HNHacker News
TopNewBestAskShowJobs

crabapple

172 karma · joined November 3, 2008

submissionscomments
crabapple··on Nick Denton: Pay-for-page-views was too successful
i was never enamored with denton's editorial tenure at valleywag (he seemed capricious and arbitrary about what you could shit-kick and what was a sacred cow)...but as a business owner he is spot-on. he isn't waiting for the recession to force him into consolidation, he's doing it proactively, on his time. the newspapers could be doing this, but instead are going to wait for the markets to back them into a corner.

in any case the bell was tolling for valleywag...there just isn't that much dirt in the bay area to dish on...i never knew how people managed even the contrived level of interest that valleywag generated

crabapple··on Sun announcing up to 6000 (18%) layoffs
yup. sun is a bespoke tech firm in an era of relentless commoditization.

my guess is that they only cling to life on support contracts from people who are still solaris/sparc shops (which i imagine they are still quite a few of)

crabapple··on Sun announcing up to 6000 (18%) layoffs
who else is astounded that sun employs more than 30k people???
crabapple··on What’s so hard about monetizing Facebook?
early adopter under-30s have become impervious to web advertising. the clickthrough rates are just dismal. yahoo and aol cater to the trailing edge...you know, the people dumb enough to click a banner ad. thats why they make money with content (non-search) ads.

fb should consider a nominal monthly fee for advanced features (a la flickr). my guess is that the service has become sticky enough to turn the screws ever so slightly on users. barring a subscription, i don't see them ever making a lot of money....unless they just go to a cpm model and advertisers just give up on clickthrough

crabapple··on 4 Languages you should learn in 2009
haskell is the only programming language that made me reconsider most of my assumptions. its not a tacked-on, hacked-on functional layer on something else, its mind-bending at a fundamental level. because haskell makes no comprimises, it also is showing some insane performance on multicore. its no mistake that the google mapreduce tutorials use haskell in some explanatory examples...

ten years from now, you will be dealing with a multicore world, and you will be using something that is conceptually tied to haskell. i see this as inevitable.

factor i have looked at again and again, worked through some trivial problems...not sure i will ever "get" it. i would brute-force understanding it were it to take off.

crabapple··on 4 Languages you should learn in 2009
i would recommend clojure if you want to learn "a lisp"...along the positives you cite, it has a growing community, very active development, hype (in a good way), and deployability.
crabapple··on The Worst Is Ahead (Roubini)
yes. i wish most sites would go on a markup diet.
crabapple··on The Worst Is Ahead (Roubini)
History promises inflation, not deflation.

BINGO

all that money being printed now with reckless abandon, and the tens of trillions more we will print as the boomers retire...its gotta chase something...it will chase commodities through the roof. oil will go to $500. your weekly grocery bill will be over $1000. now's a great time to stockpile the black stuff before it goes crazy. it will be amusing watching futures traders get gutted as they open up shorts on oil futures presuming it will keep falling...

crabapple··on The Worst Is Ahead (Roubini)
indeed, look at the great depression, look at japan in the 90s. in each of these secular bear markets, markets rebounded for gains as high as 100% from the previous low...yet still the ultimate direction of the market was downward. DJIA will rebound at some point to around 12k, then back down to 5k. it will be most amusing watching the last hopeful traders plow back into the market only to fall into an obvious bear trap.

ultimately we will get hyperinflation as the govt prints literally trillions to monetize the various debts of the govt. the strong dollar now is just a head fake as the yen carry trade and other various forex arbitrages unwind.

all real estate will lose 50%. everything. that is what happened in japan in the 90s...even desirable tokyo condos lost 50%. 50% will be the minimum downside for real estate.

actually its worth noting that when japan went down the toilet in the 90s, they were the world's biggest saver and lender. we are the world's biggest borrower. so don't expect anything as gentle as japan's ten-year recession.

expect an end to all of this around 2020, and this will only end when people see taking on debt, any debt, as about as attractive as sharing needles with a junkie. when this is over its impossible to say what the economy will look like, but you can be sure people will be dealing in tangibles...cash or barter.

the best thing you can do now is -preserve capital-. you are going to need it to survive. forget the charlatans telling you to "double down" because they have some anecdotes of companies exploding out of temporary down times....none of your luminaries have lived through anything like this, and indeed many of the superstars of the roaring twenties were literally gutted by the market turmoil of the 30s...they just refused to believe the downside would last very long, they thought it was all just a buying opportunity...little did they know how far things would fall.

crabapple··on "Sequential programming is dead. So stop teaching it"
no, haskell. clojure in the end runs on the jvm. the ghc compiler has parallelism baked into it in a fundamental way.

don't get me wrong, i like clojure, but it won't be better for multicore until the jvm is better for multicore

crabapple··on Ubuntu's #1 Bug: Microsoft has a majority market share
uh, this gets posted once a month. we get it
crabapple··on "Sequential programming is dead. So stop teaching it"
you. you aren't going to be given the choice, these are coming to a laptop near you...like on your lap
crabapple··on "Sequential programming is dead. So stop teaching it"
who is trying to make you less dumb
crabapple··on "Sequential programming is dead. So stop teaching it"
you need to stop fighting the future
crabapple··on "Sequential programming is dead. So stop teaching it"
NO. why don't you READ before you reply

a program compiled for a multicore CPU will RUN. the question is how OPTIMALLY does it run. a program with no potential for parallelism will not get any parallelism. it will run, but run slow compared to programs designed for parallelism.

