@Gilles The nipples in Friends was pretty much just Jennifer Aniston straining her shirts. Game of Thrones is a little...less strain.

@AviD No, they seem to be showing a little more variety.

It basically is a way of storing password hashes on a server (as really BIG files) so that the cost of an offline attack is essentially the same as the cost of an online attack, given reasonable assumptions.

@nealmcb I may be misreading the paper (I just woke up...) but I don't think the 2 large assumptions they make (storage is cheap, bandwidth is expensive) holds true at all.

While storage is relatively cheap, it is still a significant cost at scale. Also, exfiltrating large amounts of data isn't exactly difficult given the state of most detection systems today. It may take longer, sure, but not impossible or even difficult.

@TerryChia I need to see a practical example to decide how useful the theory is - thus the question. But it does seem reasonable that there there is some bound at which you might expect to detect a theft before it gets too significant, etc.

It also seems like you might be able to put the big password files behind a thin pipe or something. Again I want to see an implementation.

@nealmcb I guess the problem is that the level of security doesn't hold up against the worst case scenario where the attacker has indeed manage to retrieve the entire dataset. With the current KDF method you can be reasonably certain that a bruteforce attack is impossible given a high enough cost parameter and a decent password.

Also, if it's patented like you said you won't be seeing an implementation anytime soon.

But we know that users won't choose decent passwords.....

And the purpose of a patent is to let the inventor make money off an implementation. Well, usually.... So does Telcordia have an implementation?

The idea of making it difficult to exfil the password db is interesting, though not perfectly novel. An HSM is the embodiement of that concept.

@tylerl Right. But they have cool hashing algorithms to make it apply to user passwords.

Look at the appendix, a maximum of 4096 users require 1TB of storage. How the hell is that practical?

I assume you can twiddle the parameters to better fit your scenarios.

And perhaps that might be a good parameter set for an sso server for a small company.

@nealmcb I don't think their technique is practical enough to catch on. But I think you could reasonably easily implement a Password-Hashing-As-A-Service device that accomplished the goal more cheaply.

@tylerl Who do you trust to run that service?

@nealmcb I used "as a service" a bit tongue-in-cheek. I mean basically a hardware device -- even a simple computer -- that allows you to interface with it but won't dump the database.

@TerryChia And as they say, "If we limit the server to an out- going bandwidth of 8192 bits/sec = 1024 bytes/sec, it will take the adversary over 30 years to download that much data." Yet the server can be used to do approximately 32 logins/sec

it'd probably be enough to have the device NOT on the network, but rather communicate to the rest of the world through serial, or soemthing.

@tylerl That seems to be what they're suggesting, though I haven't read the whole paper.

But I'd rather just implement it myself with some off-the-shelf hardware. Until they start suing me....

I wrote them - we'll see if they have anything....

And I get the sense that an adversary would need to download nearly the entire 1 TB to get any significant number of passwords. Though I don't really understand the "β" parameter they reference only once.

I think the idea of making the hashes LARGER to make downloading them harder isn't the right way to go. Your average attacker has effectively unlimited bandwidth because he's not using his own.

The idea that it'll show up on the IDS if you make it bigger but not if you dont is, I think, wishful. And presupposes that they're using an IDS.

@tylerl Unless the attacker can physically get to the box, he has to use the defender's limited bandwidth.

and the defender puts the password box on the end of a slow line

You need to deploy an ids on the link to the password box or you're missing the point

@nealmcb I think when you run the number in a practical sense, this technique has way worse returns than your average alternative

but it isn't hard to design an ids to notice that the user has spent 20 years saturating your little internal connection trying to get the database, and you'd better act within a few more years if you want to stop the compromise of your passwords.

But I could build you a device in about 2 hours out of $35 of hardware that'll be 100% impervious to password dumping if installed correctly.

@tylerl What is the average alternative?

"if installed correctly"? I.e. no bugs? Good IDS?

@nealmcb Mine would be a raspberry pi with a usb-serial connection to the "host" computer which accepts username/password submissions and pushes back "OK" or "Not OK" based on local storage, and throttling based on simple sleeps. You could build the whole thing in less than a lunch break.

