With click-to-focus you always have to be careful not to click on a button or if you click on an editor, it automatically moves your cursor from where it should be.
oh, good point.
I've bound my mouse's fourth button to be "raise, focus, and allow me to move window". Took a couple of lines of LISP in sawfish, but it's terribly useful. I don't need to worry about passing through clicks (or about hunting for and hitting the titlebar when I want to move a window).
The welfare states of Europe are great if you just want to slide by and work part time while having everyone else pay for your existence when you get too lazy to be productive. The problem is that this means there is no innovation and no dynamic growth, both of which are the keys to the American economy. Europe is a great place to be for paper MCSE's or people with enough political pull to get a cushy union job, it is not a great place for innovating and pushing the boundaries of new technology.
I dunno. How many great advances are made by burned-out people stuck at work to the breaking point? I usually get most of my good ideas when I'm showering, where I get a chance to unwind and think about the material I was working on. I've tried writing code for more than 60 hours a week. I can do about one week of that, then I just hit "code burnout", where it's like whacking your head with a rubber bat. Your LOC per hour goes waaay down, and you start staring at the screen without seeing it. It just doesn't *work*.
I like focus-follows mouse, magic desktop borders and transparent thingies.
I like most fvwm-ish things (and zero resistance edge flipping!), but focus-follows-mouse always confused me.
It seems like this focusing system always tends to result in your mouse cursor winding up covering up what you're trying to work with. Usually, I'd prefer to have my mouse cursor elsewhere.
BTW, if anyone from Sun is reading this, that HCI work done was much appreciated. Formal testing was something that had been severely lacking and that helps out the salabilty of the environment well.
While this 3d business seems a bit...well...researchy-without-immediate-benefitis h, the HCI feedback was a big deal.
Marvellous. Just as users of current operating systems have ben doing for years anyway.
Linux is a current operating system, but the mainstream Linux GUI environment does not provide this (watch for rapid changes now that freedesktop.org is coming to the fore).
Hmm. Potentially interesting as a way to pick between open windows, but doesn't Expose perform this task in a better manner?
It's eye candy, little more. Frankly, I view Expose as eye candy -- multiple viewports are a far more powerful mechanism, and one that I'm sad that Apple chose not to support.
Except that in the real world I can never find the bloody CDs, because I can't remember where I've put them. I can navigate a media player interface far faster than I can hunt for CDs, and I can use more search criteria too (album, artist etc.)
Agreed. People were raving about real-world metaphors in research labs a couple of years ago, but nothing worked out all that well. It seems to me that all the features mentioned here could be provided better through a 2d implementation.
Uh...Germany didn't bother condemning them. They just outright banned them.
Is Vietnam condemning Battlefield Vietnam?
I haven't the faintest idea what the political situation is like in Vietnam, though I doubt that video gaming is a prominent issue there.
On a side note, imagine the fun we'd have playing a game based on the US/Canada war of 1812...
I think that sounds like fun.
On the other hand, I also thought that it would be fun to play not only as the police but as the sniper in a UT clocktower scenerio in one of the hostage rescue/counterterrorism games.
Nurses are an example of a profession even less "happy" than IT. While nurses aren't worried about their jobs being outsourced, interestingly enough they tend to be overworked (usually 10-12 hour rotating shifts throughout the week) and underpaid, especially compared to their colleagues. But, according to the articles, they should be happy, as they are very hands on. Guess that hypothesis just got shot down.
New studies have indicated that working people as far as you can stretch them *makes them unhappy*!
I could never figure out why the employment situation in the US is so screwed up.
We have this kind of go-go-GO-OR-YOU-FAIL-DAMMIT-GO! mentality that keeps being pushed. I was talking to some folks about the kind of hours that people starting off in financial services or the legal world can expect to work -- the hours are *stupid*. Sure, the jobs pay well, but what do you do with the money? Buy a bigger TV or a more expensive car, neither of which you get to use because you're at work most of the time?
