Become a fan of Slashdot on Facebook


Forgot your password?
DEAL: For $25 - Add A Second Phone Number To Your Smartphone for life! Use promo code SLASHDOT25. Also, Slashdot's Facebook page has a chat bot now. Message it for stories and more. Check out the new SourceForge HTML5 internet speed test! ×

Comment rent in Seoul (Score 4, Informative) 246

I am not sure it makes sense to talk about rent in Seoul, Korea. Many people are living on the jeonse system where you give a massive security deposit that the landlord will invest (and give you back when you leave) in exchange for the rent to be very low or even free.

So speaking of "rent" might be very biased. Any Korean around to give feed back on these numbers?

Comment Re:Is this a lot? (Score 1) 160

It is a weird definition of overhead. Overhead is usually defined as the part of administrative expenses as opposed to research expenses. But even like that 72% would not be that big necessarily.

When applying to NSF grant, public state universities have indirect cost, often labeled "overhead", which rate is roughly 50% for public universities (it is negociated per university, but that's roughly what it is.)
What that 50% overhead means is that if $1 goes to the research (paying faculty in summer time when they do additional research, research assistant, travel, hardware(sometimes)), then an additional $.5 goes to indirect cost (keeping the buildings up, the lights up, the admin staff to do accounting, ...)
But remember that public universities do not depend on research funding for many things. Some funds directly come from the state and some funds come from the students.
If you are looking at research-only institution that do not have that stable and high source of income, the overhead is usually much larger. Some places have an overhead of 100%.
I don't know in which category NASA falls, but my guess is that they look more like a national lab, than a public state university.

Comment Re:What the fuck is FMA? (Score 5, Informative) 112

FMA instructions are Fused Multiply Add instruction. Usually their are on SIMD registers and allow you to do "a += b *c". Modern cores can do that on a vector in a single cycle. Actually, they may be able to do more than one FMA on a vector register in a single cycle.

FMA are most commonly used to compute dot product, and are therefore very helpful in linear algebra. (And so they are useful in a ton of data mining algorithms.)

Comment Re:odd thing I've noticed (Score 1) 319

yeah, exactly.

I don't remember EVER talking about the bias of different projections. Never actually looked at a globe in class. I don't remember seeing different projections ever in class. I saw Mercator as a modern map and various medieval maps like T-and-O.

Never thought of the problem of projection before xkcd's famous comic.

(Note, I was educated in France.)

Regarding the difference of projection, I guess that is a nice thing to show in a geography class to show the difficulty of mapping. And of representing things in general.

Comment Re:Taxing consumption over profits would fix this (Score 1) 448

A flat tax on consumption is fine as long as you do not dig too deeply in it.

The main purpose of taxation is to provide a global pool of money to accomplish tax that can not be efficiently performed by each taxed entity.

You do not expect each citizen to build the 10 yard of road in front of his/her house. You pool money in, and the the global entity (government) will pave the road. Flat taxes are a reasonable way to gather fund for activities that touch all part of your country, like education, unemployment funds, defense spending.

The problem with a flat consumption tax is that it does not allow you to tax more activities that causes a particular stress on particular aspect of your society.

For instance, if two modes of transportation have widely different impact on the cost of maintenance of the globally owned infrastructure, you want the cost of that maintenance to be attached to the entity, goods, that are using them. If flying cost to the company $1 per product, and driving cost $2 per product, then the companies will fly. However, if the maintenance cost of flying by the government is $10 per product, while driving would only cost $2 per product, this is a catastrophy because the government bleeds money for each product flown. Because the real cost of flying does not appear in the company balance sheet, the company will not care. So you need the real cost, or at least something that looks like the real cost to appear in the companies balance sheet.

That is why we see things like taxation on gas (to recoup some of the road cost), or taxation on electronics (to recoup the cost of recycling them when they get to the garbage), or airport tax (to recoup the cost monitoring flights, the runway, security, ...).

Comment Re:Their job? (Score 4, Informative) 246

Challenge accepted. In the last 10 years:
-Malala Yousafzai is a nobel peace prize winner and she is from pakistan.
-Aziz Sancar was born and educated in turkey (difficult to tell whether he is of muslim faith or not, but he was probably at least raised in that culture) and is a chemistry nobel prize recipient.
-Maryam Mirzakhani was born and educated (up to bachelor) in Iran and received a Fields medal.

Comment Re:s/drug trials/climate change/g (Score 1) 331

Well, there is a misconception that "it is published therefore it has been thoroughly reviewed". That is not what the reviewing part of the publication process does. Peer reviewing actually start at publication when other lab will try to reproduce the result or incorporate parts of the work into their own work. And at that point you will see whether it is correct or not.

You never have complete experimental protocols today. Because it is not often clear what matters and what doesn't. Also these protocols can be very complex to write. So you provide only a high level view of the experiments. And when others will try to reproduce it, they will use their own setting. If they can reproduce it as is, perfect. If they can't, they'll start trying to narrow down the important parameters. And we will get a better understanding of the phenomenon as a result.

Comment Re:easy to fix without adding more limits (Score 1) 318

No that is good for everyone. Because the american worker (citizen or permanent resident) would only be competing against legal alien, and not the population of the entire world. These work visa could still be restricted per field and have wage lower bounds.

What we would gain with visa portability is visa holders would become indistinguishable from the american worker to the companies. Then doing statistics on them to decide how many new visa to emit the year after and how to change the minimum wages bounds.

Comment Re:Matloff's myths on "indentured servitude" (Score 1) 318

Well, yes and no. H1B are not actually that portable. The only thing that "portability" means is once you have an H1B, any new H1B application for you is no longer capped for quotas. But you still need a new H1B visa.

That means that your new employer still needs to fill a full application for your new H1B visa. That is an expensive process, and many companies do not want to do that or do not have the legal department to follow up with that.

So while yes, H1B are more flexible that people think, they are not as flexible as would be useful to level the playing field.

Comment Re:Google maps navigation option? (Score 1) 359

It may already be in google maps without you knowing. It is just a matter of weighting differently turns depending on whether they are left or right turns. Actually provided the shit ton on data that google has, they may actually be able to estimate the cost of each turn for each light.

Comment Re:Not Possible to Grade on Metrics (Score 1) 229

As usual, there is no perfect metric. But there are first cut indicators.
Things like bug reported per line of code, size of the diff after a code review, simple measures of code readability, how many bugs closed of particular categories, ...
Metrics never tell you the entire story, but that gives a quick idea of who is doing what kind of code is being written. You'll need to look deeper into particular developers to put things in context.

Comment Re:Because they're constantly generating new keys. (Score 1) 203

I have no idea how the system is built in practice. But the lock does not HAVE TO be on the network if you use a public key encryption system.
The lock could read an encrypted stream of bits from the keycard, decypher it with the lock's public key and check the expiration date of the keycard (encoded in the stream of bits) against its internal clock.
Now, that has drawback, in particular it prevents easily baning a keycard after it was emitted (since you would need to be able to tell the lock which is not on the network). But the lock does not have to be network connected.

Slashdot Top Deals

"In matters of principle, stand like a rock; in matters of taste, swim with the current." -- Thomas Jefferson