If only Yahoo! would bring back AltaVista's Prisma technology.
Actually, that never went away. Take for example the results page for this search on dog. Above the sponsored results (how cool that they actually think it's more important) and directly below the words "Search results", you'll find the Prisma stuff: "Also try: dog breeds, dog names, dog the bounty hunter, dog training More...".
I think Moore's missing the point. The reason a company hits a wall is that it stops being innovative
I think Moore actually has a better understanding of the problem. Do you think company execs sit around a table and say to each other, "let's stop being innovative now"? No, it's a situation that happens, and it tends to be inevitable. You're faced with the "innovator's dilemma, and sooner or later it'll get you. Google is just too young to have been hit with this. They're doing everything they can to dodge the bullet, but it really is just a matter of time.
I don't recall Yahoo! making anything innovative recently, but correct me if I'm wrong.
You are wrong. Google has way better PR than Yahoo, so pretty much everything they do gets spun as innovative (regardless of merit), while Yahoo's stuff tends to get ignored. Still, as an example, a lot of search industry experts have been describing Yahoo's Mindset demo as the most innovative thing they've seen in search in years.
Acutally, most places I have interviewed of late have at least 5 interviews before you get the job. People think this is unique to Google as part of the mystique, but frankly it's true of Yahoo and most other tech companies that are both good and large. Typically you spend a day interviewing with people, and if they like you, you have another day doing interviews at headquarters with the "big shots".
If you look at the interview process as just a way for them to evaluate you, then I can see why you find it annoying, but I see it also as an opportunity for me to evaluate them. The more I get to see of them before I make a decision, the better.
Sigh. You know that part of the deal with YME is that it is integrated with messenger, right? You can share music with your IM buddies and you guys can see what each other is listening too.
Just because you perceive two independent applications doesn't mean that they are two independent applications, or that someone else might perceive them as two different applications.
I mean, both Google and Yahoo have toolbars that also include pop-up blocking. What does pop-up blocking have to do with the toolbars? Well, I guess they both involve a browser, but beyond that, nothing. Why is it integrated in? Because they thought it'd be a good feature that users would want. Maybe some people just want the toolbar and others just want the pop-up blocking, but I notice you aren't saying anything about that.
I see you haven't played around with GigE. DMA doesn't solve the problem with networking. Sure, the hardware can handle the processing of the ethernet frame, but unless the network card does layer 3 (like the cards in this article), each frame still needs to go to the CPU so it can handle the network protocol (say TCP/IP). Sure, you can try delaying the processing until you have a few ethernet frames together, but of course that's already part of the trick that's going on with a fast ethernet card, and the frames are coming 10x faster with GigE. You delay too much, and it starts to actually slow down the networking.
I wonder what has changed? I have never known the CPU to get dragged down by network traffic, but maybe in the network server markets it is different
The thing that has changed is that the frequency that frames arrive at has gone up. Unless you can use jumbo frames (and even then, if the payloads are small), GigE is delivering the same sized frames as fast ethernet, just 10x faster. This tends to create a hell of a lot more interrupts for the processor to handle (a condition made worse by the deeper pipelines in processors like the P4). If you can offload the processing of the frames a bit, just enough to give a processor a chance to get something done, you could dramatically improve performance.
That being said, changes to the protocol (such as jumbo frames) can also have a positive effect in a lot of circumstances, and have the advantage of being cheaper to implement.
It was funny, because I had just read an article talking about the rising demand for folks with computer science degrees. Then I looked more closely... This Gartner study measures 2001 (you know, the.com bust was just starting to cause masses of people to lose their jobs) against 2004 (industry just starting to recover). No surprise there has been a decline in jobs. Sigh.
Note that the current design already has a POWER processor as it's core. The problem is, a lot of the chip is dedicated to the fancy vector processing units. There's no way to make up for that. The entire win with the Cell processor is that it's got so much of it's transistor budget dedicated to something that CPUs do relatively poorly: vector processing. The rest of the design is very standard. So the best you can do is match what you're doing with other POWER CPU's in terms of performance, but you'll be significantly more expensive because of all those vector coprocessors.