Furthermore, I claim that you can't be productive at the number of hours that people work. People cannot work 80 productive hours a week. They can push themselves to be *at* work 80 hours a week, but there's no way that they're getting that much done.
France and Germany both seem to have much more liberal hours-of-work and vacation policies. So what if you make a bit less money if you aren't beating yourself to death trying to claw your way ahead?
We currently have unemployment problems in the US. Lots of people out of work. We also have lots of people that are well-paid but overworked. It just seems like there should be, you know, an obvious solution to this. Hire more people and pay a lower pay rate.
I remember seeing a wooden mouse, at least (and one that looked *much* nicer than this one -- a bit more angular and with large buttons that covered the whole top). It was being sold with, I believe, either a wooden monitor or keyboard. Darned if I remember which, though. It wasn't both.
It is kind of depressing. Just as Microsoft unjustly gets credit for "giving us easy-to-use computers", Apple gets credit for "bringing us stylish case design".
(a) Have you actually read the license, or are you just assuming that there are legal violations in all three cases? RealPlayer has been structured such that the library is not part of the binary -- exactly so that third-party use like this could take place. It is possible that the distribution separate from the player app is illegal (not sure, as I said, I haven't read the license), but it can certainly be expected that the use is not out of whack.
(b) None of these companies have gone after MPlayer, despite the fact that the mplayer people have zipped copies of the DLLs from the installers on their websites. It would *not* be hard for them to do so -- companies send C&Ds all the time. One can conclude that it might *not* be in their interest to do so. REing for interprogram compatibility *is* legally protected and an exemption in the DMCA. If any of these companies force the point, they will simply suppress these codecs for six months or so, and then have have RE'd legal open source and free implementations of their codecs floating around, which will mean that they will lose the ability to force people to pay licensing fees (the other alternative is a mass of techies pushing for the non-use of a format, which is not something to sneeze at). Do not shed tears for MS/Apple/Real -- they're doing what's in their best interest.
(c) Yes, people don't get angry at MPlayer. MPlayer is the *only* way to play Quicktime or Windows Media formats on Linux, since Apple and Microsoft refuse to support Linux, and a much nicer application than Real. Why should anyone be upset about MPlayer? It only benefits them.
Personally, I think that copyright protection for a piece of software should not be granted unless the protocols and formats used by that software are made publically available (source, in the case of Open Source software, would be an acceptable form of documentation). Almost all of the grief that consumers suffer from commercial software comes from an inability of competitors to enter the market because of compatibility issues. It's not that software costing money is a problem -- hell, people spent time and money creating the software, and if they want recompense, fair enough. The thing that makes people angry is that the thing driving sales absolutely should not be artificial monopolies created by secret formats.
For starters, a lot of "NASA work" is done by contractors. I dunno what the breakdown is, but I'd guess that majority of NASA's funding goes to contractors. Which are private companies.
Microprocessors? What did NASA do for microprocessors aside from adapting them to space travel? What could they possibly sell to a terrestrial company?
Cryogenics? Yeah, that's real feasible and salable.
The people wanted Bush II, they voted for Bush II, and they got Bush II.
Well, technically they didn't vote for Bush II -- Bush lost the popular vote and whether or not he legitimately won the electoral college vote is a debate that will never end. However, a *lot* of people apparently wanted "more Bush". Clinton did feel-good stuff, and brought "immorality" to the White House. The religious right said "we don't think so", and put in Bush II, who goes in more for intimidation than feel-good stuff, and pushes religious (charities, abstinence, anti-homosexual) agendas.
I hate to break it to you, but the only reason mplayer is a viable alternative to things like Media Player, Real One and Quicktime is because it uses the Win32 DLLs in a very dubious (license-wise) manner. The Win32 DLLs written by the "crappy capitalistic companies" like Microsoft, Real and Apple (all of which have released open source software, by the way).
No, it isn't all using win32 dlls. IIRC, here's the breakdown:
* AVI and ASF (the wrapper formats) have been reverse-engineered and reimplemented natively.
