For the motors that are exposed to salt water, APL applied commercially available protective coatings. The team tested the performance of the motors by submerging them in salt water. Two months later, they showed no sign of corrosion and continued to operate while submerged.
MAC address lockdown is practically useless on its own (see: ARP poisoning attacks & DHCP spoofing), not to mention completely unscalable.
Network isolation & good firewall rules at the demarcs are important, but they aren't a panacea.
slacka is correct, protecting networks after a physical intrusion has occurred is very difficult or impossible.
802.1X can help when it comes to the scalability of port authentication, and DHCP snooping and dynamic ARP inspection can both help in securing networks against ARP poisoning & DHCP manipulation, but they still leave some holes open.
802.1AE ("MACsec") theoretically mitigates a whole lot of attacks, but it's difficult to deploy to end-user devices. Want to attack a MACsec-protected network? Just look for the nearest printer which likely doesn't support MACsec and has an exception configged for its switch port.
This doesn't even begin to address hardware keyboard loggers, cameras, or TEMPEST attacks, all of which are perfectly capable of grabbing up user credentials.
In short:
1) Security requires a defense in depth approach, and physical security is an important part of that defense.
2) No matter how smart I may think I am, there's usually someone smarter who can think up some attack I haven't.
3) The more security you have, the more capex, maintenance and failures you have to deal with -- it's a always a balancing act.
More on topic though... as others have said, dumb article with no new revelations....
Agree with everything else you're saying, except perhaps this:
They also had to move the bot so it would fit in frame from that shooting location, that's why it's on a different bench from when the local news recorded it earlier.
The "beating" would have occurred on-camera (IIUC) had they performed it in the same place as it had been found. I'd venture that they weren't in possession at the time of the staged beating because it was a few nights after it actually occurred, so they just fake-beat a non-existent bot (plus the two prop arms—who knows what they actually were) just beyond the camera's visibility.
As I said before, I think it's more likely that they just want to troll the interwebs than actually release footage of themselves destroying property. Just a guess, though.
For example, why is the sky pitch black despite it purportedly being 5:45 AM (well into civil twilight, just 15 min. before full sunrise)?
The remaining question is: are they small d-bags (staged a fake destruction after some unknown party performed the actual destruction) or big ones (did the actual destruction themselves)? I'd guess the former—in an effort to make others look foolish, that seems to be more their M.O., but the latter wouldn't surprise me all that much either....
DSA and ECDSA do share a lot. To construct both of these algorithms, you start with an abelian group (a set of elements (e.g. integers; one of these becomes your public key) plus a "group operation" (e.g. multiplication)) and a "trapdoor operation" which is easy to calculate in one direction, but believed to be hard to calculate in reverse. The trapdoor operation is a repeated application of the group operation.
With DSA, the abelien group is a set of integers between 1 and p-1 (p is prime), the group operation is integer multiplication modulus p, and the trapdoor operation is integer exponentiation modulus p. (Note that exponentiation is repeated multiplication.)
With ECDSA, the abelien group is the set of points on an elliptic curve over a finite field, the group operation is something called "point addition", and the trapdoor operation is something called "scalar multiplication" (which is just repeated point additions).
The rest of the DSA and ECDSA algorithm is the same, and can be defined by steps such as "repeat the group operation x times" which is performed using one of the two group operations above depending on which algorithm is being used.
RSA on the other hand is a completely different beast, and not at all similar to ECDSA.
the only difference between RSA and elliptic curve is the equation you use for the curve.
ECDSA uses a curve. Neither RSA nor DSA uses any form of curves or points.
Elliptic curve obviously uses the equation for an ellipse.
Elliptic curve crypto uses the equation for, well... an elliptic curve. An ellipse (oval), despite the similar name, is an entirely different equation.
FWIW I use Trendnet Powerline adapters where I have poor Wi-Fi coverage. This is an Ethernet-over-AC-power technology. Specifically, I use a pair of their 500 series. I just ran a quick test, and with a 1Mbps upload stream (from my PS3 location to my PC), I was able to pull a download stream of around 60Mbps - 100Mbps (it fluctuated a bit between those speeds).
I occasionally have to reset them when they seem to lock up (say, once every 6 months or so), but overall I've been quite happy with them.
If you do find a solution to your question for the audio/video/controls, I't definitely recommend a pair of these adapters versus any wireless option.
(Of course, YMMV.... if the two power sockets happen to be on opposite legs in a typical 2-leg 120/240V North American home, then the signal needs to go all the way out to the nearest pole-mounted service transformer and back, which will likely result in a slower speed than otherwise.)
NearlyFreeSpeech.net offers many TLDs (not all) for registration. If you use them for DNS, their config page isn't that great IMO (it's a bit slow and cumbersome), but I like just about everything else about them.
Relevant to TFA: you can configure how many "recovery actions," between 2 and 7 (default: 3), which are required before you're granted access to lost account credentials. They also offer a "scorched earth" option: if you lose access to your account, it's gone forever (any associated services will persist until the account runs out of funds).
Whatever conflict of interest may or may not exist at the ESRB (or MPAA for that matter) pales in comparison to the real issue those two organizations continue to perpetuate in the US: showing a little skin, talking about sex, and swearing is a sure way to get you an M or R rating, but gunning people down or beheading them is relatively acceptable behavior.
(FYI I'm not talking about a violent game's ability to influence the behavior of individuals IRL, which I believe is insignificant if it exists at all, I'm only talking about the ESRB/MPAA's messed up moral compass.)
They don't even have any unlimited plans, they charge for usage!! That means that my p0rn site which gets tons of hits isn't subsidized by everyone else!
They actually charge for support! Why do I have to pay for support just cause I'm too stupid to figure anything out! I want my support to be paid for by everyone else who doesn't need support!!
They have all sorts of burdensome requirements to file a DMCA takedown request, I don't have time for this when all I really want to do is silence my critics!!
Please, whatever you do, avoid NFS at all costs!!!!
If the point of the IR lights is to overexpose a camera that's IR sensitive, wouldn't this light them up like a Christmas tree? I wouldn't think that the kind of person who wants to hide their identity would be interested in wearing a big neon sign that says "Look at meee!!!" to any security guards monitoring those cameras....
Keep in mind that this is crypto code, which often has different properties and requirements than other types of code.
For example, the count of bytes to store a private key, or a finite field element, or a hash output, etc. is hard coded and never changes. It would be akin to asserting that the sizeof a uint32_t is 4 somewhere in your code.... not very useful. Perhaps some defines would be nice from a documentation point of view, but that's more a style choice.
Input validation must be handled very carefully due to the possibility of opening up side channel attacks (anywhere you see a piece of code which branches depending on its input is a potential vulnerability for a timing attack, and is intentionally avoided by this library).
But the big reason I'm a lot less concerned than you are is simply the reputation of the authors, who are among the most talented cryptographers on the planet. It's not that they can do no harm (such hubris would be self-defeating), but rather that the likelihood of you or I finding some vulnerability with just a cursory glance seems extremely unlikely....
In addition to the various other oversights already mentioned, OP doesn't seem to understand Git (or perhaps SCMs in general) given that those (now revoked) keys are stillon GitHub -- there was no need for a bot to be all that quick.
Although I wouldn't blame OP for any single one of these oversights -- nobody's perfect -- it's fair to say that it took a number of different oversights / misunderstandings on OP's part for this to become a real problem.
1. A small handful of individuals who manage to work things out in an amicable way amongst themselves.
2. A project headed by a benevolent dictator for life (e.g. Slackware).
3. A governance model that is not dependent on a BDFL, and can scale better than "a small handful of individuals".
Even though option 2 works well for some projects, it's not always ideal. This doesn't seem to be a problem with a simple solution (and it probably doesn't help that not many techies are great at politics).
According to WaPo, she claims that this iPhone video was taken in September 2013, and not related to the alleged December 2012 incident. It looks like crooksandliars jumped the gun here.
The rather blatent Dancing with the Stars episode playing in the background may have even been intentional to provide additional credence to the video (the timing is dead on with her claim).
That's not to say she's not otherwise mistaken (or outright dishonest), but this isn't the smoking gun you're looking for.
if you're going to store it on someone else's system (iCloud, etc) then this is what happens.
Maybe, but this has been argued to death, no need for a repeat.
First, if you're going to have that crap on your phone
First, if you're going to have that crap on your tablet First, if you're going to have that crap on your laptop First, if you're going to have that crap on your home PC First, if you're going to have that crap in your car First, if you're going to have that crap in your home
Is that really what you're implying?!? Without more context, it's hard to tell if you're uninformed or just trollin.
I don't see a mistakenly created certificate. It looks like it is legitimately for hotmai.com ... Or another way - if Microsoft is catching typos, why would a nation state be amateurish for doing the same thing?
Microsoft isn't doing the same thing, though. You're right that the (real) hotmai.com site does redirect to outlook.com, however it doesn't have a certificate, nor does it even have https enabled.
Furthermore, the packet capture shows that whoever created it was trying to visit "login.live.com" (it's in the SNI field of the SSL Client Hello message), and so the server should have responded with a cert for that domain, not for hotmail.com nor hotmai.com.
I'll stick by my interpretation that this was amateurish, I just don't know if it was intentionally so.
Just an FYI... I've no reason to disbelieve the story, but it would be simple to fake the evidence presented...
I also wonder why the hotmail.com certificate was mistakenly created for the hotmai.com domain... that seems rather amateurish for a nation state. (Of course, perhaps plausible deniability is the reason.)
Regardless of whether or not it's fake, it does serve to point out the intentional flaws of Qihoo’s Chinese 360 "Secure Browser" pointed out by Rosyna above -- certainly a good thing to publicize.
A "zero-knowledge" service provider (allegedly) has no access to most of the digital assets stored by their service (e.g. LastPass, SpiderOak, etc.). They store encrypted blobs of data on your behalf, and send you these encrypted blobs at your request. Your PC (and not their servers) then decrypts this data using your password (of which the service provider has no knowledge).
I scanned through the bill, and it doesn't seem to acknowledge that such services exist. It doesn't even acknowledge that passwords themselves may not be retrievable, and instead groups passwords into the same category as other "digital assets."
Now IANAL, and it's entirely possible that some other bit of language in the bill or in a service provider's ToS could help to alleviate this, but if I ran such a service, I'd be a bit concerned....
When a provider needs to decide on it's next 100,000 "free" routers to provide to new customers, it shouldn't come to anyone's surprise when "cost-effectiveness" turns out to be its first priority. So I'm all for removing as much functionality as possible from any ISP-provider CPE; no wireless, just simple bridging.
But I really must respectfully disagree when it comes to separating out the wireless from the NAT box.
From a security point of view, having two manufacturers and two devices where one would suffice increases the attack surface -- it increases the likelihood that you have a security-related bug somewhere.
It increases the management burden -- now you have twice the number of devices whose firmware you have to keep up to date (if you're security conscience).
It doesn't scale well if you want more than one extra guest SSID or VLAN - sure you could attach a USB hub and half a dozen usb nics, or buy a VLAN-capable smart switch, but do you really want 3 Wi-Fi boxes, 3 unmanaged switches, and one router when just one Wi-Fi router would have worked fine?
There are definitely some advantages to separate wireless boxes. You can run guest SSIDs on different frequencies than your trusted SSID for example for better spectral efficiency. There are also cases where it's more convenient to have a NAT box near the CPE, and a separate Wi-Fi box centrally located. However in the average home setting, a single Wi-Fi/NAT box from a manufacturer with a decent track record is more practical.
I own an RB2011 at home too, and I've used both it and other RouterOS-based products professionally, and although they're not perfect, I can certainly recommend them for many cases. Here are a couple of random thoughts off the top of my head:
New major firmware versions (once every couple of years) are always buggy, avoid. That said, they're pretty good about releasing regular bug fixes, and they continue to support older routers for quite a while (the 500 series, released in 2006ish, is still supported on their latest firmware for example).
They can't seem to get a good OpenVPN implementation, which is a common complaint (but they have a lot of other styles of VPN which generally work well).
They use some open source software (e.g. it's Linux kernel based), but they only release the bare minimum required source code. This is definitely not an open tinker-and-recompile OS.
It does support virtualization, so you can run e.g. OpenWRT as a guest of RouterOS (yup, your router can have a router). You can also replace RouterOS with OpenWRT without worry of brickage. I haven't done either in a while, so I'm making no claims of either being easy or stable, but it can be done, and reverting back is easy.
It's really more business-oriented than consumer-oriented. That means its configuration is very flexible, but also rather complex unless you're used to configuring non-web-based routers.
Despite being complex, I find the configuration quite logical. It has no fewer than 4 different configuration interfaces (Web, CLI, Windows-based client app, and an API for automation). All present pretty much the same set of options in similar hierarchical arangements.
The documentation is much better than it once was, for most uses it's quite good.
The support community (via forum) is pretty good. Occasionally one of the Mikrotik staff will be a bit rude/condescending, but for the most part they're friendly (as are other posters).
Everyone who has a favorite adventure game is sure to point out that theirs isn't mentioned (for me, it was: Full Throttle (barely mentioned), Dreamfall, and more recently, Machinarium).
However, overall, this is simply a superb article. It touches on all the bases, is exceptionally well written, and really makes me yearn to play a new adventure game (come on Ragnar... you know you want to work on that sequel....)
I hope that anyone who has (or has ever had) even the most flighting interest in adventure games reads this article.
Slashdot Mirror
I've watched the videos
Yes, but did you read the article? ;-)
For the motors that are exposed to salt water, APL applied commercially available protective coatings. The team tested the performance of the motors by submerging them in salt water. Two months later, they showed no sign of corrosion and continued to operate while submerged.
MAC address lockdown is practically useless on its own (see: ARP poisoning attacks & DHCP spoofing), not to mention completely unscalable.
Network isolation & good firewall rules at the demarcs are important, but they aren't a panacea.
slacka is correct, protecting networks after a physical intrusion has occurred is very difficult or impossible.
802.1X can help when it comes to the scalability of port authentication, and DHCP snooping and dynamic ARP inspection can both help in securing networks against ARP poisoning & DHCP manipulation, but they still leave some holes open.
802.1AE ("MACsec") theoretically mitigates a whole lot of attacks, but it's difficult to deploy to end-user devices. Want to attack a MACsec-protected network? Just look for the nearest printer which likely doesn't support MACsec and has an exception configged for its switch port.
This doesn't even begin to address hardware keyboard loggers, cameras, or TEMPEST attacks, all of which are perfectly capable of grabbing up user credentials.
In short:
1) Security requires a defense in depth approach, and physical security is an important part of that defense.
2) No matter how smart I may think I am, there's usually someone smarter who can think up some attack I haven't.
3) The more security you have, the more capex, maintenance and failures you have to deal with -- it's a always a balancing act.
More on topic though... as others have said, dumb article with no new revelations....
Agree with everything else you're saying, except perhaps this:
They also had to move the bot so it would fit in frame from that shooting location, that's why it's on a different bench from when the local news recorded it earlier.
The "beating" would have occurred on-camera (IIUC) had they performed it in the same place as it had been found. I'd venture that they weren't in possession at the time of the staged beating because it was a few nights after it actually occurred, so they just fake-beat a non-existent bot (plus the two prop arms—who knows what they actually were) just beyond the camera's visibility.
As I said before, I think it's more likely that they just want to troll the interwebs than actually release footage of themselves destroying property. Just a guess, though.
Certainly looks staged, I agree.
For example, why is the sky pitch black despite it purportedly being 5:45 AM (well into civil twilight, just 15 min. before full sunrise)?
The remaining question is: are they small d-bags (staged a fake destruction after some unknown party performed the actual destruction) or big ones (did the actual destruction themselves)? I'd guess the former—in an effort to make others look foolish, that seems to be more their M.O., but the latter wouldn't surprise me all that much either....
Perhaps you're mistaking RSA with DSA.
DSA and ECDSA do share a lot. To construct both of these algorithms, you start with an abelian group (a set of elements (e.g. integers; one of these becomes your public key) plus a "group operation" (e.g. multiplication)) and a "trapdoor operation" which is easy to calculate in one direction, but believed to be hard to calculate in reverse. The trapdoor operation is a repeated application of the group operation.
With DSA, the abelien group is a set of integers between 1 and p-1 (p is prime), the group operation is integer multiplication modulus p, and the trapdoor operation is integer exponentiation modulus p. (Note that exponentiation is repeated multiplication.)
With ECDSA, the abelien group is the set of points on an elliptic curve over a finite field, the group operation is something called "point addition", and the trapdoor operation is something called "scalar multiplication" (which is just repeated point additions).
The rest of the DSA and ECDSA algorithm is the same, and can be defined by steps such as "repeat the group operation x times" which is performed using one of the two group operations above depending on which algorithm is being used.
RSA on the other hand is a completely different beast, and not at all similar to ECDSA.
the only difference between RSA and elliptic curve is the equation you use for the curve.
ECDSA uses a curve. Neither RSA nor DSA uses any form of curves or points.