The "host" computer could expose it as a web service or whatever. But the host computer doesn't actually have the passwords, so even if it gets hacked the PWs couldn't be dumped.

And you can't hack the pi over the serial console because the system is too damn simple.

And if / when I find a bug in your box, I can quickly get the database.

HOW? Really.

Bugs are everywhere.

I'd prefer a system where the only dangerous bug is one that expands the bandwidth - much easier to harden that small surface than a whole application and networking stack

But I wonder also if your idea has been implemented already.

What is this? On-topic conversation? I knew I sensed a disturbance in the force.

@DavidFreitag :)

@nealmcb gist.github.com/tylerl/369a9c13368f80eb88c0

Bugs are everywhere, but only a narrow set of bugs are exploitable.

I can see a few improvements -- you can b64 the username/password to allow for \n in password.

But by and large, nothing you can do to this code will allow for arbitrary code execution

Quick work! Note missing text at the end of the pw_test = line, and that variable doesn't seem to be used.

yeah... changed my idea mid-stream

If I did this in a real-world setting, I'd do quite a bit more error handling, restarting of the serial connection, a better sleep mechanic, allow for password updates, etc.

Oooh - more chances for bugs! ;)

...most obviously when you handle password updates....

But the idea is to keep the wire protocol dead-simple and don't mix data and metadata. I'd probably use some simple encoding rules with some basic checksumming on messages.

But @DavidFreitag should be even more impressed that an implementation has been completed in this room, perhaps faster than it would take to even do the search I asked for.

I agree with DJB that our current mentality that security vulnerabilities are inevitable is just lazyness wrapped in platitudes.

It is possible to write secure code.

Give me an hour and I can hash out a protocol and implementation that provides the necessary functionality without providing any wiggle-room for exploitation. The problem is so simple that this is easily possible. We're not writing a web browser here.

@tylerl Do you have a link for that? How broadly applicable is this notion? Of course lots of buggy software will always be written. How many can write how much bug-free code?

Ok, so I missed a bunch of this when I was gaming.

@nealmcb Erm, I don't know if I can point to any one statement; it's really his broad philosophy. It's why he wrote Qmail and djbdns. People back in the day looked at bind and sendmail and said "vulnerabilities are just unavoidable," and he said "the hell they are." And went ahead and wrote these competing programs to prove it.

@nealmcb Here's one bit of commentary he wrote WRT qmail though: cr.yp.to/qmail/guarantee.html

@nealmcb Again, the idea is nice in theory but I don't see it being practical especially in the environment today where everything is cloud this cloud that. This idea will never be widely deployed.

He's also a bit (in)famous for requiring his students to go out and find new undiscovered security vulnerabilities in production software as part of their homework. I don't know if he still does that.

And if you have the resources to roll out something custom and keep it secure, like @tylerl said there are much better alternatives.

@tylerl Really? That's cool. Any particular class of production software?

I imagine it won't be that difficult if you can pick and choose from all the software in the world.

@TerryChia I don't understand why it seems impractical to you, at least for the one use case and worked example we discussed. Except for the patent issue which is a huge red flag for me, though it could just be defensive.

We just need a standard implementation. By DJB....

...or @tylerl

@nealmcb Again, the terabytes or even gigabytes of storage required to store the credentials for a handful of people is just too expensive.

@TerryChia it.slashdot.org/story/04/12/15/2113202/…

@TerryChia Spending a few hundred dollars on securing an enterprise sso to get the kinds of security guarantees they offer is a nearly negligible cost. What enterprise doesn't have that kind of budget for their crown jewels? Now how much added security you get is a different issue. But the expense is not the problem for that use case.

@tylerl Ah impressive!

Note 4000 != handful....

The logistics and expense of off-site backup is another issue (for both @tylerl's gist and the original scheme in question)

Anyway - fun to hang out here again - good discussions, folks!

@nealmcb There is of course the cost of the server, actually writing the software because no one else is using it and that there's no implementation available, paying a security auditor to make sure that the software itself has no bugs and getting a developer to maintain it. Oh, the alternative grab an implementation of PBKDF2, Bcrypt or Scrypt available in literally every single language there is? Why not just use that instead.

But time to get to other stuff....

