90% of people who have bought an IPOD never thought of it in terms of "I want cool app X" when purchasing. Instead they thought "I want the gadget from that cool commercial that plays digital music files for me; bonus if it can also play movies too, although I don't even know how that works and probably won't use it anyway".
How do you know this? Can you read the minds of 90% of people? You realize there are installable apps. for iPods, right? http://www.apple.com/games/ipod/
First of all, when I say "virus" I mean "malicious code", which would include trojans, worms, scripts, etc intended to do something illicit and undesired by a user.
Say what you mean then. You can't blame me for the misunderstanding.
Considering no one has even been able to crack the OS loading procedure since at least the Gen 4 devices, I'd say they've done a pretty good job of securing the system.
Don't mistake lack of interest with lack of ability. What motivation would anyone have to crack it? Waste of time without return on money. The iPhone crack was motivated by money first and then fame.
A SDK would mean any person could write any code which allocates memory or uses wifi or the harddisk. You could simply WRITE a code INTENDED to buffer overflow, to zombie ping, to launch a mini-server, etc.
Not any person. Trusted people. Like I said. They have to be registered. Maybe pay Archos a registration fee.
So what if it's not "digitally signed"? Does that mean that people are only going to run Archos Digitally Signed apps???
Of course it does! The Archos OS should only run signed applications. It should reject any application that doesn't have a proper signature.
And even if the apps were somehow encrypted with Archos keys, on what basis would apps be judged to be fitting? What moral proof does a developer have to supply in order to be allowed to develop? Despite many hackers being bagged up for their actions, it still hasn't stopped people from developing viruses and getting caught, receiving double digit year prison sentences time and again.
Another red herring. Virii developers are usually anonymous. If you make everyone register, you know who they are! You can sue them immediately.
Moreover, code could unintentionally leave holes in an otherwise secure OS. Do Archos apps run under a typical user based schema or simply as a root user?
Non-root user, duh.
Also, you must consider the very real possibility that the TI chip inside is probably reaching the limits of what it can do already. The current SDK and OS probably allow very low level code to be written and optimized so both of these would have to be completely rethought and adapted for public use - a costly and extensive procedure that probably wouldnt get Archos any more customer satisfaction/support than say MKV support or Flash 8 or gapless mp3 playback. Finally, what's to stop a user from skipping the $70 codec fees in favor of free open source versions interfaced to work with the same existing API calls currently used by the apps?
We've been through this before. Non-compete clause.
Hmm let's see; because most linux apps are distributed as source code with makefiles that most linux users don't actually read or understand enough to validate whether or not they contain backdoors and other exploitable sections.
Wrong. Most are distributed as signed binaries. Get with the times. I haven't compiled a Linux app. in years except on rare occasions.
apt-get upgrade, etc.
Plus even if they build their own apps., they can't install it without a valid signature so they can't infect themselves.
And a Virus scan can still find known viruses it just like on a Windows. So it's not any more difficult to detect.
And many are compiled and installed via a root user (some, such as servers, require this to be done).
That's irrelevent. You're associating Linux desktop with Linux on Archos. Archos can always __not__ allow root installation.
Saying that "you have had 0 virii in 11 years" proves almost NOTHING about linux security as a whole, but rather just that YOU MAY NOT have had any virii in this time period (or had some virii/worms/trojans and didn't even know about it).
Ask me how many Windows viruses I've had. Proves a lot.
When was the last time any true RATIONAL system admin required an "academic paper" to prove the security vulnerabilities of an OS????
Are you kidding me? There are tons of papers on Linux security. The knowledge is out there. It's irrational _not_ to read it.
Do you know why the NSA helped develop SELinux? Specifically because so many linux systems have been compromised that exploits didn't need to be very complex (as Linux comes prepackaged with so many networking tools, servers, and functionality that make it very powerful but also very dangerous) and were starting to give cyber attackers massive power in taking down very sophisticated sites. Key loggers, packet sniffers, FTP servers, pingers, etc are all readily attainable and can allow a single compromised system to open up doors to entire networks.
Think about scope here. You think my PMP with 5 hours of battery uptime is gonna be a good zombie pinger? Get real. Hackers will hit desktops which are vastly more powerful and easier targets. It's just not economical on this scale. Also, what _sane_ admin is going to let people put their Archos on their Enterprise net? We're talking home users here with a PMP and maybe one or two computers. The computers are an even more likely target than the tiny PMP with barely any processing power.
Why is it harder to detect? Well there are few realtime anti-virus solutions for linux
I count no fewer than 10 on this page. http://virusall.com/downprodavmac.shtml
Now, I haven't verified them all and personally, I use AVG from Grisoft.
and the exploits developed for linux can sometimes be very obscure (even embedded within valid source code),
Trojans embedded in open source code are _easier_ to find than Trojans embedded in binary.
Need more "proof", well you're going to have to dig, but start with some of these keywords:
format string bug
Use lint. Can happen with or without an SDK.
Good luck finding this with a virus scanner. This DOS attack can be mitigated in the OS.
Port scanner is a diagnostic tool, not malicious code.
Umm. You can do that now remotely. Irrelevent to a SDK.
malicious tar archives
Irrelevent can't run without being signed.
The Ramen Worm
Irrelvent to an SDK. These are _remote_ exploits. Basically, you're saying you got nothing.
Better yet, pick up one of those Hacking Exposed books. They're a bit dated these days, but still contain very valid information. I remember one day when I enabled the telnetd and in less than 24 hours some clown from china had hacked my system and gained root using some simple POC code and scripts obtained from sites like rootshell.org.
Your fault for running telnetd. Use sshd.
I also know someone who failed out of highschool but was a systems genius and actually hacked the second largest electronic trading network in existence at the time (he basically broke in and left a message describing how vulnerable they were; a week later they hired him as security lead and he now runs his own IT security firm). Linux is not bulletproof and a SDK is really like providing a gun and hoping that those using it only require their bullets for target practice. I doubt a SDK will paint a target on the Archos for hackers, but it does open up the possibility, something any corporation would have to include in their risk assessment and planning.
Oh geez, now we're getting into magical friends. I give up.