Elliptic curve obviously uses the equation for an ellipse.
Elliptic curve crypto uses the equation for, well... an elliptic curve. An ellipse (oval), despite the similar name, is an entirely different equation.
FWIW I use Trendnet Powerline adapters where I have poor Wi-Fi coverage. This is an Ethernet-over-AC-power technology. Specifically, I use a pair of their 500 series. I just ran a quick test, and with a 1Mbps upload stream (from my PS3 location to my PC), I was able to pull a download stream of around 60Mbps - 100Mbps (it fluctuated a bit between those speeds).
I occasionally have to reset them when they seem to lock up (say, once every 6 months or so), but overall I've been quite happy with them.
If you do find a solution to your question for the audio/video/controls, I't definitely recommend a pair of these adapters versus any wireless option.
(Of course, YMMV.... if the two power sockets happen to be on opposite legs in a typical 2-leg 120/240V North American home, then the signal needs to go all the way out to the nearest pole-mounted service transformer and back, which will likely result in a slower speed than otherwise.)
NearlyFreeSpeech.net offers many TLDs (not all) for registration. If you use them for DNS, their config page isn't that great IMO (it's a bit slow and cumbersome), but I like just about everything else about them.
Relevant to TFA: you can configure how many "recovery actions," between 2 and 7 (default: 3), which are required before you're granted access to lost account credentials. They also offer a "scorched earth" option: if you lose access to your account, it's gone forever (any associated services will persist until the account runs out of funds).
Screenshot of NFS.net account recovery settings
Whatever conflict of interest may or may not exist at the ESRB (or MPAA for that matter) pales in comparison to the real issue those two organizations continue to perpetuate in the US: showing a little skin, talking about sex, and swearing is a sure way to get you an M or R rating, but gunning people down or beheading them is relatively acceptable behavior.
(FYI I'm not talking about a violent game's ability to influence the behavior of individuals IRL, which I believe is insignificant if it exists at all, I'm only talking about the ESRB/MPAA's messed up moral compass.)
These guys are crazy!
Please, whatever you do, avoid NFS at all costs!!!!
If the point of the IR lights is to overexpose a camera that's IR sensitive, wouldn't this light them up like a Christmas tree? I wouldn't think that the kind of person who wants to hide their identity would be interested in wearing a big neon sign that says "Look at meee!!!" to any security guards monitoring those cameras....
Keep in mind that this is crypto code, which often has different properties and requirements than other types of code.
For example, the count of bytes to store a private key, or a finite field element, or a hash output, etc. is hard coded and never changes. It would be akin to asserting that the sizeof a uint32_t is 4 somewhere in your code.... not very useful. Perhaps some defines would be nice from a documentation point of view, but that's more a style choice.
Input validation must be handled very carefully due to the possibility of opening up side channel attacks (anywhere you see a piece of code which branches depending on its input is a potential vulnerability for a timing attack, and is intentionally avoided by this library).
But the big reason I'm a lot less concerned than you are is simply the reputation of the authors, who are among the most talented cryptographers on the planet. It's not that they can do no harm (such hubris would be self-defeating), but rather that the likelihood of you or I finding some vulnerability with just a cursory glance seems extremely unlikely....
In addition to the various other oversights already mentioned, OP doesn't seem to understand Git (or perhaps SCMs in general) given that those (now revoked) keys are still on GitHub -- there was no need for a bot to be all that quick.
Although I wouldn't blame OP for any single one of these oversights -- nobody's perfect -- it's fair to say that it took a number of different oversights / misunderstandings on OP's part for this to become a real problem.
What are the options, then?
Even though option 2 works well for some projects, it's not always ideal. This doesn't seem to be a problem with a simple solution (and it probably doesn't help that not many techies are great at politics).
According to WaPo, she claims that this iPhone video was taken in September 2013, and not related to the alleged December 2012 incident. It looks like crooksandliars jumped the gun here.
The rather blatent Dancing with the Stars episode playing in the background may have even been intentional to provide additional credence to the video (the timing is dead on with her claim).
That's not to say she's not otherwise mistaken (or outright dishonest), but this isn't the smoking gun you're looking for.
if you're going to store it on someone else's system (iCloud, etc) then this is what happens.
Maybe, but this has been argued to death, no need for a repeat.
