Hacker News new | comments | show | ask | jobs | submit login

This is about the TPM in windows 8. It's the same argument about treacherous computing that goes around, except the article seems to be suggesting people think it's a extent problem now because the TPM is always on, not a hypothetical in the future/ Microsoft's long term plan. Further, there is a nebulous assertion linking this to the NSA.

1) The TPM still can't control your computer(yet). It can only measure it's state, allow you access to keys you created in some state, and attest to things about the state (which would allow other parties to mandate what state your system is in when interacting with it, but presumably those entities would be bond by German law and likely be German themselves)

2) If Microsoft wanted to backdoor your system, they don't need the TPM to do it. In fact, the TPM can be used to protect against a whole bunch of malware that various intelligence agencies might use: it can protect keys with passwords (with rate limiting/self destruct for guessing), make sure the system is in the same state(i.e. malware free) when you created you PGP key as it is when your using it to decrypt an e-mail, and it can isolate an application from the rest of your system.

I have a Windows 8 notebook and for the first time in my life I have no control over my own general purpose computer:

- I can't enter BIOS before entering OS.

- Once I enter the BIOS from the OS I can't activate the hard disk password.

- I can't install the Windows 8 OS clean. The MSFT has the deal with the computer producers that doesn't allow them to deliver the pure OS medium, you can only backup the already present installed files to some external HDD.

- Because of the previous and the fact that the binaries are controlled on the hardware level (TPM), I have no control of what's running on my computer -- I can't know, to be precise.

- It's even worse than that, there is some Intel built-in technology on the hardware/BIOS/drivers level which also has built-in "features" that allow communication with some external "command and control center" which I don't control. It supposedly allows, among other "features" disabling the notebook once it's stolen. But I don't control how it's done, and I don't know if it has additional backdoors. It proudly claims to facilitate "remote access."

It's scary how it looks like all together. I haven't even figured out how I'd be able to install Linux on the computer. In some forums people claim that the OEM should allow that, but apparently a lot of people haven't managed to actually install it on different specific computers -- there are BIOS problems that can't be avoided, and the OEMs don't give you support or the updates. Mine is an Intel i5 processor-based modern Acer. It's fast, but I have no control. Definitely not FUD.

Personally I like Apple approach more: thanks to their approach of the OSX or iOS (no third party pre-installed crap) at least I have to just trust Apple. Here I have to trust Microsoft, Intel and every company who has the drivers on my machine. Much more chance for some of them to do what they want, in the name of "cloud." Remember routers that are controlled from the producer of the router, even "protecting" you from browsing all the sites? Remember Android phones which upload all your passwords to the cloud of the mobile operator? That's where the "cloud" support of the driver writers goes now. It is scary.

(Globally, we're talking about this: http://xkcd.com/743/ -- We've been giving up the control of "infrastructures" for a long time)

What brand notebook is that? Just so I can make a mental note to avoid purchasing that brand in the future :)

It's Acer, but I as far as I know almost nothing it Acer specific -- it's a Windows 8, all OEMs must accept what MSFT wants, plus the concept of third party additions, plus the Intel technologies. I'm surprised how little coverage there is on this all aspects.

The discussions of kernel-level "giving up control" existed in Palladium and "technologies formerly known as Palladium (http://en.wikipedia.org/wiki/Next-Generation_Secure_Computin...) even 10 years ago but with Windows 8 they start to be ubiquitous and nobody even notices.

It was Apple that implemented Palladium in on iPads and iPhones and many technical folks even cheered it on.

I like iPad and iPhone as they are. Apple devices don't come with random crud from the third parties preinstalled. Windows computers have problematic things even in BIOSes: different software from companies that claim to "protect" your computer but can even provide remote access for third parties.

My roommate bought a Windows 8 equipped Asus a few months back and installed Mint on it. Seriously, Google 'How to install Linux on a Windows 8 laptop' sometime.

My point is still: new made-for-Windows 8 computers (especially notebooks) are bigger security risk than older Windows 7 computers. Installing Linux on the Windows 8 hardware can help if you avoid some issues, still even then there has to be independent evaluation of the modern BIOSes, appearing more and more in Windows 8 computers, that know how to connect or answer to the Internet and that provide the level of execution which user can't observe.

I had Windows 8 RTM installed on an old ThinkPad that I eventually wiped clean and put Ubuntu on. Granted its not a new laptop with Windows 8, but still, its not impossible to install Linux on a Win 8 laptop.

The notebook hardware made for Windows 8 behaves differently than all PC hardware before. It's irrelevant that the old ThinkPad works, it didn't magically change because the new software arrived.

No it doesn't. There were plenty of pre-Win 8 motherboards that had UEFI on there. Apple's been using it in their computers since the Intel switch in 2006. I don't know what you think is going on with your computer, but every other Windows 8 computer I've touched behaves identically to all other PCs, with the exception that Secure Boot is turned on by default.

I talk about the new Windows 8 notebooks not "motherboards."

Notebooks use motherboards, just like desktops do. And they run the same firmware. Apple's Macbooks have shipped with EFI since 2006.

Your second point is exactly what the German government is afraid of according to the article. According to the article there seem to be security vulnerabilities on 3 levels in TPM 2.0, which might be used by intelligence agencies - the article states NSA and China who is actually producing most of the TPM chips.


> If Microsoft wanted to backdoor your system ...

... they already have Windows Update. It cannot be null-routed (respective entries in /etc/hosts are simply ignored), it is virtually always on and it can be trivially used to deliver custom patches to specific boxes. What more can you ask for?

You are right that MSFT has the "update". The bigger problem with Windows 8 computers is that similar things are in practice available to all the "third parties" who have hardware or kernel-driver components. And you have less control than before about them all. It's a broader problem than just Microsoft.

The new "you as the user can't control the kernel stuff, even with the debugger" concept is really about the user (you) giving up the control. The excuse is that you as the user aren't supposed to be able to copy movies. In practice, you have no more control of your own computer whereas the companies have real-time control even of the content by directly controlling your computer. Some routers already did such stuff. It is really worrying -- having the part of the "great firewall of China/whateverothercountry" on your own computer which you paid with your own money.

What if you disable it? Doesn't it respect that setting?

Who says the original windows binaries don't have backdoors in them? So far as I can see the only difference TPM makes is that it potentially opens vulnerabilities in non-MS operating systems you run. If you're running any version Windows, or in fact any software you don't compile yourself from source(1), you just have to trust on faith it's not back-doored up to the eyeballs from the get-go. This has always been true.

(1) And in fact also have total confidence in the compiler itself: http://scienceblogs.com/goodmath/2007/04/15/strange-loops-de...

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact