Hacker Newsnew | past | comments | ask | show | jobs | submit | more throwawayjava's commentslogin

This was my original reaction as well; it's not at all surprising that a policy supported by the left would have supporters who score lower on RWA relative to their right-wing counterparts on the other side of that policy. It would be interesting to see more granular survey results, since there is a bucket of questions that do a better job of measuring the A in RWA.


> I have no idea what this is supposed to measure

On its own, it's not supposed to measure anything. There's more than 1 question in the survey for a reason. And incidentally, the questions that people are most likely to criticize -- the first two -- aren't even used in the assessment; they're throw-aways.

> To see why thia question is ridiculous: Imagine asking this question to people in Tehran during the Shah. In that case, the “proper authorities” were the western-aligned Shah government, while the “rabble rousers” would be the Islamists. But it was the rabble rousers’ view that was authoritarian.

From chapter 1 of the book in which this scale was introduced: the test, which was designed to measure right-wing authoritarianism in North America, will probably fall apart in markedly different cultures.

That same chapter goes on to directly address almost all of the criticisms I'm seeing in this thread.

I think it's wrong to construe one measure of "right-wing" (explicitly not "conservative") authoritarianism with "authoritarianism" generally, as the title of this piece does. But there's nothing wrong with designing assessments that are intended to work only in one particular culture.


The question I quoted was not one of the two throw-aways. Even with the caveats you mentioned, this survey does not actually measure authoritarianism in any meaningful way. Take three of the things addressed in the questioning: animal rights, abortion, and LGBT rights. You can frame any of these things to make both sides seem “authoritarian” or anti-authoritarian. For example, public accomodation laws can be viewed as an impingement on the right to freedom of association, or a vindication of the right to participate in the economy free of discrimination. Abortion restrictions can be viewed as vindiction of the fetus’s right to life, or an impingement on the mother’s bodily autonomy. The article simply picks a framing for each issue, and labels the conservative point of view “authoritarian.”

But in reality, these disputes aren’t about authoritariansm versus non-authoritariansim. Both sides believe that the government may restrict the natural freedom people would have in the state of nature to protect certain things that society recognizes as “rights.” These disputes are about what rights society is willing to recognize and enforce, and what to do when those rights conflict.


> The survey falsely portrays these disputes as being about “authoritarianism” versus (presumably), “non-authoritarianism.”

You seem to feel that this survey is a definition about individuals? That's not how the survey is intended to be used, and the fist chapter of the book that introduces it is full of caveats that the survey is not intended to be treated as a definition about individuals.

The author of the survey is very explicitly NOT providing a definition of authoritarianism or a litmus test for individuals. He is providing a survey whose scores correlate with right-wing authoritarianism in North America in the late 20th century.

The survey is a tool, the author of the tool is very explicit about what that tool is and is not for. I, for one, don't get angry when my hammers don't work as screw drivers. And also don't get angry at the inventor of the hammer when some random third party messes up my stuff by using a hammer on my screws...

> Take three of the things addressed in the questioning: animal rights, abortion, and LGBT rights

The survey is designed to measure RIGHT-WING Authoritarianism. That is, the conjunct of being both authoritarian and also right-wing in that authoritarianism.

If we asked 20 questions about authoritarianism but never mentioned anything about being right-wing, we might have a good test for authoritarianism but not a good test for RIGHT-WING authoritarianism...


I’m not assuming this survey is a definition about individuals. The survey takes the beliefs of a group (conservative christians) and simply labels them “authoritarian.”

As to your point about right-wing versus left-wing authoritarianism: you’re missing the point. Imagine a counter-part to this test that labeled support for the Civil Rights Act of 1964 as “left-wing authoritarianism.” Technically, yes, that law restricts the freedom of individuals to conduct their business and associate with people according to their own choices. But labeling that “authoritarianism” would (I think rightly) be perceived as biased by most liberals. The article makes exactly the same mistake. It takes a lot of views and unfairly labels them “authoritarian.”


> The article makes exactly the same mistake. It takes a lot of views and unfairly labels them “authoritarian.”

And as I said in my original post:

"I think it's wrong to construe one measure of "right-wing" (explicitly not "conservative") authoritarianism with "authoritarianism" generally, as the title of this piece does. But there's nothing wrong with designing assessments that are intended to work only in one particular culture."

> The survey takes the beliefs of a group (conservative Christians) and simply labels them “authoritarian."

No, it doesn't, and I think that's an unfair characterization of conservative Christians.

1. I only count 11 items that say anything at all about conservative Christian beliefs. The other 9 are totally orthogonal. Even if we score those items -4 (which most conservative Christians wouldn't; see item 2) and the other items +4, we end up in the low 100s. And more realistic answers typical of non-authoritarian conservative Christians in the midwest end up with scores in the 90s (again, see item 2).

2. of those 11 items that make reference to "Christian/family/traditional values", many would not elicit a "strong" score from all or even most conservative Christians following the survey's instructions.

Example 1: "Everyone should have their own lifestyle, religious beliefs, and sexual preferences, even if it makes them different from everyone else."

Most conservative Christians will strong-disagree with "sexual preferences".

But what about "lifestyle" and "religious beliefs"? Many conservative Christians support different lifestyles; e.g., homeschooling. And a STRONG majority support religious freedom.

According to the survey's instructions, which tell use to average the -4 re: homosexuality with +N for the other two items, this question would elicit a weak disagree or maybe even a weak agree from many conservative Christians.

Example 2: "God's laws about abortion, pornography and marriage must be strictly followed before it is too late, and those who break them must be strongly punished."

Many conservative Christians believe that divorce should be legal and pornography, though damaging, should not be outlawed. So using the survey's averaging rule, this might even end up as a "weak disagree".

And there are many questions that ask for moral comparisons between "traditional" and "non-traditional" people, but the bible is very clear on exactly this question: we are not justified by our acts.

It is true that there is a certain brand of conservative Christian who believe we should have a "strong leader" who uses the state to forcibly impose their personal conservative values on the rest of society. Those people would score highly on this test. But that set of people is definitely not the same as the set of "conservative Christians", because many "conservative Christians" have non-authoritarian values (and many right-wing authoritarians are not Christian.)

> As to your point about right-wing versus left-wing authoritarianism: you’re missing the point. Imagine a counter-part to this test that labeled support for the Civil Rights Act of 1964 as “left-wing authoritarianism.”

Exactly zero questions in this survey mention a piece of legislation, and not a single question on this test automatically labels RWA for agreeing with a given piece of legislation. There are multiple types of questions for a reason.

There exist left-wing authoritarian tests. And yes, they do ask questions about typical left-wing issues. And yes, they should. Because measuring LEFT-WING authoritarianism without even bothering to ask if someone's beliefs are consistently left-wing would be silly and would miss the point...

Maybe DeMorgan can help. A low (L or R)WA score just means:

NOT (LR-wing AND authoritarian) == NOT LR-wing OR NOT authoritarian.

A below-average score could mean (LR-wing AND NOT authoritarian).

A below-average score could mean (NOT LR-wing AND authoritarian).

A below-average score could mean (NOT LR-wing AND NOT authoritarian).

A low LR-WA score does not imply "not authoritarian". It implies "not authoritarian in this particular way". A very high score indicates "authoritarian in this particular way". Borderline scores could mean "authoritarian but not in this way" or "this way but not authoritarian".

Survey design is difficult, and interpreting survey results is also difficult. A survey designed to measure the intersection of two properties will give higher scores to people who have one of those properties than people who have zero of those properties. That does not mean the survey is flawed; it means that the results have to be used with this effect in mind.


Again, I think this comparison is a bit contrived -- an entire industry vs. one product at one company. Google's search engine disappearing is more analogous to a railroad or pipeline subsidiary of an oil company. Life would continue, the railway/pipeline would be rebuilt, and in 10 years everything would be back to normal.

If you want to make an accurate analogy, compare the disappearance of oil to the disappearance of all modern computing. Both would be catastrophic.

Society is complex and has lots of interdependencies.


>Again, I think this comparison is a bit contrived -- an entire industry vs. one product at one company. Google's search engine disappearing is more analogous to a railroad or pipeline subsidiary of an oil company.

You're focusing on the wrong thing on my argument. I used Google as a stand-in for web search in general -- regardless of company. Because for me, and most of the west, Google is that: all of search.

My argument wasn't really about the one company, or about not being able to replace the company (we could just use Bing and make do with it if that was all I meant it for).

What's more, I also think my argument would hold even if taken to mean the web in general (and not just search). It would still be less catastrophic (and quite mild after a small re-adjustment period) than losing the fossil energy sector.

>If you want to make an accurate analogy, compare the disappearance of oil to the disappearance of all modern computing. Both would be catastrophic.

Both would be catastrophic but the latter less so. We did fine with minimal to no computers in the 60s. We can always go back to that level, which is not that savage or even old. Without fossil fuels (and no transition period to slowly replace them in toto with alternative sources) there would be zombie apocalypse levels of mayhem.


> Without fossil fuels (and no transition period to slowly replace them in toto with alternative sources) there would be zombie apocalypse levels of mayhem.

I generally agree with your argument, but I'm not sure if, with sudden loss of general-purpose computing, we wouldn't have small-scale zombie apocalypse on our hands. The world has grown in many ways since the 60s - including population and complexity of supply chains. All of this would have to be scaled down, and here "scaled down" means mass loss of life.


For starters, the sudden loss of all modern computing would cripple oil extraction, transportation, and refining for years on end.


Would it? We transported as much oil in the 70s without "all modern computing". Faxing maybe...


> Would it?

Yes.

> We transported as much oil in the 70s without "all modern computing".

No, we didn't. Global production (and consumption) has close to doubled since 1970.

Also, "is this possible to do without computing?" and "would the sudden disappearance of modern computing seriously interrupt the way that we currently do this?" are very different questions.


>No, we didn't. Global production (and consumption) has close to doubled since 1970.

Doubled is not that much to cover.

Besides did it double because of increased demand (more population, more developing nations, etc) and efficiencies in drilling (e.g. fracking or how it's called), or because of "modern computing"?


Demand and production of oil are relatively coupled throughout history and both have stayed relatively flat in the west (with substantial caveats, below). So over the past 30 years, both were driven primarily by developing economies (on the demand side) and new oil discovery/extraction techniques (on the supply side).

> Besides did it double because of increased demand... or because of "modern computing"

My answer is basically "yes". The rest of my argument contains a lot of inter-related claims, so I'm going to number them to make things easier to follow.

But my fundamental claim is that modern computing played a large role on both the supply side and the demand side in the oil industry over the past 30 years.

1. Effect of computing on supply: Computing helped provide the steady flow of cheap oil.

1a. Computing enabled more efficient and more accurate discovery, driving down the price of extracting oil.

1b. Computing enabled new drilling techniques, driving down the price of extracting oil.

1c. Computing improved the efficiency of both old and new drilling techniques, driving down the price of extracting oil.

1d. Computing enabled improvements in refining, driving down the price of refining oil.

1e. Computing enabled improvements in predicting demand, decreasing the risks of destablizing cycles of under-production followed by glut.

2. Effect of computing on demand in developed economies: Computing helped decrease d(demand)/dt for oil in developed economies.

2a. Fuel efficiency in vehicles -- both planes and cars -- was driven in part by modern computing. Multiple mechanisms explain this: improved design (esp in the case of cars), more efficient usage (esp in the case of planes and plants), and efficiencies directly enabled by computing (esp in cars).

3. Effect of computing on demand in developing economies: Computing increased the rate of development.

3a. Indireclty: computing helped keep the price of oil more stable and lower than it would've been without computing (see: 1 and 2). Predictable and relatively lower prices enabled a faster rate development, which itself drove more demand. This cycle would've happened in any case, but by keeping prices lower and more stable, computing increased the rate of economic development inb these countries.

3b. Directly: In some of the most rapidly developing economies, development was directly driven by offshoring of IT/computing and especially of manufacturing related to IT/computing.

To conclude:

1. computing increased the availability and consistency of oil supply, driving down both volatility and prices (note: is is the subtle point. In the case of oil there's a feedback loop between supply and demand, so we have to consider prices for equal levels of demand in the real world vs. the hypothetical no computing world. The claim is that, when we fix the level of demand, prices are significantly lower than they would've been without computing. The claim is not that absolute prices are lower; I think without computing, demand would be significantly lower because large swaths of the world would be significantly less developed),

2. computing decreased the rate of growth in demand for oil in developing economies, allowing growth in supply to enable more rapid development in other economies, and

3. computing increased demand for oil by both directly and indirectly driving development.

None of this is to discount the role of other innovations: the spread of public health practices, vaccinations, the most recent wave of globalization, relative peace, decolonization, and so on all played major roles as well. My point is only that, without computing, the state of oil production and consumption would be substantially different with absolutely massive implications for all of the non-western world, and perhaps for the west as well.


Yeah, that's about right. crazy thing: still not worth it if your only goal is to make money. do not get a phd to max out lifetime earnings...


If viewed through risk-averse lenses, it is an excellent way to raise the earnings floor considerably.


TIL Starbucks coffee has a ton of caffeine. This explains why I'm super jittery every time I get Starbucks even though I drink (non-Starbucks) coffee every day.

So yeah, two freaking massive 20oz pours of highly caffeinated coffee will get you to around 1 gram.