* RTSP has been implemented natively.
* RealMedia's codecs are implemented by using a Linux-native shared library that Real exposes *specifically* to allow third-party software to do decoding. Seeking in RealMedia content is not supported.
* divx (not technically from MS, though the original codebase originated from MS code) is implemented natively.
* Quicktime (the wrapper format) is implemented natively for older versions, but newer versions require use of Win32 DLLs.
* Sorenson v1 and v2 are reverse-engineered and implemented natively.
* Sorenson v3, I believe, requires use of a Win32 DLL.
* Indeo requires use of Win32 dlls.
The Win32 DLLs written by the "crappy capitalistic companies" like Microsoft, Real and Apple (all of which have released open source software, by the way).
The problem has nothing to do with the company releasing open source software. The problem is that, while it's difficult-but-doable to make your own video codec, it's extremely hard to produce an exactly compatible player without format information. This has nothing to do with Apple, Real, or Microsoft having better designers -- it has to do with none of them having to reimplement someone *else*'s codec without technical information.
I've got to take exception with your classifying quicktime under 'horrible stuff'.
* It's proprietary. Apple has made up the biggest collection of bullshit to avoid releasing specs on the format to allow mplayer to implement a decoder (Apple: "We can't do that -- only Sorenson can" and Sorenson: "We can't do that -- only Apple can").
* The player sucks. There has been enough analysis of the problems with the player interface (especially on Windows) that I don't think I need to analyze it any more.
* It's closed-source, and hence a pain in the ass for Linux users.
* It doesn't support Linux.
* It is nagware. Even the awful, awful RealPlayer, with all the corporate links and the calling back home, doesn't keep popping up ads for the software, and they're the next-most-annoying player.
How about the simple argument that planets are gravitationally strong enough to pull themselves into nearly spherical objects, whereas asteroids are not. Pluto, BTW, Sedna, and many of the largest moons can all do this.
Doesn't that tend to favor gas planets like Jupiter?
Some browsers do not send referrer information or have it blocked.
I would suspect that a link from an email would result in no referrer data being sent in most web browsing environments.
Since the two possibilities (legit user, no referrer data and bogus user, clicked on email) can't be distinguished easily, I would expect that Google just waits for a lot of hits to suddenly come in on an AdWord, with a high percentage of click-through. Especially if most do not have host-referer.
It amazes me the number of people that talk about IP spoofing. All their wild theories can be shot down by asking "What happens when you send out thatfirst packet and it comes back to confirm it?"
TCP spoofing is quite possible. It's just difficult, and has become progressively more difficult.
Say we have just the idea of a basic handshake (without worrying about the way TCP works for a moment). Host A sends a packet to Host C purporting to come from Host B. Host C sends a packet to Host B saying "you really want to open this connection?". Host A waits a short period of time, then sends another packet to Host C claiming to be from Host B saying "yup, open it". Handshake completed.
Now, in TCP world, there are a couple of complications. First, Host B is supposed to respond back when it gets the "do you want to open this packet" question from Host C with a "Nope, blow away the connection" response. So, just for starters, Host B has to be unresponsive. That means that it might be a good idea for Host C to compromise a bunch of hosts and flood Host B starting sometime before sending that first forged packet to Host A. This bumps Host B's packet loss rate up to, say, 90%. That means that there's an awfully good chance that the "Do you want to connect" message never gets through to B...all C has is the forged response from A, so it considers the response valid and opens the connection.
Then we have sequence numbers. TCP uses sequence numbers to ensure that packets don't get lost or out of order. A's bogus response to C has to have a sequence number based on the number that A included in its "do you want to connect" message to B. The traditional way to get around this was to have C try to open a (non-forged) connection to A. A's response contains a sequence number. C ignores this response, and when A sends out its first packet to B, as long as nobody else has opened any connections in the interim, it uses a starting sequence number that is, say, one greater than the previous starting sequence number. At least, there is some form of correlation that C can use to determine the sequence number being sent to B that will allow it to forge a packet with a valid sequence number.