@TerryChia Well assuming the idea is useful, getting a good implementation is mostly a question of the patent approach they're taking, as I already said. But they offer something that the other stuff you all are talking about doesn't really address....

@nealmcb It's only useful to a very very very limited subset of places where passwords gets used. At that point, hell just get or build a HSM...

Which is essentially the thing @tylerl was describing earlier.

Hey - any other explainxkcd.com contributors out there? The ISEE3 reboot stuff has been amazing....

Backing up an HSM is not easy....

hasta la.....

Jesus @Neal is here! Must have been a year since I last saw you in DMZ. And you mention ISEE-3? I think I need a drink. It's been amazing, still is actually. And SDR is now in the loop, that's huge too. I'm still trying to figure out what all kinds of good things that means for future of space exp in general.

You guys watched the interview with Snowden? theguardian.com/world/video/2014/jul/17/…

BTW, cc @Rоry's and the rest of Highlands: Scotland could be base for spaceport, says UK government

6/8 locations mentioned are in Scotland

This must be the least coherent scientific paper abstract I've ever read: link.springer.com/article/10.3103%2FS1068335611050034 Warning, you might develop self-inflicting lobotomy tendencies!

Can't they just use MASERs like the rest of the sane world?

I

AM

BOOOOOOOOOOOOORRRRRRRRRRRRRRRRREEEEEEEEEDDDDDDDDDDDDDDDD

also: meh, morning

also: what

@nealmcb Ahh - that was the other place I saw your username :-)

@TildalWave Yes - now that will be fun...if it ever happens

@kalina s/bored/boring :-)

oh

world war 3 is starting

who thought it was clever to shoot down a passenger plane?

fuck it, I'm moving to Antarctica, I'll be back in a few hundred years once nuclear winter is over

@kalina I thought you didn't like the cold? Gets down to a healthy -50 degrees down that way :-)

@kalina Did you see the rebel group that did it tweeted they had taken out another Antonov! Then rapidly removed the tweet once they realised...

@kalina you know just because everywhere else will be colder doesn't make Antarctica any warmer ;)

@RoryAlsop It'll happen, it has to. Because I just added to my bucket list to work in space industry in Scotland

@TildalWave Yay!

and deep space deep-frying pigs in space to crisp bacon of course

damn

@RoryAlsop would rather be cold than dead

at least in the short term

@RoryAlsop If this is true why are none of the news sites picking it up?

@kalina I saw it on two news sites last night. Hang on - will see if I can find a link

here it is kyivpost.com/content/ukraine/…

I've found a link

I know how to google

don't insult me

was in my history and I thought to lend you a hand, I was not trying to be patronising

actually, that's a wrong link, it's this one I wanted to post dailymail.co.uk/news/article-2696389/…

anyway, have to run ... t/c

@TildalWave ... Daily Mail. Really?

@FEichinger sometimes you have to dance with the devil

@RoryAlsop I would suggest it's a carrot to encourage a no vote....

@RоryMcCune probably :-)

Everything's political </SkunkAnansie>

@RoryAlsop well at the moment in Scotland I can't help seeing any announcement as part of the indyref..

although that said you'd think the tories were dying to lose given some of their euroskeptic rhetoric at the moment..

"Indyref" - almost makes it sound cool :-)

@RоryMcCune I know. Both sides have some utter morons ...

always the way

The solution discussed with siblings last night: ban politicians, or at least allow hunting of them on Tuesdays :-)

@RoryAlsop heh

@RoryAlsop or ban anyone who shows the desire to become a politician from becoming one

@RоryMcCune yup

see the Web Crypto API is getting turned on by default in Chrome 37 w3.org/TR/WebCryptoAPI ..

@RоryMcCune JS crypto. :(

Crypto in a language without integers.

@TerryChia I was thinking it might introduce some useful possibilities, for example in detecting MITM attacks...

Hmm wait, Web Crypto doesn't have to be implemented in JS right?

Since it ships with the browser.

It just has to expose the API to JS?

yeah I think that's right

That should be fairly useful then.

looking at their use cases at lot of it seems to revolve around reducing reliance on SSL/TLS

@TerryChia BTW think I've found the right tool for my deployment tasks github.com/sprinkle-tool/sprinkle

lightweight and does pretty much what I want..

@RоryMcCune Interesting. It sounds a lot like Ansible.

@TerryChia a bit so yep but there's no real concept of it being scaled up. ansible/puppet/chef all struck me as being designed for quite large scale

with the baggage that goes with that

for this I just need to get my script working and install the ruby gem

@RоryMcCune Nah, Ansible certainly scales decently but it's very simple to use as well.

It's definitely not as heavy as Chef or Puppet.

I like that it uses YAML instead of a custom DSL.

@TerryChia dunno when I went to ansible.com and they're starting off with 20 minute videos and talking about "free trials" it felt a bit OTT for my modest needs :)

@RоryMcCune Heh. That's the supported version. The FOSS one is available and easy to use. :)

@TerryChia sure but it always makes me nervous to see that approach, how long before things start popping up in the paid version only...

which is cool if I need it for a big purpose but for me this is a "nice to have" automation task...

must say though I'm liking being able to use vagrant for testing this

makes creating/destroying the test VM nice and easy

@RоryMcCune It looks like the paid version of Ansible is the shiny web interface.

As well as support of course.

@RоryMcCune Oh yeah, Vagrant is definitely nice.

And then there's always Docker.

@TerryChia for some stuff sure dockers likely better but the networking support seems a bit hinky at the moment..

whereas vagrant is just vmware networking..

@RоryMcCune Vagrant is really for dev stuff.

You can combine Docker + something like ansible or chef or whatever for production stuff. Different target markets imo.

@TerryChia yarr different solutions work for different scenarios, I'm sure for people deploying n-tier clouds docker+chef/puppet/ansible would make sense...

@RоryMcCune And OpenStack of course. :)

@TerryChia and all the other cloudy style initiatives (there seem to be several...)

@RоryMcCune OpenStack is the one with the biggest names behind it though.

Red Hat, Rackspace, IBM I think.

@TerryChia indeed, I tried out an install of devstack in a VM which worked nicely but scary as smeg to watch the install process and it's just pulling loads of semi-random software down from a wide variety of places

It's probably one of the largest Python projects now.

I mean an admin of that system has no chance of understanding what they're running and being able to patch it manually

they'll be 100% dependent on the openstack people to keep on top of patching...

@RоryMcCune Ehh, the project really isn't meant for any small or even medium sized company.

If your business needs require an openstack installation you most likely already have a team of competent sysadmins.

@TerryChia d'ya reckon that'll stop them deploying it, so they can say they "have a cloud"?

@RоryMcCune That's really their own problem isn't it? ;)

Interestingly enough the cryptography project I'm contributing to was written for OpenStack's use.

@TerryChia harsh

@RоryMcCune Hey, they can always buy support from Red Hat. ;)

If they have money to blow by all means. :P

@kalina Is that you?

Nope, that's German comedy. Martina Hill.

Also, morning. sigh

@FEichinger nah - it's afternoon now :-)

on British Summer Time, anyway

@kalina Don't you ever again tell me this weather isn't "hot". My exam was delayed by an hour and moved to different rooms because of this damned heat.

@Rory >.>

@FEichinger it's even hot in Scotland this week

@FEichinger 20s isn't hot

@kalina is so

:-)

lightweights

@kalina I have evolved to my environment

people walking around with no shirts on, I'm still fully dressed

@kalina at least it's dress down Friday, so I have my bicycles.SE t-shirt on, which is much better than a formal shirt

Dress down Friday?

I'm still in my pyjamas

I'm debating stopping watching stargate atlantis and actually getting out of bed

@FEichinger oh really, you no doubt have a crush on her?

she seems like the crazy sort of girl you'd go for

@kalina As evidenced by his obsession with you.

@TerryChia how rude

I am not crazy, you guys are just too boring

@kalina What would you like to do today? Mock @FEichinger about PHP or have another OS debate?

or publish a song ...

@RоryMcCune dotdotdot

@TerryChia exactly

It's Friday - why not write two songs

and then get out of bed

althought looks like @kalina might get beaten to it... translate.google.com/#ja/en/…

@kalina - do you do songs with lyrics, or are they musical/instrumental?

hey @RoryAlsop see sharon has a book out.. amazon.co.uk/…

@kalina I do not have a crush on her. Her work in satire is pretty good, though.

@RоryMcCune £18 on kindle! It better be good...

@RoryAlsop yeah that's not cheap but I guess with more niche books they're not expecting to sell big numbers of copies... Perhaps we should try to get signed pBooks

@RоryMcCune I have just sent her a note to that effect

She does come to Edin every August

@RoryAlsop heh

@RoryAlsop oh yeah I forgot the annual festival pilgrimage..

@RoryAlsop looks like Eireann did the tech review on it

^ nice bio

where did you find that pic?

I like Eireann's bio best :-)

@RoryAlsop eireann twitted it out twitter.com/blackswanburst

which I'm guessing was based on his bio being cool. I actually didn't know he knew sharon..

ahh - everyone knows Sharon :-)

@RoryAlsop That statement is false.

@TerryChia you will, my friend. You will

SHAAARON

@LucasKauffman You must know Sharon, right?

not that Sharon

and amusingly, not the girl on the cover of Sharon's book

@RoryAlsop think I might have heard a few interviews with her PaulDotComs podcast

@LucasKauffman ah - yes

she is lovely. Was our best social engineer in EY

@RoryAlsop yeah odd that cover

@RoryAlsop She worked at EY?

@LucasKauffman yup, for many years

@RoryAlsop cool

@LucasKauffman Probably left before you were born. ;)

I remember her demonstrating her new spy handbag, with covert video camera...

@RoryAlsop :O

why she leave?

@RoryAlsop That's so much better than the overt ones.

to be honest, I'm surprised she managed to retrieve the camera afterwards, from what I have learned about handbags is that they work a lot like black holes.

@LucasKauffman likely a similar reason that EY_UK lost a load of techhies :)

@RоryMcCune which was? :p

@RоryMcCune reckon so

@LucasKauffman The constant mocking you receive as a Big Four consultant? ;)

did I mention they are actually considering a techie track next to a manager track?

"you like being a techy. Great, during audit season if you're at all not busy, you're now an auditor". "also when you get experienced as a techy please become a salesman too"

@RоryMcCune yea that's typical

'cause all techies love being salespeople

Sorry, I'm just incredibly bored with my work right now. :P

@RоryMcCune sales is still ok for me :p

I don't mind it

@LucasKauffman compare and contrast to security consultancies where techies can stay techies and nerf weaponry abounds

what I do hate is people not understanding internally that our tech consultants can't do financial audit/account.

@LucasKauffman sure but in your limited available time which would you prefer doing....

@RоryMcCune Sales compared to audit?

I really don't like IT Audit

@LucasKauffman sales compared to techy stuff

so testers on downtime should be doing research or tooling work

not responding to RFPs

<shudder>

@RоryMcCune I don't mind going over to sales actually, but not everyone is willing to do that

^ Llama boy has joined the dark side.

RFPs are still ok but yea I do think we should get more time for tooling and research

@LucasKauffman sure some people will be cool, so as long as their flexible that's fine. From what I saw of it, they weren't

@RоryMcCune depends on the partner and the directors mostly

@LucasKauffman well the partner/director has to be willing to put in the effort to explain to HR / other partners that techies need to be handled differently for retention to happen...

@RоryMcCune they already know that now

it took a few techies leaving though :p

@LucasKauffman progress!

@RоryMcCune We have our training budget and they realized they shouldn't dare touching it

@LucasKauffman heh

@RоryMcCune We have the heighest training budget of the whole of EY Advisory

the accountants don't really understand that

it doesn't compute

@LucasKauffman The budget must be really tall.

@RоryMcCune u on skype?

@LucasKauffman at the moment they're improving my skype experience :op but it should load up in due course...

Awwh, @LucasKauffman had enough of my mocking. :P

@TerryChia no

@RоryMcCune no

@RoryAlsop maybe

@FEichinger yes

@kalina but what if it is tho'

@RоryMcCune no

Me trying to dual-boot Ubuntu on a pre-installed Windows 8 UEFI laptop last night:

OF COURSE THE GIF WON'T UPLOAD

@Simon Do you know how to computer?

@TerryChia Do you have any idea how much of a problem UEFI + W8 is for Ubuntu?

