Posts by Michael Wojcik

So you're doing an IoT project. Cute. Let's start with the basics: Security

Re: The biggest problem

Thing is it's difficult to buy dumb now.

Yes. Last time I bought a TV, Target had only one non-"smart" model on sale, and only two of them in stock.

You'll have a much easier time buying smart and not connecting it.

I hear anecdotally that some models won't work unless they're allowed to connect on initial power-up and occasionally thereafter. While it might be possible to reduce how often it's allowed to phone home, or spoof its server (I'm betting many manufacturers fuck up certificate validation), that sort of thing quickly becomes onerous for experts and impossible for regular consumers.

Appliance manufacturers have razor-thin margins, particularly at the low end. Data collection from "smart" devices is going to be very hard for them to resist.

Re: Am I the only person...

The OWASP Top 10 (updated for 2017, kids!) is great, particularly in the associated resources on their wiki. But it's web-focused, even if many of the issues have non-web analogues. Many IoT devices have web interfaces, but not all, and that's not the extent of their problems.

I'd suggest starting with the SANS Top 25 or the Howard / LeBlanc / Viega 24 Deadly Sins. Then hit 'em with some actual software security theory and SDLC practices.

Amazon, eBay and pals agree to Europe's other GDPR: Generally Dangerous Products Removed from websites

I'm sure some American will sue Amazon claiming exactly that.

I don't know why you were downvoted. I haven't bothered checking LexisNexis or Westlaw or anything, but I would imagine more than a few lawsuits have been filed against Amazon on all sorts of spurious bases.

Ridiculous lawsuits are an American tradition, and there's no reason to think that will change.

Re: Dangerous?

They have toll roads in the US...

And in the UK, but they are the exception.

At least by mileage. If Wikipedia is to be believed on this topic, the US has around 5000 miles of toll roads. Presumably most of those are on the numbered highway systems (defense highways and Interstates), of which there are around 161000 miles. So about 3%.

By number of vehicles or passenger-miles or the like, who knows? Toll roads do tend to be some of the more heavily-traveled routes. But mileage is one reasonable metric: you can do a lot of driving in the US without hitting a toll road.

And there are generally alternative routes around the tolls if you really want to avoid them. I often skip I-70 when I drive to Kansas, incidentally bypassing the tolls, simply because the back roads are more interesting. It adds perhaps half an hour to a 13-hour drive.

SUSE Linux Enterprise turns 15: Look, Ma! A common code base

Re: Cultural cloning and diminishing returns

Haven't the cultures that have developed negative associations with certain numbers thought of changing the word or pronunciation to remove the association.

What's fascinating is that this got two upvotes. Apparently we have at least three forum participants who have no understanding of what a culture is, or how one works.

I do like the prediction of economic doom, though. That's the most hyperbolic snowball argument I've seen in some time.

Re: How about Windows skipping 'Windows 9'?

Is 9 an unlucky number?

In the case of Windows, not nearly so unlucky as 10. At least for users.

(Just had my corporate Win10 Enterprise laptop do another unannounced forced reboot last night. Second time this week. Unforgivable.)

Re: SuSE Linux

And SuSE is sponsored by M$, so better not trust them with your data.

SUSE is a division of Micro Focus. I don't know what sponsorship you're referring to (Microsoft isn't a current openSUSE sponsor, for example), but believe me, we give Microsoft a lot more money than we might get from them.

It's been "SUSE" for years now. The mixed case was abandoned long ago.

Re: not uncommon

What is the opinion of the native Americans on the subject of floor labeling?

Depends whether you're talking multi-level tipi, wikiup, longhouse, .... With hundreds of indigenous American ethnic groups, you'll get lots of different customs, mores, superstitions, and so on. As I'm sure you know - but it shows how foolish the GP's rant was. Even with the homogenizing effects of mass media and widespread travel there are hundreds of distinct local customs and the like just in the US.