First, if you're going to have that crap on your phone
First, if you're going to have that crap on your tablet
First, if you're going to have that crap on your laptop
First, if you're going to have that crap on your home PC
First, if you're going to have that crap in your car
First, if you're going to have that crap in your home
Is that really what you're implying?!? Without more context, it's hard to tell if you're uninformed or just trollin.
I don't see a mistakenly created certificate. It looks like it is legitimately for hotmai.com
...
Or another way - if Microsoft is catching typos, why would a nation state be amateurish for doing the same thing?
Microsoft isn't doing the same thing, though. You're right that the (real) hotmai.com site does redirect to outlook.com, however it doesn't have a certificate, nor does it even have https enabled.
Furthermore, the packet capture shows that whoever created it was trying to visit "login.live.com" (it's in the SNI field of the SSL Client Hello message), and so the server should have responded with a cert for that domain, not for hotmail.com nor hotmai.com.
I'll stick by my interpretation that this was amateurish, I just don't know if it was intentionally so.
Just an FYI... I've no reason to disbelieve the story, but it would be simple to fake the evidence presented...
I also wonder why the hotmail.com certificate was mistakenly created for the hotmai.com domain... that seems rather amateurish for a nation state. (Of course, perhaps plausible deniability is the reason.)
Regardless of whether or not it's fake, it does serve to point out the intentional flaws of Qihoo’s Chinese 360 "Secure Browser" pointed out by Rosyna above -- certainly a good thing to publicize.
:-) Very nice pun there, thanks for that.
Oops, missed a "print" in there, but you get the point.
This also affects other scripting languages executed via CGI if the code spawns a shell, e.g.:
#!/bin/perl
`cat header.html`
It doesn't necessarily affect scripting languages executed via other means, e.g. mod_*
like a smack in the face to loyal Apple consumers.
I do believe that's called an iSmack.
A "zero-knowledge" service provider (allegedly) has no access to most of the digital assets stored by their service (e.g. LastPass, SpiderOak, etc.). They store encrypted blobs of data on your behalf, and send you these encrypted blobs at your request. Your PC (and not their servers) then decrypts this data using your password (of which the service provider has no knowledge).
I scanned through the bill, and it doesn't seem to acknowledge that such services exist. It doesn't even acknowledge that passwords themselves may not be retrievable, and instead groups passwords into the same category as other "digital assets."
Now IANAL, and it's entirely possible that some other bit of language in the bill or in a service provider's ToS could help to alleviate this, but if I ran such a service, I'd be a bit concerned....
When a provider needs to decide on it's next 100,000 "free" routers to provide to new customers, it shouldn't come to anyone's surprise when "cost-effectiveness" turns out to be its first priority. So I'm all for removing as much functionality as possible from any ISP-provider CPE; no wireless, just simple bridging.
But I really must respectfully disagree when it comes to separating out the wireless from the NAT box.
From a security point of view, having two manufacturers and two devices where one would suffice increases the attack surface -- it increases the likelihood that you have a security-related bug somewhere.
It increases the management burden -- now you have twice the number of devices whose firmware you have to keep up to date (if you're security conscience).
It doesn't scale well if you want more than one extra guest SSID or VLAN - sure you could attach a USB hub and half a dozen usb nics, or buy a VLAN-capable smart switch, but do you really want 3 Wi-Fi boxes, 3 unmanaged switches, and one router when just one Wi-Fi router would have worked fine?
There are definitely some advantages to separate wireless boxes. You can run guest SSIDs on different frequencies than your trusted SSID for example for better spectral efficiency. There are also cases where it's more convenient to have a NAT box near the CPE, and a separate Wi-Fi box centrally located. However in the average home setting, a single Wi-Fi/NAT box from a manufacturer with a decent track record is more practical.
I own an RB2011 at home too, and I've used both it and other RouterOS-based products professionally, and although they're not perfect, I can certainly recommend them for many cases. Here are a couple of random thoughts off the top of my head:
Everyone who has a favorite adventure game is sure to point out that theirs isn't mentioned (for me, it was: Full Throttle (barely mentioned), Dreamfall, and more recently, Machinarium).
However, overall, this is simply a superb article. It touches on all the bases, is exceptionally well written, and really makes me yearn to play a new adventure game (come on Ragnar... you know you want to work on that sequel....)
I hope that anyone who has (or has ever had) even the most flighting interest in adventure games reads this article.