HNHacker News
TopNewBestAskShowJobs

miloshh

420 karma · joined October 25, 2008

submissionscomments
miloshh··on The one sure way to advance software engineering
A culture of quality has a significant cost. This is justified if lives are at stake, but that's pretty much the only case where it's justified. Grocery stores also don't check each apple 3 times independently if it's not rotten. :)
miloshh··on Economics Is Not A Natural Science
Whether or not these concepts are really proven wrong, I would claim they're not that centeal to economics. The most central idea of economics is that there aren't enough goods for everybody to have what they want. Therefore, you have to distribute the goods somehow.

Some (more or less) fair ways to do that would be to give the goods to those that pay more, or work harder, or wait longer. There are other ways - stealing, cheating, haveing influential friends.

In any case, the scarcity simply won't go away by declaring economics unscientific.

miloshh··on Economics Is Not A Natural Science
Indeed, the main concept of economics is scarcity, not free market or anything like that. Actualy, the basic rules of economics are felt even stronger in non-free markets.

I still remember when I was a kid growing up in communism, there would be days with no potatoes in any store in the city, but there were heaps of rotting oranges that nobody wanted, while the store was not allowed to lower their price. Butchers and gas station operators were among the richest citizens, since those jobs allowed for the easiest cheating on both customers and the state.

While economics is nowhere near an exact science, some people would do better to learn about its very foundations, rather than write lengthy, post-modern-sounding, uninformed criticisms of it.

miloshh··on Tim Sweeney: The end of the GPU roadmap [pdf]
Note that now he's not saying that GPUs are becoming margunal, it's the fixed-function pipeline he dislikes. The "software rendering" he calls for will most likely run on a GPU.
miloshh··on Tim Sweeney: The end of the GPU roadmap [pdf]
I think he mostly hopes for something like REYES for games, which would improve quality somewhat (motion blur and DOF look nice), but that's probably never going to become part of the fixed-function pipeline. That's why he's calling for a "back to software" approach.

WRT Larrabee, I agree with you, and I wonder if the scientific market can recover their development costs.

miloshh··on The Genetically Modified Food You Eat
The Monsanto issue makes me think - might there be a place in the world for "open-source seeds"? :)
miloshh··on The Genetically Modified Food You Eat
I agree with you - the article just implies that GM foods raise questions, but appears almost afraid to take a stand on those questions.
miloshh··on The unstoppable Google
Do you have any concrete examples?
miloshh··on Software Engineering ≠ Computer Science
The only problem with that comparison is that Willie does not help Fats to get his fortune at all. However, software engineering heavily depends on and benefits from computer science.
miloshh··on Ask HN: which functional language to learn?
Definitely Haskell. May I add one more resource for learning it that worked great for me, the official "Gentle Introduction to Haskell":

http://www.haskell.org/tutorial/

For those already proficient with programming that just wnat to pick up Haskell, it's ideal.

Then I would suggest an exercise that involves manipulating abstract syntax, for example parsing and interpreting a little language. ("Build yourself a Scheme..." might be good though I didn't read it myself.) This is where the language really shines, and you don't need any of the recently popular trickery like arrows or GADTs.

miloshh··on Is P=NP an Ill Posed Problem?
OK, I see what you mean, but I would argue that P=NP is a question of a very different nature than CH.

CH is a statement about the existence of an abstract infinite set, so it is not that surprising that its truth depends on your very definition of a set, an elusive object with multiple possible meanings.

On the other hand, P=NP is a statement about boolean formulas, Turing machines and polynomials. None of these are elusive, they are very cleanly defined, trivial concepts. So it's a very well-posed question.

Of course, we could still not be able to prove the (non)existence of a polynomial algorithm for SAT in ZF set theory. In that case, too bad for ZF, but whether P=NP is still a meaningful question.

miloshh··on Is P=NP an Ill Posed Problem?
Well, even if a given logical theory cannot decide P=NP, it is still a well-posed problem. Either there is a polynomial-time algorithm for SAT or there isn't - no middle-ground here.

But I agree, the blog isn't about this issue at all.

miloshh··on Google mocks Bing and the stuff behind it
Of course, Google's infrastructure is great, whether MapReduce or GFS.

However, if Bing gets to a point where their search results are clearly better, then Google will be in trouble, and no amount of screaming "we have a better infrastructure!" will help. Users want the best results, and that's why Google originally won against other search engines.

I don't know if that's ever going to happen. Probably not, but I would not bet on it.

miloshh··on The evolutionary origin of depression
Interesting hypothesis. However, isn't depression sometimes getting in the way of pursuing perfectly attainable goals? For example, finding more friends or losing a bit of weight are definitely not overly ambitious goals, but depressed people often despair that they will never be able to reach them.
miloshh··on What the Heck Is Cloud Computing?
Agreed, I also don't see much advantage in word processing or spreadsheets on the cloud.

The only thing I can think of is some light-weight collaboration, but more involved collaboration is better done using version control anyway. This is outweighed by the disadvantages of having to deal with clunky Javascript/Flash apps and slow internet connsections.

There are many cool things to be done on the cloud, but Office-style apps just don't seem to be a good example.

miloshh··on Admitting that Functional Programming Can Be Awkward
Yes, and it is a weak point.

My point, instead, was that FP does not break down when things get ugly; on the contrary, it gives much stronger guarantees on correctness. Of course, it takes effort to express the problem in the more restrictive functional paradigm.

miloshh··on Admitting that Functional Programming Can Be Awkward
The main benefit, as I mentioned above, is that if a pure function works once, it works all the time. This is captured by the saying "in Haskell, if it compiles, it works", which is surprisingly often true.