(And, frankly, anyone who eats in the US today should be really, really happy that we didn't stick with either native cuisine or what the initial European settlers brought with them.)

But then the AC you were responding to is clearly an ideologue uninterested in actual facts or reason, as all of this is prima facie obvious.

Re: not uncommon

None of the Americans would take a room on the 13th floor.

How many Americans were there? The triskadecaphobic thing is somewhat common in the US, I suppose - though I don't think I've ever met anyone who admitted to it - but it's hardly universal. Most of the people I know treat it as a joke.

And then, in fine untenable fashion, we have the "baker's dozen", which has positive connotations. Even though it's the same number.

Re: "moronic ancient superstition"

Ha ha ..... my upvote made the count 13 !!

You fool, you've doomed us all. I wager not one reader of this forum will survive the millennium.

(If I'm wrong, let me know in 3001 and I'll send you a Dogecoin, which I believe will be the common currency then.)

Amazon staffers protest giant's 'support of the surveillance state'

Re: Not really knowing...

Having worked for Amazon (out of pure curiosity) I suspect they'll be shipped off to the arctic when nobody's looking.

But will shipping be free? Is Bezos a Prime member?

Simply disappearing the employees seems inefficient. Surely Whole Foods could find some shelf space for Amazon Soylent Food Product.

How a tax form kludge gifted the world 25 joyous years of PDF

Re: A beast of many things...

a freaking filesystem, because that's really what PDF is

When you want to flatten complex documents into a single file, you're probably going to end up with a compound file format of some sort. Open Document Format is a compound file format - it's just a zip archive, in fact. OOXML and XPS are compound file formats. EPUB is a compound file format.

The alternative is a single non-compound format that encompasses all the types of data you might want. That's worse: it's more cumbersome to define, document, implement, etc. With a compound file format, it's trivial to build toolchains that operate on only some parts of the entire document - the explode / filter / implode pattern.

HTML and its siblings can get away with not being compound because they present a de facto remote filesystem to the user agent. They don't try to flatten everything into a single byte-stream blob.

Re: Open format !!

If we want things that can be read for a long time to come, is there anything as robust as PDF?

Digital document preservation and archiving is a large and very active field. As with any such, the guesses non-specialists make about it are not likely to be particularly accurate or useful.

There's a decent short introduction to the subject by David Anderson in the December 2015 issue (58.12) of CACM. Anderson mentions the #nodigitaldarkage discussions on Twitter that were sparked by Vint Cerf's "Digital Dark Age" arguments, and such projects as POCOS and E-ARK. Interested readers may also want to investigate historical efforts such as Acid-Free Bits or the long debates about human-readable versus machine-readable formats, and so forth.

Re: PDF can be cool... if you stay away from Adobe

PoC||GTFO is a wondrous thing (and let us not forget that two volumes are also available as lovely hardbound books). But using it as an example of the virtues of PDF is a bit like using the Bugatti Chiron to argue that cars are pretty fast. It's something of an edge case, surely.

It nicely illustrates how narrow and limited many commenters experience is; you wouldn't use Word to write a musical score, however, PDF allows those without the relevant application to read your score.

1. Terrible thing X is useless for application A.

2. Sometimes-useful thing Y is useful for application A', which is related to but distinct from A.

3. Therefore people who do not believe Y is wonderful have limited experience.

I think your syllogism needs work. Or, preferably, nuking from orbit. Care to try again?

Re: PDF has its uses I suppose

So you want to do away with a standard, so when I refer you to page 404 of the HTML status code manual, you get something totally different because in your rendering of the manual the relevant material is on page 418 or even 1415...

The vapidity of this example (there is no "HTML status code manual") aside, the problems with using page numbers for citation have been well known since long before there were computers. That's why, when we're using responsive-layout documents, we don't use page numbers to cite passages.

This straw man was scattered to the winds long ago.

Re: "Placement and styling is important" ...

The good old printed book is much more than "just the text"!!!