But FWIW the daily recommended dose is 400mg. 1g is a ton of caffeine and way beyond what you would get to by sipping on coffee throughout the day (it's 10+ cups of drip).


I don’t know where you got 400mg/day. Last I checked it’s 160 mg/day.


https://health.gov/dietaryguidelines/2015 mentions that 400mg/day is the upper bound for moderate caffeine consumption but provides no citation.

I cannot find where the number originates, but there is at least one study @ https://www.sciencedirect.com/science/article/pii/S027869151... which explicitly claims "vidence supported consumption of ≤400 mg/day in adults is not associated with overt, adverse effects."


> If yes, proof assistants have to be improved on conciseness, imho

TBF the proof is written for readability, not concision.

But also, this happens in programming as well:

"A search function looking for `x` in a sorted list that starts in the middle of the list, compares the element to `x`, and then recursively searches the front half or back half of the list"

"A valid HTML document with one button that, when clicked, downloads a text file containing the md5sum of the page"

"A facebook clone"


True, but in other languages, developers try to fill the gap with numerous libraries and languages on top of each others, with abstractions on top of abstractions. Proof assistants feel "stuck" at a low level, with high expertise required to use them.


Short term problems need short term solutions and benefit (from) short attention spans.

Long term problems need long term solutions and benefit from longer attention spans.

Your reminder to think about those problems is well taken. But, aside from heeding your reminder, I'm okay with this.


I wish I could upvote this a million times.

Your first point is 100% spot on.

Your second point deserves a caveat: CS does mean something at colleges/universities. they almost all follow the ACM model curriculum:

- CS I -> CS II -> Data Structures

- Discrete Math and Calculus I

- A smattering of available electives from a few fairly uniform buckets (SE,DB,OOP; OS,PL; ML/AI; theory,algorithms).

Contents of courses change (true in Math also! Biz Calc vs. Eng Calc vs. Honors/Advanced Calc). But overall shape stays the same. Can't guarantee outcomes, but can guarantee intent.

Unlike colleges/universities, at the national level, CCs have a serious cruft issue in CS/IT. It's totally impossible to tell what someone might know from the name of the degree program alone. Not even overall shape of curriculum is obvious.

IMO there's a simple solution: CCs nation-wide should pare down everything into a few specific named degrees:

1. Associates in CS Foundations: basically ACM without electives. Designed for transfer to a BA/BS.

2. Associates in Software Development: the non-math portion of ACM + some pragmatic courses (e.g., databases and web dev). Can have various version of this marketed as "AS in SD with Emphasis in ___" where ___ is chosen according to local/regional labor market. E.g., most bootcamps are somewhat like "AS in SD with an Emphasis in Web Development using <stack here>".

3. Associates in Information Technologies: catch-all for the stuff that doesn't even fulfill the non-math, non-elective portion of the ACM model. So "VBA scripting", "Networking aka CISCO certs", "Database Administrator aka Oracle&MS certs", "Applications aka MS Office+Web publishing", etc. degrees currently offered by CCs. Again, various versions can be marketed as "AS in Information Technologies with Emphasis in ___"

These issues might not be recognizable to folks in states like California with amazing CC systems, but in most of America, just figuring out which of the half dozen similar-sounding "IT/CS" associates degrees you want is a serious problem. And very often the answer is "none" because all gazillion of them fit into bucket #3 and you're really looking for bucket 1 or 2.


You don't really need an associates in CS foundations, since all the "CS" part of CS is upper level (except for the most advanced students who start elite Bachelors universities with 1-2 years of university coursework from high school). Just take some basic liberal arts math and intro programming during your AA, then you are ready to apply for CS Backelors program.


> Just take some basic liberal arts math and intro programming during your AA, then you are ready to apply for CS Bachelors program. You don't really need an associates in CS foundations

Well, you can say the same for pretty much every curriculum ever: "we don't need a named degree, just have students figure out what courses they need to take".

Concretely, here's what happens in practice. A student does not know what to take and tells their counselor they are interested in transferring into a four-year CS degree. So student then ends up in:

- standard college prep courses (Algebra, Calc, and liberal artsy stuff), and

- some totally random stuff for the CS part. Might be programming. Might be VBA (won't transfer) or Java (might transfer). Might be a CISCO prep course. Who knows! And unless the student knows and fights, they end up wasting time.

Making a separate track that's explicitly "pre-college CS" avoids this confusion. And after all, part of the value add of a formal education is that someone who knows what you need to know organizes the curriculum for you.

So what you say is true, but there are still strong advantages for both the student and the institution to having a separate, explicit pre-college track.