Presumably if one wants an ad in the information-only (!) section one has to pay an extreme premium.
You're thinking old world style advertising. Paid search is all about being able to target customers who actually want to buy something that you have to sell. I suspect the reality would be just the opposite of what you say. If I'm do an "research" search, what are the odds that I'm going to go buy something when I click on a link? Pretty rotten right? So, the advertiser doesn't get much value from it. I suspect the premium would be for people who are really looking to buy something.
The slider doesn't really do blending. It controls how much priority the "shopping" or the "research" gets in the sorting over how the search engine would normally sort the results. If you have it all to the research side, it shows the most "researchy" page, regardless of how relevant the page is to the search.
I wonder how this comapares to the ages at which they get married? There is that theory that once you get married it's hard to have the singlemindedness that leads to great achievements. Of course, it's hard to seperate cause and effect, because a lot of scientists wait until after they've had some career success before getting married.
Um, if it doesn't make code run faster, what's the point of including it?
Sigh.. I wasn't saying that it doesn't make code run faster. I was saying that it doesn't necessasrily make code run faster. Auto-vectorization is only a win in certain circumstances. There are a whole host of optimizations that only apply in specific circumstances and/or only improve performance in certain circumstances and slow things down in others. If there weren't trade-offs with optimizations, compilers would just have "-O" and wouldn't bother with tons of other optimization flags.
Actually, even if you completely turn off auto-vectorization, or compile on IA64 where such concepts are effectively meaningless, the Intel compiler still performs very well.
I wish people would stop talking about autovectorization like it's an instant "code runs faster" thing. Yes, in some cases it makes a huge difference. In many others, it doesn't even apply.
Yeah, but when Microsoft does it, it must by definition be evil. If Google does it, it must by definition be good. You can't fool us with your rationality. This is slashdot!
Otherwise every politician within spitting distance would be in jail
Well, if anything they tend to commit slander rather than libel. That being said, politics and politicians don't get the kind of protections from slander and libel that us private citizens enjoy.
It's a little hypocritical for Larry to complain about other people riding on his coat tails when Bitkeeper is, like most successful products, a really good implementation of a bunch of ideas that were invented by a lot of other people over a lot of time.
Might I humbly suggest you RTFA again? The "riding on our coattails" statement was specifically with regard to reverse engineering, and he specifically said he had no problem with people building a competitive product.
Now, one might not agree with Larry about reverse engineering, but it's libelous to take his statements out of context and claim they're hypocritical.
Hehe. You might want to reevaluate them then, as the underlying engine for both has been completely revamped in that time with a significant increase in quality. Same with MSN (although most people would agree that MSN still has a ways to go).
This is an interesting demonstration of just how much subjectivity effects perceptions about search engines.
Alltheweb.com produces the same results as Yahoo search (basically ever since Yahoo merged with Overture). Yet you describe them as being distinct and with different qualities. You even will search on one after searching with the other.
...and yet strangely you'll find even companies selling solar power equipment will put the cost at least eight cents per watt. Maybe you need to think things through?
First, there are maintence costs, but those are arguably quite minimal. The bigger problem is that the solar cells don't actually last indefinitely. It appears that nobody can judge very well how long the cells can last (although most estimates are around 30 years), that's no reason to assume they'll last forever.
Heck, there was a time (like 20 years ago or something) where the energy used to make a solar cell was actually greater than all the energy said cell would produce over it's lifetime.;-)
Slashdot Mirror
Actually, part of the problem is that Google is ignoring my robots.txt
Okay, if this is the case, then it seems pretty rediculous to offer this as "proof" of much of anything.
If only Yahoo! would bring back AltaVista's Prisma technology.
Actually, that never went away. Take for example the results page for this search on dog. Above the sponsored results (how cool that they actually think it's more important) and directly below the words "Search results", you'll find the Prisma stuff: "Also try: dog breeds, dog names, dog the bounty hunter, dog training More...".