Yes, but that excess is often irrelevant to readers. Anyone with even cursory knowledge of textual scholarship knows that audiences generally consider all editions of prose books to be essentially the same, even though they may be typeset completely differently.

There are certainly cases where typesetting matters to more than a small subset of the audience, but those cases are the minority. And most of the professional book designers and typesetters I've heard discuss the subject are well aware of that.

Precision layout is mostly important to the people who lay things out. For most other audiences its effects tend to be detectable but not hugely significant.

Re: Format of choice for immediate offline reading, easy sharing or simple portability

you can be 99% certain that it will display properly and legibly on any and all computing devices

I don't know what magical unicorn devices you use, but the vast majority of PDFs I have aren't legible on my (Android) smartphone or my Kindle. A small rectangular subset of a given page may be legible at any given moment, but scrolling half a dozen times just to read a few lines is not a usable reading experience.

PDF is a non-responsive format, and as such is inherently limited on what device form factors a given document can be usably rendered.

(I won't even bother noting that the vast majority of "any and all computing devices" don't even have a display, and chalk that phrase up to lazy thinking.)

Re: Format of choice for immediate offline reading, easy sharing or simple portability

Placement and styling is important.

Anybody who cares about communication should appreciate how design affects interpretation.

I have a Master's in digital rhetoric, so I'm well aware that design affects interpretation. I've read scores of scholarly articles on the subject, presented on it at academic conferences, done user research, etc.

It's naive to claim that the rhetorical effects and additional information channels afforded by precise control over layout are an absolute good, or that they outweigh the tremendous advantages of responsive layouts, particularly when addressing a large and diverse audience using a wide array of devices. Unreadable documents have proven very poor at communicating and persuading.

Frankly, based on your comment, I rather doubt you've studied this area in any depth.

Re: Jobs Didn't Introduce Typography to Computers

I guess the author was referring to the NeXT's display system.

He specifically mentions the LaserWriter in the same paragraph.

Re: Jobs Didn't Introduce Typography to Computers

Yes. The line about "Steve Jobs introduced typography into computing" is complete rubbish.

TeX was released in 1978, so a good 7 years before Steve Jobs and the LaserWriter. Even the first version of PostScript was only released in 1982.

troff was just one descendant of CTSS RUNOFF, from 1964. Arguably RUNOFF didn't do much in the way of "typography", but it did lay out text. troff appeared a couple of years before TeX (circa 1976) and did quite a lot of typesetting.

Perhaps the "Apple fanboi legend" and some paywall-protected Forbes page aren't ideal sources.

JURI's out, Euro copyright votes in: Whoa, did the EU just 'break the internet'?

Re: Citation needed

Encryption to specific recipients and groups, and serving files off one's own machine are perhaps going to get more interesting.

It's easier than that. Use a public hosting firm that lets you configure TLS. Run your own CA (it can be a toy/demo CA like OpenSSL's) and issue client certificates to anyone you want to grant access. Leave index pages and their resources unrestricted, but require a client certificate for anything you don't want visible to hoi polloi, Google, the IP bottom-feeders, etc.

In other words, "encryption to specific recipients" is already provided for in existing web standards and infrastructure. Browsers and servers will handle it automatically; in fact, many TLS stacks will handle it automatically for all the applications that use them.

I've never figured out why lobbying is legal.

In the US, because it's protected by the Constitution, specifically by the First Amendment. The freedom of speech clause obviously applies, and it's not difficult to argue that a law attempting to excessively restrict lobbying would fall foul of the petition-the-Government clause.

It's even possible that SCOTUS would find a prohibition on lobbying a violation of the assembly clause, as they did with California's blanket-primary law in California Democratic Party v. Jones.

More generally, how do you think lobbying should be restricted under a democratically-appointed republican form of government? What mechanism would prevent "lobbying" (already difficult to define in a sufficiently specific way) while still letting constituents interact with their representatives?

If you're concerned about the money, how do you eliminate all quid pro quibus? No representative, nor any party body, nor any recipient the representative might care about, can benefit in any way from any contribution made by anyone, ever? Such a proposal is patently absurd.

Re: bad for small sites

The vast majority of the ones rejected are because they URL has already been reported. IOW they aren't malicious they are duplicates, the URL in question has already been removed.

Citation, please. To actual data.

Are your IoT gizmos, music boxes, smart home kit vulnerable to DNS rebinding attacks? Here's how to check

Re: 192.168.1.xxx

The proof of concept exploit is hardcoded to 192.168.1.1/24. He should have mentioned that, and if he had, perhaps The Registers might have mentioned it

The article does now mention it. I'm assuming that's an edit, since a number of people complained.

Personally, I'm far more worried that nearly all the responses here are about the proof-of-concept rather than the actual problem. Those who do not understand DNS Rebinding are doomed to remain vulnerable to it.

Of course, that's what happens when you make typical consumers de facto network administrators. It's not reasonable to expect even most people in the IT industry to be aware of and understand all these vulnerabilities; most people simply don't have that luxury, even if they had the inclination. The onus has to be on the manufacturers of these IDIOT¹ devices and the infrastructure² they rely on.

¹Internet of Dumb and Inappropriately Online Things.

²Including poorly-designed protocols like SSDP and crap devices like consumer-grade routers.

'90s hacker collective man turned infosec VIP: Internet security hasn't improved in 20 years

Re: 56k bullshit

the days of Procomm & Qmodem

Telebit Trailblazers were my drink of choice, before I had a 56K leased line. SLIP over those for interactive stuff, then drop the SLIP connection and use the modems' uucp g-mode spoofing for bulk transfer. Worked fine for editing code with vim and the like.

Re: Well

I am not surprised Internet security has not improved.

There's probably no useful definition of "Internet security" that's acceptable to actual security experts, and claiming the security of any non-trivial system has or has not "improved" is a dubious proposition as well. But under any reasonable threat model, software security has improved significantly over the past few decades, in the senses of removing many prominent branches from the attack tree and increasing costs for attackers. It simply has a long way to go yet.

Re: Liability

All code is written by offshore idiots to the lowest price

Even just the "offshore" part of this is patently untrue, probably for any continent. I haven't verified that there's anyone writing code in Antarctica at the moment, but unless that's where you live, you're prima facie wrong.

Re: bzzt yourself

BIX and Delphi both had some commercial Internet access in ''92.

True, and we could certainly quibble about whether 1992 was pre-commercial-Internet. I think most people who remember the historical details would be more likely to call 1991 the watershed year for commercial Internet; that's when CIX was formed and ANS CO+RE opened for business.

But 1992 was when ANS and CIX agreed to interconnect, and when the SAT Act changed the NSFNET usage terms to allow general commercial traffic. (There had been limited "experimental" use of NSFNET for some commercial traffic as early as '88.)

So it's while it's inaccurate to say that there were no commercial Internet users in 1992, most commercial users got connections after that year.

What's all the C Plus Fuss? Bjarne Stroustrup warns of dangerous future plans for his C++

Re: Disagree....Because it's been done

Have a look at RUST.

Rust (it's not an acronym) suffers from programmer resistance to the borrow model, a community widely noted for hostility to being questioned, and (as someone else noted below) a tendency to change without preserving backward compatibility.

It may yet outgrow those challenges, but history suggests that new programming languages have an uphill battle, and Rust hasn't done a great job so far of building momentum.

I have no dog in this fight myself - I haven't done anything significant with Rust, and I'm perfectly happy with the borrow model. (I designed a toy language with something conceptually similar years ago.) But I'd be quite surprised if Rust is one of the major languages in, say, ten years.

Re: C and C-style C++

Stroustrup has always been a blowhard, for me his ship sank almost 20 years ago.

I disagree with Stroustrup on a number of points. I've argued with him in public, on Usenet. I'm certainly not an unalloyed fan of C++.¹

But your comment is small-minded and foolish. Stroustrup has made many excellent contributions to computing, a good portion of which have nothing to do with C++, such as his essay (written while Chair of CS at Texas A&M) deploring the resistance to programming among academic computer scientists.

The article links to his papers on the history of C++ and programming languages in general, which are a good example of Stroustrup as an academic. I'd like you to point out where in them he's being a "blowhard".

¹A decent, fairly clean language, hidden under a huge mound of ugly and unintuitive syntax, grievous legacy features, unfortunate complications, and obvious failings (some of which S. mentions in the article) which have yet to be remedied; most frequently seen in fevered visions after looking at far too much extant C++ code, which is nearly always execrable.

Re: C and C-style C++

The "two languages" problem you describe is exactly what Julia is designed to overcome. Code in a modern language, with a REPL and Juyputer notebooks.

I like Julia, but I don't see Julia growing outside the HPC and data science domains. As with most programming languages, its advantages aren't compelling enough to retrain large groups of developers, much less convert existing codebases.

Jupyter (which is what I assume you're referring to) definitely has its applications - if I were doing quantitative research I'd definitely be considering using Jupyter notebooks, whether we were using Julia or Python or some other supported language. But I don't immediately see much use for it in typical system or business programming, even if that happens to be done in a language Jupyter supports.

Unbreakable smart lock devastated to discover screwdrivers exist

Re: Yeah - but if I am a "common criminal" I'll definitely find another non-indiegogo to pawn

A pair (male and female) of these is as effective as two 24x7x365 guards armed with submachine guns.

Really? What if the attacker is, oh, let's say, in a car?

Mind you, I'm not advocating for attack dogs or attack humans. But I think your threat model is a little simplistic if it finds those two mitigations equivalent.

(There are, of course, attack classes in which the dogs are more effective. They're less susceptible to threats against family members, for example.)

Boffins offer to make speculative execution great again with Spectre-Meltdown CPU fix

Re: Strange Charcters

Just some of our local kooks. What has the deal ever been with online kooks? They are a species unto themselves.

Personally, I'm glad to have them, as long as they don't become too disruptive. They give the dish some flavor.

Hell, I even enjoyed Eadon. ("Overwhelming, am I not?")

Re: I'm going to use up my stupid question of the day quota here...

Am I correct in thinking that if speculative execution obeyed memory access restrictions (no user process peeking at lower ring address spaces like the kernel) regardless of whether or not that memory is cached, then these problems would go away?

No.

The Spectre vulnerabilities use side channels to extract information. They don't "peek[] at lower ring address spaces".

Meltdown is a Spectre variant that leaks privileged memory, and the (obvious) fix for it was to prevent spec-ex from crossing privilege boundaries. But Meltdown is only one of many Spectre variants.

Re: Speculative versus parallel execution

Another solution is to pause execution until the outcome is known. While this pause lasts, you can have another thread use the execution units.

You want to have a context switch every time a branch causes a cache miss? That would be a Bad Thing.

There's a reason why Tomasulo created his eponymous algorithm for OOO in the '60s, and hardware implementations followed shortly thereafter; and commercial spec-ex machines became available in the '80s. (The Stretch did it even earlier, in the '50s, but its misprediction recovery was so expensive that it would have been better off without prediction and speculative execution.)

There's a reason why nearly 20 years of SpMT research hasn't done a hell of a lot to improve thread-level parallelism.

And that reason is that parallelism is both difficult and expensive. It's expensive because independent parallel units eat up your hardware budget quickly, and keeping those units fed requires expensive context switches. You can play with the time/hardware trade-offs of context switches (using extra register sets and whatnot), but you're still paying one way or the other.

There is no practical, cost-free, bumper-sticker solution to Spectre-class problems. It doesn't matter whether that word is "SafeSpec" or "parallelism" or "unicorns".