There's a freaking page on Ubuntu's website dedicated to that.

@Simon My statement is just a general one.

Ah. Then my answer is no.

UEFI isn't a problem with just Ubuntu.

Oh I could have bet on that.

@LucasKauffman I got that implemented in the UK. It worked for about 4 years

@RoryAlsop oh crap

@RoryAlsop it didn't work?

@LucasKauffman mostly for the reasons @unpingable mentioned

They started trying to bring in a sales or management bit

and the tech stream ended up having to be tech specialists/experts plus the normal manager skillset

and got paid less, generally, as techies tend to be less good at negotiating salary boosts than salesy types

@Simon It is possible to do that, but painful. Usually, you try to install, and either the machine won't boot the Linux, or it won't boot at all.

I did find some utility which "fixed" my hard disk, and I could boot either Linux or Windows 8.

Then I ditched the hard disk, put a SSD, configured the BIOS to stop being UEFI and revert to old-style booting, and installed Linux only.

Dual-boot sucks anyway.

@Simon I think @ThomasPornin is telling you to buy 2 machines.

@ThomasPornin I couldn't even get to boot on the USB key in UEFI, it was stuck on a boot loop. When I switched to Legacy mode, it did boot but dual-booting will be all messed up if I install it in legacy.

@TerryChia I have concluded years ago that if you want two OS on one machine, then you need VM, not dual-boot. With dual-boot you only have one OS up at a time, and often enough it is not the right one.

It's for the gf, I don't want this shit to be too complicated even though she's not a dumbass.

@RoryAlsop yep that's also typical big 4

So dual-booting is the simplest option, if I can manage to set it up properly.

@Simon Don't dual-boot if you want things to stay simple.

Blargh.

Don't dual-boot your gf's laptop if you want your gf to keep on being your gf.

Haha.

@ThomasPornin I doubt his right hand is going anywhere.

I might just convince her to wipe W8.

Oh and yesterday was another interesting experience with the Metro UI. I managed to shut down the PC when I didn't even want to do that.

@ThomasPornin I agree

@Simon only a retard button masher spam clicker would be able to achieve this

@Simon VMs are still easier than dual booting

@kalina s/retard button masher spam clicker/@Simon

@kalina Funny how that never happened to me on any other OS.

@Simon the thing you're saying happened is several really obvious menus in, whichever way you do

you can't "accidentally" bring up the charm bar and then "accidentally" click power and then "accidentally" click shut down

just like you can't "accidentally" win+R and you can't "accidentally" type "shutdown -i"

just like you can't "accidentally" right click on the start button and you can't "accidentally" bring up the shut down menu and you can't "accidentally" click shut down

@kalina I have no damn idea what happened, a box popped and when I tried to move the mouse with the track pad, it just selected an option even though I was far away from the box.

and you certainly can't "accidentally" hold down the power button for 5 seconds

@Simon right, Windows update

@kalina I beg to differ.

yeah, Windows update's default settings in Windows 8 are pretty intrusive

you want to disable auto restart when windows updates are scheduled to be applied

I didn't have to do that on my Mac.

I don't care if you didn't have to do it on your Mac

#yolo

I didn't have to explain to the last guy I spoke to that he was a spam clicking retard

if only we could always all get our own way

Moby made a collab with the creator of Subground \o/

The kick will probably be as hard as a @kalina kick but I guess I can live with that as long as it's not too Housey.

@kalina I dunno, my cat appears to be able to do just that. But maybe this is, indeed, in no way accidental.

@ThomasPornin no, cats are evil, was fully aware what it was doing

@Simon my kicks are quite hard, just no distortion at all

I don't like distorted kicks

I prefer nice, clean and punchy

I somewhat miss my Siamese cats :(

@kalina That's fair.

I don't mind pushing some distortion on my riffs or effects or... well, literally anything but the kicks

my basslines are always 3-6 instruments going through various distorters

I went through the documentation of ES1 (synth) and it's quite interesting what every single option can achieve.

yep

always useful to know what all the buttons do

With the sub-oscillator being 2 octaves lower and can be a different wave.

@kalina Which makes this whole thing quite overwhelming but I guess it's just a matter of time/playing with the knobs.

Hence when I said