programs written to exploit parallelism will be programs that bring new approaches to data and state. functional languages provide this today, which is why lots of people think they will be the way forward for multicore.

honestly i think you are just bordering on being a troll. why don't you do some reading on this topic before writing more uninformed replies

crabapple··on "Sequential programming is dead. So stop teaching it"
what is in the cpus of those machines will be in the cpus of all machines. jesus, how much more legit can you get than intel telling you this is coming?
crabapple··on "Sequential programming is dead. So stop teaching it"
search "shared nothing concurrency"...read sutter's "the end of the free lunch"...
crabapple··on "Sequential programming is dead. So stop teaching it"
you should stop posting until you take classes in operating systems and compilers. really
crabapple··on "Sequential programming is dead. So stop teaching it"
uh, yeah, i know that. i thought the "coolzzz" would relate my sarcasm
crabapple··on "Sequential programming is dead. So stop teaching it"
I don't write OS kernels, I write rails application. I don't give a rat's ass about the OS kernel

then stop spouting off uninformed comments about how processes are scheduled

The Rails applications themselves don't need to be altered to run in a multi-core environment.

nor do any other program compiled for that architecture. its the OS that schedules processes, not your userland program. the point is, some programs can be written in a way that makes it easier for the OS to exploit multicore. since ruby is not a functional language, my guess is that it would tend to not help the kernel exploit these resources. but obviously in the worst case, a process can run inside one core and never get the advantages of the rest of the chip architecture. this is about exploiting multicore

Got it?

yes, i get that you know very little about how computers function

crabapple··on "Sequential programming is dead. So stop teaching it"
Sorry, but I don't buy that

then you clearly aren't buying new high-end servers for data crunching either, because these are already multicore

crabapple··on "Sequential programming is dead. So stop teaching it"
no, java programmers pick up haskell is more likely.
crabapple··on "Sequential programming is dead. So stop teaching it"
That's ridiculous. Programming is not all about multi-core performance.

it will be. you're getting beaten over the head by chip designers telling you that your future cpu is going to consist of a (possibly large) array of processing cores with a high-capacity bus connecting them. they are telling you this is the only way they can give you higher performance. you had better start believing them because these systems are starting to get delivered now.

A great example of that is Rails, when set up with mongrel processes, each of which can run on its own core if necessary

?????? so mongrel comes with its own OS kernel that has better support for multicore than linux and freebsd? wow!! coolzzz!

crabapple··on "Sequential programming is dead. So stop teaching it"
I'm not sure how you can talk about concurrency without talking about locks

that is because you have slept through the entire revolution of share-nothing concurrency

crabapple··on A Scary Line Has Been Crossed For VCs
What's supposed to replace VC, exactly?

its possible that VCs can't assemble the level of funding needed to attack the energy market (for example)...these investments will have to scale out to the hundreds of billions with very long time horizons for returns. most likely this stuff will have to get federal funding, and much of it already does from nsf, doe, darpa, etc

crabapple··on A Scary Line Has Been Crossed For VCs
boo-fucking-hoo. a down market hurts everyone. well maybe they should use this lull to re-educate themselves for the future. guess what folks, sandhill road will be as dead as wall street unless you try expanding past websites. let yc fund this low-iq crap. time to bone up on all that chemistry you skipped in high school, you'll need it if you are going to evaluate greentech, nuclear, etc
crabapple··on Ask HN: How do I handle a million hits a minute?
lots of hardware

there are a few best practices to scaling, you can find those archived at places like highscalability etc...but in the end you just need hardware. hardware buys you fault tolerance, test environments, sharding, blah blah blah. if you are looking to meet this traffic on your own box or on a vm-hoster, forget it

crabapple··on Apple may use Carbon Fiber in Next Macbook
wow apple is desperate. they are using unibody aluminum chasis to convince people to pay $2k for laptop internals they could get from dell for $600. case mods must be where its at since the macbook pro is not meaningfully different under the hood than the previous macbooks.

carbon fiber?? why not try to see if you can freeze-frame star trek iv at the point where scotty has his forumla for transparent aluminum on the screen. that should catch a cool $5k per unit at least.

crabapple··on Loopt Hires Allen & Co. For Financing Or Sale
the valuation they cite is a joke. $500 million? haha i'm sorry, 2006 is gone and won't be returning. loopt would be better off claiming they are a bank and going to the TARP...probably make more with fewer questions asked!
crabapple··on Science's Alternative to an Intelligent Creator: the Multiverse Theory
indeed. the article seems to be asking a false question - "just how did we get so lucky!?!?!" which seems ridiculous, and as you point out, presumes that a universe that produces us is more optimal than one that doesn't.
← PreviousPage 3 of 4Next →