As somebody who works in this field, these conclusions seem unlikely to be true.

The performance claim is based on running a suite of benchmarks against an emulator. Hard to say what would happen with real silicon, but it's not a complete SWAG.

The hardware overhead, on the other hand... the paper says 17% increase in area and 26% in power. That may well be acceptable for many applications, but it's a pretty big price to pay for others.

There is, after all, no free lunch. Resources you take to create shadow structures can't be used for additional sets of primary structures. The paper describes some clever ways to optimize shadowing and reduce the overhead (a naive implementation would make much less efficient use of cache lines), but as you suggest, there's going to be a cost.

And since this approach requires significant CPU redesign, it's not likely to produce a chip that lets the OS flip a switch - SafeSpec or use-all-resources-for-primary. So CPU manufacturers would have to decide whether they want to cater to security-focused or performance-focused customers. Or manufacture two design families.

And, of course, we still have other side channels ("other structures would have to be hardened too").

Don't break out the party hats yet.

Re: It's time to remove "Speculation" from CPUs

Back in 1996 when the Pentium 1 with 200 MHz was modern, we had no such thing as "Speculation" inside the CPU.

Perhaps you didn't. Plenty of us did.

Then came the Pentium Pro and Pentium II and introduced flacky "Speculation" to improve speed on behalf of security.

The PPro and P2 did not invent speculation. And it is probably not "flacky", though it's hard to say, since that appears to be your personal coinage.

Re: Hard as I try...

One solution is to make high precision timers inaccessible to JavaScript.

There are many alternate timing mechanisms. I've posted the link to the paper in Reg discussions of Spectre-class attacks before.

Or to just wonder how, and why, an interpreted language that is designed to execute within any compliant webbrowser, on any Operating System, on any hardware/virtual architecture can somehow access the low level CPU state.

Why do you think it needs to? The original Javascript Spectre attacks did not "access the low level CPU state".

The various Spectre variants are well-documented and described in detail in many places. It's not hard to learn how they work, or why bumper-sticker solutions don't significantly reduce the class of side-channel attacks.

Re: Hard as I try...

At least one person proposed something even simpler, in essence saving the speculative state along with thread/task state, and restoring it on thread/task switch.

I'm curious why you think this would close the cache timing channel. Or indeed most other side channels.

Thought the AT&T Time-Warner tie-up was scary? Comcast says 'hold my beer'

This merger is the first of many other mergers which will see just a few mega corporations controlling every facet of our lives.

Er... I can think of some facets that will remain unaffected by this sort of merger, unless Comcast are going to buy, say, Kohler and Georgia Paper too.

I'm no fan of the ATT+TW or Comcrap+Faux mergers, but I think it's a little soon to declare the sky is falling. Even in the realm of broadcast entertainment I have no trouble finding competing content providers. And (hypothetical) attempts by "mega corporations" to use their ISP services to favor their own content streams wouldn't materially affect my life; if I can't get broadcast entertainment I want, I'll do without. There are plenty of other entertainment sources I'm far more interested in.

(I don't bother with broadcast news. Synchronous media is expensive, in terms of opportunity cost - its information rate is miserable compared to reading. Since the quality of broadcast news appears to be quite low, it's simply not valuable enough to watch.)

Intel chip flaw: Math unit may spill crypto secrets from apps to malware

Re: Pedantic spelling

I guess people are lazy and just want to type the fewest characters possible hence 'maths' ?

If you do the m, you'll find "m" is even cheaper.

(On a more serious note, the thesis quoted above is prima facie incorrect. People will use the term they're accustomed to, which will be the term predominant in their speech communities. Just as with any other regional variation in diction and usage.)

Dinosaurs permitted to mate: But what does AT&T Time merger mean for antitrust – and you?

Re: "The US approach maximises consumer welfare"

@Andrew: Even with that simplistic definition of Consumer Welfare, it does not necessarily follow that prices will be lower under a monopoly.