> (and subsequent implosion)

Can you elaborate?


Probably referring to articles like this: http://hackeducation.com/2017/07/22/bootcamp-bust


> If you can’t define the operational semantics of a system then you can’t rigorously convince yourself or anyone else that it is “secure.”

Well. This means no one can provide any security guarantee for any remotely realistic system because no modern stack has a completely understood&formalized operational semantics. But I definitely know that some systems are much more secure than others! So this seems like a situation where perfect is very much the enemy of "1000x better than it could've been".

Furthermore, formalization is "above and beyond" best practice, so you're unlikely to be accused of negligence if you do "better than best practice but still not formalized".

Finally, the original claim is not true. There are many systems with non-operational semantics that are useful for proving security properties. And sometimes no semantics is needed at all for large swaths of the system. Sandboxing is an excellent example of the latter. For some very reasonable attacker models, you need an operational semantics for the sandbox but don't need an operational semantics for whatever's running inside the sandbox in order to provide fairly strong security guarantees.


There are realistic systems that are operationally secure by the given standard. Nuclear launch systems are the obvious example.

“Best practice” is a euphemism for whatever it is the majority does. Nobody ever got fired for buying IBM. That’s not even reasoning.

Finally, your paragraph about sandboxing is ill considered, but not that wrong. First all programs have operational semantics, the only question is how well understood they are. Second sandboxing is a constraint that is easily formally expressed by way of the logical consequence. I’m surprised you’re taking issue with that since you give a great example thereof.


> There are realistic systems that are operationally secure by the given standard. Nuclear launch systems are the obvious example.

AFAIK there aren't really any examples of production systems that have meaningful, formally verified security properties without gaping holes.

Do you have a good paper about the full stack formalization/verification of a nuclear launch system?

> “Best practice” is a euphemism for whatever it is the majority does. Nobody ever got fired for buying IBM. That’s not even reasoning.

1. Companies are sometimes sued, successfully, for not following "best practices". And on technical merits. Ex, Toyota lawsuits.

2. It most definitely is reasoning! Here's the cartoon derivation: "If I buy IBM I do not get fired. Therefore, I will buy IBM". This is reasoning about the social system, not the technical system. But then, security is both a technical problem and a social problem. $Billions on formal verification can't stop the simplest of phishing attacks.

> First all programs have operational semantics, the only question is how well understood they are.

Well... I guess technically. But that's not typically what people mean when they say something "has a(n operational) semantics".

Typically when people say "X has an operational semantics" they mean "someone has actually tex'd/coq'd/pencil'd the transition rules and maybe proved things about them".

Example: If I implement a programming language without doing any theory and you ask if I have an operational semantics, the only non-confusing answer is "no" or perhaps "the semantics is defined by the implementation of the compiler", which is just a tongue-in-cheek way of saying "no". This doesn't mean that no operational semantics exists, it just means I haven't written it down in the form of transition rules or a coq file or whatever.

If I give my language a denotational semantics and you ask if I have an operational semantics, I'll say "no". But again, that doesn't mean that the operational semantics don't exist. It just means I haven't written them down and proven a correspdonence.

> I’m surprised you’re taking issue with that since you give a great example thereof.

I think you missed the fundamental moral of the sandboxing example: perfect is the enemy of good enough.

Sandboxes allow for security guarantees without formalizing every aspect of the system. In fact, I conjecture that these sorts of "don't try to formalize everything" approaches toward formal security guarantees are really the only ones that scale.

If you try to formalize everything, you'll drown. If you strategically concede defeat and admit that some parts of the system aren't possible to formalize, you can get a lot of strong guarantees. As long as the concessions are strategic. Ex, sandboxing whenever the permissions interface is simple but the implementation is complex and the failure modes permit sandboxing.

Or, to put that observation in a pithy phrase: "perfect is the enemy of good enough."


> Do you have a good paper about the full stack formalization/verification of a nuclear launch system?

Bwahahaha you're funny.

> If you try to formalize everything, you'll drown. If you strategically concede defeat and admit that some parts of the system aren't possible to formalize, you can get a lot of strong guarantees.

You're contradicting yourself again. There's a reason why weakness is a strength in formalisms.


> Bwahahaha you're funny.

No, I'm curious and would use that citation! Even just a paper saying "this has been done" without any details? I have a hard time believing this has been done, and people are allowed to talk about it on online forums, but there isn't a single citation even acknowleding its existence.

> You're contradicting yourself again. There's a reason why weakness is a strength in formalisms.

I'm not sure what this means, but it sounds a lot like what I've been saying all along?


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: