
> hosts will then fall sharply, as the platform is abandoned en-masse.
> 
> At this time, crackers who have been depending upon a certain amount 
> of cracks per week for income, will find themselves short.  They will 
> then, if they have not already, refocus their activities on more 
> profitable revenue streams.
> 
> If every computer is running a diverse ecosystem, crackers will have 
> no choice but to resort to small-scale, targetted attacks, and the 
> days of mass-market malware will be over, just as the days of the 
> mass-market platform it depends on, will also be over.
> 
> And then, crackers will need to be very good crackers, to generate 
> enough income from their small-scale attacks.  If they aren't very 
> good, they might find it easier and more profitable to get a 9-to-5 
> job.  The number of malware authors will then fall sharply.
> 
> The world will awaken from the 20+ year nightmare that was Windows, 
> made possible only by manipulative market practices, driven by greed, 
> and discover the only reason it was wracked with malware, was because 
> it had all its eggs in one basket.
> 
> Certainly, vulnerabilities will persist, and skilled cracking groups 
> may well find new niches from which to operate.  But diversifying the 
> ecosystem raises the barrier to entry, to a level most garden-variety 
> crackers will find unprofitable, and that will be all that is 
> required, to encourage most of them to do something else with their 
> lives, and significantly reduce the incidence of cybercrime.
> 
> (now I phrase it like that, it might be said, that by buying 
> Microsoft, you are indirectly channelling money to organised crime 
> gangs, who most likely engage in other kinds of criminal activity, in 
> addition to cracking, such as identity theft, money laundering, and 
> smuggling. That is, when you buy Microsoft, you are propping up the 
> monoculture, and that monoculture feeds criminals, by way of its 
> inherent flaws.  Therefore, if you would like to reduce criminal 
> activity, don't buy Microsoft.)
> 
> -EOF
> 
> On 27 Aug 2009 at 13:45, lsi wrote:
> 
> From:                 "lsi" <stuart@xxxxxxxxxxxxxx>
> To:                   full-disclosure@xxxxxxxxxxxxxxxxx
> Date sent:            Thu, 27 Aug 2009 13:45:01 +0100
> Priority:             normal                                               
>             
> Subject:              [Full-disclosure] windows future
> Send reply to:        stuart@xxxxxxxxxxxxxx
>       <full-disclosure.lists.grok.org.uk>                                  
>       
>       <mailto:full-disclosure-
> request@xxxxxxxxxxxxxxxxx?subject=unsubscribe>     
>       <mailto:full-disclosure-request@xxxxxxxxxxxxxxxxx?subject=subscribe> 
>       
> 
> > [Some more extrapolations, this time taken from the fact that malware 
> > mutation rates are increasing exponentially. - Stu]
> > 
> > (actually, this wasn't written for an FD audience, please excuse the 
> > bit where it urges you to consider your migration strategy, I know 
> > you're all ultra-l33t and don't have a single M$ box on your LAN)
> > 
> > http://www.theregister.co.uk/2009/08/13/malware_arms_race/
> > 
> > If this trend continues, there will come a time when the amount of 
> > malware is so large, that anti-malware filters will need more power 
> > than the systems they are protecting are able to provide.
> > 
> > At this time, those systems will become essentially worthless, and 
> > unusable.
> > 
> > You can choose to leave now, or later.  But you cannot choose to 
> > stay...
> > 
> > (I mean, that the Windows platform seems destined to fill, 
> > completely, with malware, such that your computer will spend ALL its 
> > time on security matters, and will have no CPU, RAM etc left for 
> > actual work.  At the end of the day, the ability of malware to infect 
> > Windows machines is due to the fact that Windows is a monoculture, a 
> > monolith, built by a single company, with many interconnections and 
> > hidden alleyways.  It's hard to imagine a platform LESS vulnerable - 
> > compare with open-source efforts, which are diverse, homogenous and 
> > connect via open protocols.  Malware finds life hard in the sterile, 
> > purified world of RFCs, where one of many different programs may 
> > process your malicious payload, all of which have been peer-reviewed. 
> >  In Windows, malware knows that a specific Microsoft EXE will process 
> > its data, knows that the code has not been thoroughly checked, and 
> > can make use of undocumented mechanisms.
> > 
> > So basically Microsoft, by hoarding their source, by tightly 
> > integrating functionality, and by seeking to monopolise the various 
> > markets created by the platform (browser, media player, office 
> > software), have doomed Windows, and everything that runs on it.  The 
> > lack of diversity in the Windows ecosystem means that it is highly 
> > vulnerable to attack by predators.  The fact that malware mutation 
> > rates are accelerating is a clear indicator that the foxes are 
> > circling.  This is the beginning of a death spiral; the malware 
> > numbers we've seen in the past 20 years were the low end of an 
> > exponential curve, and we're now getting to the steep part.
> > 
> > The problem is that any given computer is only capable of so much 
> > processing.  It has an upper limit to the amount of malware it can 
> > filter, those limits being related to CPU speed, RAM, diskspace, 
> > network bandwidth.  This upper limit looks like a horizontal line, on 
> > the chart that shows the exponential curve mentioned above.
> > 
> > So my point, is that eventually, the exponential curve is going to 
> > cross that horizontal line, for any given computer, and when that 
> > happens, that computer will no longer be able to filter malware.  It 
> > will only be able to filter a subset, and thus be vulnerable to the 
> > rest. Consequently it will not be usable, for instance, on the web, 
> > and will essentially become a doorstop...
> > 
> > The only escape from this inevitability is to ditch the platform that 
> > is permitting the malware - that is, the only escape is to ditch 
> > Windows. It is being eaten alive, by predators that only have a 
> > foothold because there are weaknesses in the platform.
> > 
> > Given that it can take years to migrate to a new operating system, I 
> > do recommend, if you have not already done so, that you commence 
> > planning to ditch Windows.  I might be wrong about the exponential 
> > curve, but if I'm not, then there may not be a lot of time in between 
> > when malware levels seem managable, and the time when they are not.  
> > If your business depends on Windows machines and they all become 
> > unusable, you will have no business.  What you definitely must NOT 
> > do, is assume that Windows is going to be around for a long time.  It 
> > is a dead man walking.
> > 
> > - Of course, there might be a few years yet.  You can spend those 
> > years running up your IT bill, with lots of new computers that are 
> > required to filter all that malware while still performing at a 
> > useful speed.  Or, you can ditch Windows, and keep your existing 
> > hardware - it runs perfectly well, when it's not weighed down 
> > defending the indefensible.
> > 
> > [If Microsoft dooming Windows isn't ironic enough, consider that 
> > every time malware authors pump out another set of mutations, they 
> > are nailing one more nail in the coffin of the platform that they 
> > depend on to make their living! Ahh, there is justice in the world 
> > after all.]
> > 
> > [And the end game?  Well, M$ could open-source Windows, but frankly, 
> > why would anyone bother trying to fix it?  As the old saying goes, 
> > don't flog a dead horse...]
> > 
> > ---


---
Stuart Udall
stuart at@xxxxxxxxxxxxxx net - http://www.cyberdelix.net/

--- 
 * Origin: lsi: revolution through evolution (192:168/0.2)

_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.grok.org.uk/full-disclosure-charter.html
Hosted and sponsored by Secunia - http://secunia.com/