Who claimed otherwise?

Andrew's point is that the official guiding principle of US antitrust efforts is "consumer welfare", which in this case is a term of art with a specific definition. As he wrote in the article, European antitrust is based on a different official guiding principle.

And this official guiding principle is one reason for the historical lack, relatively speaking, of antitrust activity in the US. That's what he was writing about in that section of the article. He did not claim that US antitrust activity is necessarily of actual benefit to its citizenry.

Why so many readers have trouble comprehending this rather simple point is beyond me.

Sorry, wait - I just remembered this is an Internet forum, where thinking first is considered gauche. Never mind.

Creepy software knows what you are about to do... to that poor salad

Re: In the mean time on our planet

Food availability is widely held to be a political problem,¹ and it will likely be some time before AI is of much help in solving those.

Though having said that, I'll note that many political problems are at least in part a matter of making persuasive arguments,² and work in computational rhetoric is progressing. Whether that results in solutions to your or my liking is an open question.

¹Though I've seen arguments to the contrary.

²Indeed one might characterize most of the political problems in the US this way.

Re: Training set selection

"Hello, Microsoft? We asked Cortana^TM for a tossed salad, and for some reason it keeps telling us it needs a pair of buns."

One day there'll come a point where we'll no longer need to exist

Why do we need to exist now?

I mean, I'm happy enough to keep doing so, but I can't define any external need for it.

Re: Did they invent a magical image recognition system, or is something left out?

The image recognition system was provided information and has managed a great training set that has actually allowed it to automatically determine, within limits, what culinary task I'm doing. This would be revolutionary news...

Er ... no, it wouldn't. They had a large labeled training set, which trained the past-action RNN to label, with some accuracy, new inputs (videos). That's the sort of thing we do with RNNs all the time. Why do you think that's revolutionary?

Here is another example of using NNs to label video. It's a CNN-RNN network instead of an RNN-RNN one, and it's tagging input videos rather than predicting future content in the input video, but conceptually it's not much different.

Or see this post on continuous classification with TensorFlow-based RNNs. (RNNs are used for this purpose rather than CNNs because they're more amenable to learning vector sequences.)

So, we have substantial prior art showing we can use NNs to label video sequences. Then we output a series of labels, and use another model (this team used an RNN, but for something like this even a HMM might do pretty well) to predict the next label. Getting decent accuracy might be tough, but the basic structure is well understood.

Apple will throw forensics cops off the iPhone Lightning port every hour

The utility of this information can be tremendous.

The utility of having an effective privacy right against the government by the citizens is several orders of magnitude greater, however.

Well said. Some time back I saw a presentation by a US ADA about the use of phone data in a number of actual investigations, including what was gathered and what permission (warrant, subpoena, nothing) was required to get it. The focus on cases like San Bernardino doesn't do a good job of representing the whole picture.

In a civil society there's always a complex trade-off between civil rights and law enforcement. Reasonable people will disagree on what that trade-off should be. Personally, I favor strong privacy rights, but I don't believe that position is prima facie correct.

Re: Wait for the AI...

No matter what scheme you come up with now for devising passwords at some point will be crackable.

Handwaving bullshit. A completely untestable and thus vapid claim.

Ultimately, the cracking of passwords comes down to the N vs NP problem.

Your argument might be a little more persuasive if you knew that the name of this problem is "P versus NP", not "N versus NP". But probably not, so I wouldn't worry about it too much.

Your (corrected) claim is common, but it's also wrong. Even if P=NP (very unlikely), there are still functions with asymmetric effort, where both F and F-bar (its inverse) are in P but F-bar has a worse polynomial growth rate.

for those who may not be aware of the N vs NP problem, It asks if a problem whose solution can be quickly verified can also be solved quickly.

No, it really doesn't. The P-versus-NP problem asks whether a particular (isomorphic) class of functions that are poly-time verifiable also have a poly-time solution. Polynomial time is not necessarily "quick".¹

Even with large-scale general quantum computing and problems that are also in BQP, functions of this sort generally only get a square-root improvement on running time. So you double the length of the "hash".

It's true that passwords are terrible authenticators, under pretty much any metric other than ease of implementation and familiarity. And passphrases really aren't that much better. And yes, multiple authenticators (preferably weighted N-of-M authentication, not the sort of half-assed 2FA or 1.5FA we typically see) looks like the only viable direction to go, given the current IT landscape. But trying to derive those conclusions from P-vs-NP is cargo-cult analysis.

¹Matt Scala had a good example of this once: he derived and proved an algorithm, for an actual problem he was working on, which had a best-case polynomial time with a huge exponent. It's in P, but completely unusable for non-degenerate cases.

Re: Easy good passwords, here I go again...

if the password is alphanumeric and an actual word or combination of words, dictionary based attacks drastically shorten the time to "guess" [blah blah blah]

Sigh.

Arguments like this are just handwaving without some actual statistics, or at least back-of-the-envelope approximations.

A recent version of the aspell US English dictionary contains around 204800 words. Using an xkcd-style four-word phrase (which gives a passphrase on the order of 20 characters, quite easy to type reliably for many users; I routinely use passphrases twice that long) gives about 70 bits of entropy. That's assuming words are chosen with equal distribution from the list; it assumes nothing about, say, the per-symbol entropy of English.

Note it also assumes the passphrase contains no spacing, punctuation, or non-letter symbols, except the ones that appear in the aspell US-English dictionary (things like apostrophe and hyphen). Those can easily be added by the user in a meaningful fashion, increasing the entropy. It also assumes monocase, or a case-insensitive verification mechanism; if the system is case-sensitive, we can use mixed case as well.

What's 70 bits of entropy worth? Compare it with a random (equal distribution) password drawn from mixed-case English letters, numerals, and a dozen non-alphanumerics. That's 64 symbols, or 6 bits of entropy per symbol. So 70 bits of entropy for the passphrase is just shy of a 12-character password using this scheme.

If you can make a million attempts per millisecond, brute-forcing a 70-bits-of-entropy passphrase takes a little under 19 thousand years, on average.

The trick with xkcd-style bag-of-words passwords is to generate a number of unbiased phrases from the dictionary, then pick one you can remember by visualization, "newspaper headline" interpretation, or whatever. The relatively low per-symbol and per-word entropy of natural language really doesn't matter when it comes to resistance to brute forcing, once the phrase gets to be even a few words long. Models only do well against plausible natural-language phrases.

There's a commonplace among infosec folks that xkcd-style passphrases are not particularly strong. Schneier subscribes to it in this post, for example, talking about the password-cracking bake-off Ars Technica hosted back in 2013. But it's not the scheme itself that's broken. The weakness comes from weak use of it - from users choosing words from too small a dictionary,¹ or creating passphrases that are too small.

(Also, the Ars piece only worked with one attack mode - cracking a corpus of unsalted MD5 hashes. While Schneier generalizes that to "password crackers know to combine words from their dictionaries", even with smarter candidate generation, stronger key-derivation functions such as Argon2 are going to slow brute-forcing tremendously.)

Even then, terms like "broken", "weakness", and "too small" are misleading. Absolutes are always inaccurate when discussing security. What we need to talk about is the risk (probable loss) under a threat model. My probable loss for someone brute-forcing my Reg password is very low - I don't have much at risk here, under my threat model. And the probability of someone brute-forcing it is relatively low, because most attackers have little incentive to do so. So my password only has to be strong enough against brute-forcing to lower that risk to a point that I'm comfortable with.

¹Generally that means "user has a larger dictionary, but only chooses familiar words, and has a relatively small working vocabulary in the first place". For a random-word-phrase scheme, the user's "dictionary" is the set of words they're willing (with high probability) to use.