If you have a function that moves a player in a game such as movePlayer :: PlayerState -> PlayerState, you know that it only changes the state of this player, and nothing else. In an OO language you would do something like player.doUpdate(), which has absolutely no guarantees.

Plus there are many facilities available to deal with threading and hiding state.

miloshh··on Admitting that Functional Programming Can Be Awkward
Oh, I agree with the strengths and weaknesses, and obviously I don't use FP for all my programming.

But I somehow cannot see any functional programming hype. Where are the people heralding it as solving all problems? Even in a place like HN with a highly non-average community of programmers, the FP links are quite rare and very intro-level. I would say FP is a high-hanging fruit that's only starting to find a few niches in real-world programming.

So I son't see anything close to an FP hype, nowhere near to the Java hype or Ruby hype.

miloshh··on Admitting that Functional Programming Can Be Awkward
languages that let you do short parlor tricks

Quite the opposite - functional programming is becoming very appealing as a practical alternative for building large, multi-threaded, correct systems.

True, it might be a bit more tedious to write a program that type-checks, but as many (including me) have been repeatedly amazed, it often works correctly right away, with minimal debugging.

But it also may be that almost no one has been thinking about problems like this

Mutable state is a fundamental issue that every functional programmer deals with every day. Indeed, it might be sometimes frustrating to pass around state that in C would be directly modifiable, but the benefits are huge.

Pure code has been called a "scarce resource", and that's exactly right - once it works, it will always work, no matter how huge is the system you build on top of it.

miloshh··on Why "next-gen games" went gray, brown, and grey.
Oh yes, absolutely, but there's no reason why real-time would imply brown/gray color hues...
miloshh··on Why "next-gen games" went gray, brown, and grey.
Downvoting because of disagreement is really getting out of control at HN.
miloshh··on Why "next-gen games" went gray, brown, and grey.
Exactly,movie studios use pretty much the same lighting tricks as games - direct illumination with some ambient occlusion and color bleeding hacks here and there. Yet they do not look brown/gray. I think it might be because they have armies of artists (they call them "technical directors") that tweak the materials and lighting for thousands of man-hours until it looks good. Game companies probably invest much less in this kind of tweaking.
miloshh··on Why "next-gen games" went gray, brown, and grey.
So he says that games are becoming more realistic in most aspects, but not in the lighting. That is probably true, but there is no reason why that should cause the game to become grey and washed-out.

Most CG movies and effects (from Pixar, Dreamworks, ILM, ...) also do not have anything close to photorealistic lighting (i.e. correct global illumination), but they do not look grey.

One common mistake that leads to washed-out images is to apply gamma correction (which is the right thing to do) but not apply inverse gamma correction to your textures. That way your surfaces are essentially gamma-corrected twice, which is clearly wrong. I don't know if they're really making such a simple mistake, but who knows...

miloshh··on [dead]
Plus the first source they cite is the "Heartland Institute". :)
miloshh··on [dead]
1. Temperatures are clearly rising.

2. CO2 concentrations are clearly rising.

3. Well-understood physics suggests that rising CO2 should lead to rising temperatures.

Does this make it 100% certain that we're causing global warming? Maybe not, but if you want to disprove the link, you're going to need better arguments than "41% of laypeople think global warming is exaggerated".

miloshh··on Unmanned sub touches deepest part of world's ocean
Maybe by simply stretching the wire by the force of water movement (like pulling a guitar string up really hard).
miloshh··on Ask HN: Good C++ Numerical Libraries
The Intel Math Kernel Library is decent, and very fast. It has disadvantages - not optimized for AMD processors, not free (but the student version is cheap, and the free trial does not really expire).
miloshh··on Software jobs, PhDs and over-qualification
I agree with both of you. I would advise Ph.D. not to take entry-level coding jobs, and employers not to admit Ph.D.s to such jobs. It will be a misfit, and everyone will suffer.

I was just a bit upset at the dichotomy of "industry = responsibility, having to deliver" and "academia = lack of such responsibility". I think we know better now what each of us meant.

miloshh··on Software jobs, PhDs and over-qualification
We are not disagreeing. Here you're referring to your position of leadership and responsibility that you have risen to after 30 years, but I was talking about code monkeys.

"Do x, y, and z by the first of next quarter or you're out of business." Sure, but new developer hires in large companies are absolutely not in a position to decide where the business is going, and won't be for quite some time. The article and my comment were about the entry-level developer positions.

"And for everyone of those, there is someone else changing the world." Absolutely, and I greatly admire such people. But you won't change the world in an entry-level developer job. I tried such jobs, they were very easy and I suffered a lot the lack of opportunity to change the world.

miloshh··on Software jobs, PhDs and over-qualification
As someone just finishing Ph.D. that also had industry jobs, I can say most of what you say seems wrong. There indeed are plain old developer positions, consisting of e.g. making some frontend talk to some backend - most real-world corporate programming is like that.

"Deliver something every day to someone who is depending on you" - this is exactly what you have to do in a Ph.D. program all the time - you have coauthors on papers that have to be shipped, and yours and their future depends on it.

In the corporate world, you also have to deliver, but at least what you're doing is well known to be possible. In academia you have a constant feeling that you might be against an impossible problem, and the results will never come together.

Plus in the corporate world, you can work 9am-7pm at most, rather than 11am-2pm as in grad school. And the pay is 4x higher. Now what sounds more like a vacation?

(Of course, here I was referring to a PhD in Computer Science. I don't know what's Strategic Planning, so I can't judge there.)

← PreviousPage 4 of 6Next →