Most modern machines, to avoid exactly this sort of attack, generate an "unpredictable" number. However, since entropy (I guess you'd call entropy "stored unpredictability" -- data based entirely on unpredictable events from outside the computer's operating environment) comes in at a pretty limited rate in a typical machine, machines tend to just mangle some data in a hard-to-predict manner and use it to derive a starting sequence number for the next connection. Ideally, this sequence number cannot be predicted by host A -- in reality, it's possible that host A might manage to do so, if controlled by someone that's figured out a way to predict the output of the algorithm being used by host C.
If the A and C machines are both on a fast network (a business or university, say), it might even be possible to forge a connection through brute-force guessing of the next sequence number.
So, spoofing a TCP connection is difficult, but feasible. TCP is definitely not considered to be secure as a cryptographer would consider something to be secure.
So it's a good move to avoid using IP-based authentication.
Slashdot Mirror
Google doesn't store images. Right back at square one.
The Wayback Machine doesn't have it, and it's probably too late for anyone to mirror it.
With click-to-focus you always have to be careful not to click on a button or if you click on an editor, it automatically moves your cursor from where it should be.
oh, good point.
I've bound my mouse's fourth button to be "raise, focus, and allow me to move window". Took a couple of lines of LISP in sawfish, but it's terribly useful. I don't need to worry about passing through clicks (or about hunting for and hitting the titlebar when I want to move a window).
The welfare states of Europe are great if you just want to slide by and work part time while having everyone else pay for your existence when you get too lazy to be productive. The problem is that this means there is no innovation and no dynamic growth, both of which are the keys to the American economy. Europe is a great place to be for paper MCSE's or people with enough political pull to get a cushy union job, it is not a great place for innovating and pushing the boundaries of new technology.
I dunno. How many great advances are made by burned-out people stuck at work to the breaking point? I usually get most of my good ideas when I'm showering, where I get a chance to unwind and think about the material I was working on. I've tried writing code for more than 60 hours a week. I can do about one week of that, then I just hit "code burnout", where it's like whacking your head with a rubber bat. Your LOC per hour goes waaay down, and you start staring at the screen without seeing it. It just doesn't *work*.
Re:Hey! Asses! (Score:3, Redundant)
*Please* tell me that there isn't *another* post that I missed talking about six developers gagging and spitting coffee.
The problem is that 2d acceleration is a Good Thing and plain old framebuffers don't give you much by way of that.
I like focus-follows mouse, magic desktop borders and transparent thingies.
I like most fvwm-ish things (and zero resistance edge flipping!), but focus-follows-mouse always confused me.
It seems like this focusing system always tends to result in your mouse cursor winding up covering up what you're trying to work with. Usually, I'd prefer to have my mouse cursor elsewhere.
Did I miss something when I installed Linux?
Where are the nekkid chicks on my wall paper?
here. You probably used Red Hat or SuSE. Given time, Open Source fills all niches.
BTW, if anyone from Sun is reading this, that HCI work done was much appreciated. Formal testing was something that had been severely lacking and that helps out the salabilty of the environment well.
s h, the HCI feedback was a big deal.
While this 3d business seems a bit...well...researchy-without-immediate-benefiti
Sun needs to work on precognitive technology? :-)
Marvellous. Just as users of current operating systems have ben doing for years anyway.
Linux is a current operating system, but the mainstream Linux GUI environment does not provide this (watch for rapid changes now that freedesktop.org is coming to the fore).
Hmm. Potentially interesting as a way to pick between open windows, but doesn't Expose perform this task in a better manner?
It's eye candy, little more. Frankly, I view Expose as eye candy -- multiple viewports are a far more powerful mechanism, and one that I'm sad that Apple chose not to support.
Except that in the real world I can never find the bloody CDs, because I can't remember where I've put them. I can navigate a media player interface far faster than I can hunt for CDs, and I can use more search criteria too (album, artist etc.)