I think Moore's missing the point. The reason a company hits a wall is that it stops being innovative
I think Moore actually has a better understanding of the problem. Do you think company execs sit around a table and say to each other, "let's stop being innovative now"? No, it's a situation that happens, and it tends to be inevitable. You're faced with the "innovator's dilemma, and sooner or later it'll get you. Google is just too young to have been hit with this. They're doing everything they can to dodge the bullet, but it really is just a matter of time.
I don't recall Yahoo! making anything innovative recently, but correct me if I'm wrong.
You are wrong. Google has way better PR than Yahoo, so pretty much everything they do gets spun as innovative (regardless of merit), while Yahoo's stuff tends to get ignored. Still, as an example, a lot of search industry experts have been describing Yahoo's Mindset demo as the most innovative thing they've seen in search in years.
To give an idea of what I'm talking about, here's yesterday's search engine stats from my webserver:
;-)
Hehe. Maybe all the other search engines are just better at identifying relevant sites.
Acutally, most places I have interviewed of late have at least 5 interviews before you get the job. People think this is unique to Google as part of the mystique, but frankly it's true of Yahoo and most other tech companies that are both good and large. Typically you spend a day interviewing with people, and if they like you, you have another day doing interviews at headquarters with the "big shots".
If you look at the interview process as just a way for them to evaluate you, then I can see why you find it annoying, but I see it also as an opportunity for me to evaluate them. The more I get to see of them before I make a decision, the better.
Sigh. You know that part of the deal with YME is that it is integrated with messenger, right? You can share music with your IM buddies and you guys can see what each other is listening too.
Just because you perceive two independent applications doesn't mean that they are two independent applications, or that someone else might perceive them as two different applications.
I mean, both Google and Yahoo have toolbars that also include pop-up blocking. What does pop-up blocking have to do with the toolbars? Well, I guess they both involve a browser, but beyond that, nothing. Why is it integrated in? Because they thought it'd be a good feature that users would want. Maybe some people just want the toolbar and others just want the pop-up blocking, but I notice you aren't saying anything about that.
I see you haven't played around with GigE. DMA doesn't solve the problem with networking. Sure, the hardware can handle the processing of the ethernet frame, but unless the network card does layer 3 (like the cards in this article), each frame still needs to go to the CPU so it can handle the network protocol (say TCP/IP). Sure, you can try delaying the processing until you have a few ethernet frames together, but of course that's already part of the trick that's going on with a fast ethernet card, and the frames are coming 10x faster with GigE. You delay too much, and it starts to actually slow down the networking.
I wonder what has changed? I have never known the CPU to get dragged down by network traffic, but maybe in the network server markets it is different
The thing that has changed is that the frequency that frames arrive at has gone up. Unless you can use jumbo frames (and even then, if the payloads are small), GigE is delivering the same sized frames as fast ethernet, just 10x faster. This tends to create a hell of a lot more interrupts for the processor to handle (a condition made worse by the deeper pipelines in processors like the P4). If you can offload the processing of the frames a bit, just enough to give a processor a chance to get something done, you could dramatically improve performance.
That being said, changes to the protocol (such as jumbo frames) can also have a positive effect in a lot of circumstances, and have the advantage of being cheaper to implement.
It was funny, because I had just read an article talking about the rising demand for folks with computer science degrees. Then I looked more closely... This Gartner study measures 2001 (you know, the .com bust was just starting to cause masses of people to lose their jobs) against 2004 (industry just starting to recover). No surprise there has been a decline in jobs. Sigh.
Note that the current design already has a POWER processor as it's core. The problem is, a lot of the chip is dedicated to the fancy vector processing units. There's no way to make up for that. The entire win with the Cell processor is that it's got so much of it's transistor budget dedicated to something that CPUs do relatively poorly: vector processing. The rest of the design is very standard. So the best you can do is match what you're doing with other POWER CPU's in terms of performance, but you'll be significantly more expensive because of all those vector coprocessors.
Presumably if one wants an ad in the information-only (!) section one has to pay an extreme premium.
You're thinking old world style advertising. Paid search is all about being able to target customers who actually want to buy something that you have to sell. I suspect the reality would be just the opposite of what you say. If I'm do an "research" search, what are the odds that I'm going to go buy something when I click on a link? Pretty rotten right? So, the advertiser doesn't get much value from it. I suspect the premium would be for people who are really looking to buy something.
The slider doesn't really do blending. It controls how much priority the "shopping" or the "research" gets in the sorting over how the search engine would normally sort the results. If you have it all to the research side, it shows the most "researchy" page, regardless of how relevant the page is to the search.
The tool tends to work best when you don't put the slider all to one side or the other. This is presumably why they have slider.
I wonder how this comapares to the ages at which they get married? There is that theory that once you get married it's hard to have the singlemindedness that leads to great achievements. Of course, it's hard to seperate cause and effect, because a lot of scientists wait until after they've had some career success before getting married.
Copies are available on O'Gara's sites, as was mentioned in the article. For example:
http://www.clientservernews.com/
Um, if it doesn't make code run faster, what's the point of including it?
Sigh.. I wasn't saying that it doesn't make code run faster. I was saying that it doesn't necessasrily make code run faster. Auto-vectorization is only a win in certain circumstances. There are a whole host of optimizations that only apply in specific circumstances and/or only improve performance in certain circumstances and slow things down in others. If there weren't trade-offs with optimizations, compilers would just have "-O" and wouldn't bother with tons of other optimization flags.
Actually, even if you completely turn off auto-vectorization, or compile on IA64 where such concepts are effectively meaningless, the Intel compiler still performs very well.
I wish people would stop talking about autovectorization like it's an instant "code runs faster" thing. Yes, in some cases it makes a huge difference. In many others, it doesn't even apply.
Yeah, but when Microsoft does it, it must by definition be evil. If Google does it, it must by definition be good. You can't fool us with your rationality. This is slashdot!
Otherwise every politician within spitting distance would be in jail
Well, if anything they tend to commit slander rather than libel. That being said, politics and politicians don't get the kind of protections from slander and libel that us private citizens enjoy.
It's a little hypocritical for Larry to complain about other people riding on his coat tails when Bitkeeper is, like most successful products, a really good implementation of a bunch of ideas that were invented by a lot of other people over a lot of time.
Might I humbly suggest you RTFA again? The "riding on our coattails" statement was specifically with regard to reverse engineering, and he specifically said he had no problem with people building a competitive product.
Now, one might not agree with Larry about reverse engineering, but it's libelous to take his statements out of context and claim they're hypocritical.
Hehe. You might want to reevaluate them then, as the underlying engine for both has been completely revamped in that time with a significant increase in quality. Same with MSN (although most people would agree that MSN still has a ways to go).
This is an interesting demonstration of just how much subjectivity effects perceptions about search engines.
Alltheweb.com produces the same results as Yahoo search (basically ever since Yahoo merged with Overture). Yet you describe them as being distinct and with different qualities. You even will search on one after searching with the other.
A very good point. Oh wait...
/ /
http://www.alltheweb.com/
http://www.teoma.com
http://search.yahoo.com/
http://search.msn.com
http://www.altavista.com/
Yeah. You might want to get out more.
Actually, that was a typeoh on my part.
s/watt/kilowatt hour/
...and yet strangely you'll find even companies selling solar power equipment will put the cost at least eight cents per watt. Maybe you need to think things through?
;-)
First, there are maintence costs, but those are arguably quite minimal. The bigger problem is that the solar cells don't actually last indefinitely. It appears that nobody can judge very well how long the cells can last (although most estimates are around 30 years), that's no reason to assume they'll last forever.
Heck, there was a time (like 20 years ago or something) where the energy used to make a solar cell was actually greater than all the energy said cell would produce over it's lifetime.