Agreed. People were raving about real-world metaphors in research labs a couple of years ago, but nothing worked out all that well. It seems to me that all the features mentioned here could be provided better through a 2d implementation.
did Germany condemn any Wolfenstein game(s)?
Uh...Germany didn't bother condemning them. They just outright banned them.
Is Vietnam condemning Battlefield Vietnam?
I haven't the faintest idea what the political situation is like in Vietnam, though I doubt that video gaming is a prominent issue there.
On a side note, imagine the fun we'd have playing a game based on the US/Canada war of 1812...
I think that sounds like fun.
On the other hand, I also thought that it would be fun to play not only as the police but as the sniper in a UT clocktower scenerio in one of the hostage rescue/counterterrorism games.
Nurses are an example of a profession even less "happy" than IT. While nurses aren't worried about their jobs being outsourced, interestingly enough they tend to be overworked (usually 10-12 hour rotating shifts throughout the week) and underpaid, especially compared to their colleagues. But, according to the articles, they should be happy, as they are very hands on. Guess that hypothesis just got shot down.
New studies have indicated that working people as far as you can stretch them *makes them unhappy*!
I could never figure out why the employment situation in the US is so screwed up.
We have this kind of go-go-GO-OR-YOU-FAIL-DAMMIT-GO! mentality that keeps being pushed. I was talking to some folks about the kind of hours that people starting off in financial services or the legal world can expect to work -- the hours are *stupid*. Sure, the jobs pay well, but what do you do with the money? Buy a bigger TV or a more expensive car, neither of which you get to use because you're at work most of the time?
Furthermore, I claim that you can't be productive at the number of hours that people work. People cannot work 80 productive hours a week. They can push themselves to be *at* work 80 hours a week, but there's no way that they're getting that much done.
France and Germany both seem to have much more liberal hours-of-work and vacation policies. So what if you make a bit less money if you aren't beating yourself to death trying to claw your way ahead?
We currently have unemployment problems in the US. Lots of people out of work. We also have lots of people that are well-paid but overworked. It just seems like there should be, you know, an obvious solution to this. Hire more people and pay a lower pay rate.
I remember seeing a wooden mouse, at least (and one that looked *much* nicer than this one -- a bit more angular and with large buttons that covered the whole top). It was being sold with, I believe, either a wooden monitor or keyboard. Darned if I remember which, though. It wasn't both.
This must have been at *least* five years ago.
I use an old SGI Crimson as a coffee table.
It is kind of depressing. Just as Microsoft unjustly gets credit for "giving us easy-to-use computers", Apple gets credit for "bringing us stylish case design".
You know, we had a discussion in the Games section on how sex and videogaming journalism had a lot of ties.
The MaxitMag site, the one doing the HL2 article, really drives that discussion home.
(a) Have you actually read the license, or are you just assuming that there are legal violations in all three cases? RealPlayer has been structured such that the library is not part of the binary -- exactly so that third-party use like this could take place. It is possible that the distribution separate from the player app is illegal (not sure, as I said, I haven't read the license), but it can certainly be expected that the use is not out of whack.
(b) None of these companies have gone after MPlayer, despite the fact that the mplayer people have zipped copies of the DLLs from the installers on their websites. It would *not* be hard for them to do so -- companies send C&Ds all the time. One can conclude that it might *not* be in their interest to do so. REing for interprogram compatibility *is* legally protected and an exemption in the DMCA. If any of these companies force the point, they will simply suppress these codecs for six months or so, and then have have RE'd legal open source and free implementations of their codecs floating around, which will mean that they will lose the ability to force people to pay licensing fees (the other alternative is a mass of techies pushing for the non-use of a format, which is not something to sneeze at). Do not shed tears for MS/Apple/Real -- they're doing what's in their best interest.
(c) Yes, people don't get angry at MPlayer. MPlayer is the *only* way to play Quicktime or Windows Media formats on Linux, since Apple and Microsoft refuse to support Linux, and a much nicer application than Real. Why should anyone be upset about MPlayer? It only benefits them.
Personally, I think that copyright protection for a piece of software should not be granted unless the protocols and formats used by that software are made publically available (source, in the case of Open Source software, would be an acceptable form of documentation). Almost all of the grief that consumers suffer from commercial software comes from an inability of competitors to enter the market because of compatibility issues. It's not that software costing money is a problem -- hell, people spent time and money creating the software, and if they want recompense, fair enough. The thing that makes people angry is that the thing driving sales absolutely should not be artificial monopolies created by secret formats.
Dammit, that's "Bush bootlicker", not "US bootlicker".
Most of us up here in the US would like to see the architect of the War for Oil out as much as you'd like to see Howard out, if not more.
Wait a minute.
For starters, a lot of "NASA work" is done by contractors. I dunno what the breakdown is, but I'd guess that majority of NASA's funding goes to contractors. Which are private companies.
Microprocessors? What did NASA do for microprocessors aside from adapting them to space travel? What could they possibly sell to a terrestrial company?
Cryogenics? Yeah, that's real feasible and salable.
I don't have a clue about medical telemetry, but I haven't heard of "remote surgeons" existing on the Shuttle, that's for sure. The problem with remote surgery has little to do with technical issues, and more to do with the fact that people don't trust robots to do potentially fatal medical work (thank you Therac-25 programmers). I was at a talk given by someone that was working on robot-assisted medical work -- the best thing that they could hope to have introduced was to have programmed systems with brakes, where the system was incapable of providing any force at all, just preventing the surgeon from accidently moving a scalpel outside of the parameters he inserted.
As for improvements in systems analysis software, it's possible, but I have no idea what you're referring to.
The people wanted Bush II, they voted for Bush II, and they got Bush II.
Well, technically they didn't vote for Bush II -- Bush lost the popular vote and whether or not he legitimately won the electoral college vote is a debate that will never end. However, a *lot* of people apparently wanted "more Bush". Clinton did feel-good stuff, and brought "immorality" to the White House. The religious right said "we don't think so", and put in Bush II, who goes in more for intimidation than feel-good stuff, and pushes religious (charities, abstinence, anti-homosexual) agendas.
If they take a dislike to you, they can make life very difficult for you immediately by ordering a body cavity search.
Yeah, that's going to happen.
"So...you've got *one* undeclared laptop...maybe you've got a *second* one wedged *up your ass*. Bend over, sir, we're going to have to take a look."
Even customs officials can lose their jobs.
I hate to break it to you, but the only reason mplayer is a viable alternative to things like Media Player, Real One and Quicktime is because it uses the Win32 DLLs in a very dubious (license-wise) manner. The Win32 DLLs written by the "crappy capitalistic companies" like Microsoft, Real and Apple (all of which have released open source software, by the way).
No, it isn't all using win32 dlls. IIRC, here's the breakdown:
* AVI and ASF (the wrapper formats) have been reverse-engineered and reimplemented natively.
* RTSP has been implemented natively.
* RealMedia's codecs are implemented by using a Linux-native shared library that Real exposes *specifically* to allow third-party software to do decoding. Seeking in RealMedia content is not supported.
* divx (not technically from MS, though the original codebase originated from MS code) is implemented natively.
* Quicktime (the wrapper format) is implemented natively for older versions, but newer versions require use of Win32 DLLs.
* Sorenson v1 and v2 are reverse-engineered and implemented natively.
* Sorenson v3, I believe, requires use of a Win32 DLL.
* Indeo requires use of Win32 dlls.
The Win32 DLLs written by the "crappy capitalistic companies" like Microsoft, Real and Apple (all of which have released open source software, by the way).
The problem has nothing to do with the company releasing open source software. The problem is that, while it's difficult-but-doable to make your own video codec, it's extremely hard to produce an exactly compatible player without format information. This has nothing to do with Apple, Real, or Microsoft having better designers -- it has to do with none of them having to reimplement someone *else*'s codec without technical information.
I've got to take exception with your classifying quicktime under 'horrible stuff'.
* It's proprietary. Apple has made up the biggest collection of bullshit to avoid releasing specs on the format to allow mplayer to implement a decoder (Apple: "We can't do that -- only Sorenson can" and Sorenson: "We can't do that -- only Apple can").
* The player sucks. There has been enough analysis of the problems with the player interface (especially on Windows) that I don't think I need to analyze it any more.
* It's closed-source, and hence a pain in the ass for Linux users.
* It doesn't support Linux.
* It is nagware. Even the awful, awful RealPlayer, with all the corporate links and the calling back home, doesn't keep popping up ads for the software, and they're the next-most-annoying player.
How about the simple argument that planets are gravitationally strong enough to pull themselves into nearly spherical objects, whereas asteroids are not. Pluto, BTW, Sedna, and many of the largest moons can all do this.
Doesn't that tend to favor gas planets like Jupiter?
Some browsers do not send referrer information or have it blocked.
I would suspect that a link from an email would result in no referrer data being sent in most web browsing environments.
Since the two possibilities (legit user, no referrer data and bogus user, clicked on email) can't be distinguished easily, I would expect that Google just waits for a lot of hits to suddenly come in on an AdWord, with a high percentage of click-through. Especially if most do not have host-referer.
It amazes me the number of people that talk about IP spoofing. All their wild theories can be shot down by asking "What happens when you send out thatfirst packet and it comes back to confirm it?"
TCP spoofing is quite possible. It's just difficult, and has become progressively more difficult.
Say we have just the idea of a basic handshake (without worrying about the way TCP works for a moment). Host A sends a packet to Host C purporting to come from Host B. Host C sends a packet to Host B saying "you really want to open this connection?". Host A waits a short period of time, then sends another packet to Host C claiming to be from Host B saying "yup, open it". Handshake completed.
Now, in TCP world, there are a couple of complications. First, Host B is supposed to respond back when it gets the "do you want to open this packet" question from Host C with a "Nope, blow away the connection" response. So, just for starters, Host B has to be unresponsive. That means that it might be a good idea for Host C to compromise a bunch of hosts and flood Host B starting sometime before sending that first forged packet to Host A. This bumps Host B's packet loss rate up to, say, 90%. That means that there's an awfully good chance that the "Do you want to connect" message never gets through to B...all C has is the forged response from A, so it considers the response valid and opens the connection.
Then we have sequence numbers. TCP uses sequence numbers to ensure that packets don't get lost or out of order. A's bogus response to C has to have a sequence number based on the number that A included in its "do you want to connect" message to B. The traditional way to get around this was to have C try to open a (non-forged) connection to A. A's response contains a sequence number. C ignores this response, and when A sends out its first packet to B, as long as nobody else has opened any connections in the interim, it uses a starting sequence number that is, say, one greater than the previous starting sequence number. At least, there is some form of correlation that C can use to determine the sequence number being sent to B that will allow it to forge a packet with a valid sequence number.
Most modern machines, to avoid exactly this sort of attack, generate an "unpredictable" number. However, since entropy (I guess you'd call entropy "stored unpredictability" -- data based entirely on unpredictable events from outside the computer's operating environment) comes in at a pretty limited rate in a typical machine, machines tend to just mangle some data in a hard-to-predict manner and use it to derive a starting sequence number for the next connection. Ideally, this sequence number cannot be predicted by host A -- in reality, it's possible that host A might manage to do so, if controlled by someone that's figured out a way to predict the output of the algorithm being used by host C.
If the A and C machines are both on a fast network (a business or university, say), it might even be possible to forge a connection through brute-force guessing of the next sequence number.
So, spoofing a TCP connection is difficult, but feasible. TCP is definitely not considered to be secure as a cryptographer would consider something to be secure.
So it's a good move to avoid using IP-based authentication.