Configurability is not necessarily a good thing. A few months ago i was visiting a couple who are true mac users (i think they even met working at the local mac store), we were planning a vacation with the help of their two somethingbooks running a version of OS X. From time to time everybody seemed to browse this, calculate that, on each of the systems.
"how do i switch apps with this thing?" - "press {some key combination}"
of course i tried to memorize that, it's good to know the basics of operation of as many systems as possible. a few minutes later i had the other somethingbook in my hands, pressed those keys and and, you probably know it, i did not, nothing happened. her mate used a different shortcut than her for such a basic thing.
i enjoy customizing my system as much as anybody else (beware of touching anything but the center of the synaptics' on this dell), but if it's driven as far as (or if the system even forces you to go as far as) making it impossible for you to fluidly operate a vanilla system it turns evil.
besides, the shortcuts in windows are not half as bad as most switchers think who, simply did not bother to learn about them back in their windows days. it has happened often enough that i routinely pressed one of the few combos i know (many more than the alt-tab, alt-f4 and the notorious, often misunderstood ctrl-alt-del, but still less than 30% of what's there) in front of a switcher and to my complete surprise the switcher looked at me as if i was a magician who made windows do impossible things.
granted, microsoft does an incredibly bad job at advertizing those things and they seem to be so low-priority for the marketing fools at redmond that i would not be surprised if MS was stupid enough to leave most of the more useful ones out in vista ("keyboard shortcuts do not fit into the experience we have envisioned for vista" or some other crap like that - or just forgetting) but they are there, waiting for me on every windows machine i encounter (ok, i would not find them on cyrillic or asian keyboards). i don't demand that they work on macs, or blame macs for their shortcuts not working on windows, but macs don't even seem to get "on other macs" properly done.
older cars would not be that bad for a project like this, since they used put way less steel into tunring the car into a crash-proof tank 20 years ago than they are doing today (at least speaking of european cars like the smaller ranges of VW, a 50ies US cruiser with cast-iron tail fins would probably be a different story). "passive security" might be nice for people who are even scared of walking, but it poses a bad hit on fuel economy.
but that aside, i question the whole idea of converting cars to electricity. as long as most of our energy comes from fossile sources, it is the most environmentally friendly thing to allocate that fossile fuel to mobile applicatoins, where their high energy density makes a real difference in overall energy consumption.
hybrid is a different story, it still sources from the dense fossile fuel but uses some electric components to kind of "low pass filter" the overall energy demand over time, leading to more efficient burning of the primary energy carrier.
that all being said, there are electric cars that are very friendly to the environment (like this), but they are friendly to the environment because they use less total energy thanks to being extremely lightweight and low-powered and not because of being electrically powered. i would really love to see a hybrid built on a base like that, with a low-powered, high-efficiency primary power source based on fossile/bio high density liquid fuel. could give amazing range of operationat extremely low cost, but i guess there is not enough money in developing a sufficiently small engine far enough to reach the efficiency of modern diesel engines.
no chance, since the trend to smaller mobile phones has reversed a few years ago, when the started to converge away the markets of all the other gadgets, one at a time. personally i like this, because it is easier to carry one big phone than to carry a (writs)phone, an electric organizer, an mp3 player and a cheap clickpicthing. the volume of all those things mainly consist of a display and a battery, why would anyone not want to multiplex as much functions as possible through those bottlenecks?
"as every other people of the world the iranian people have the undeniable right to use violent video games for peaceful purposes!" heh, i want him to say that in some hyped up speech in front of the UN
well, rockbox shows us that many of those boxes are fast enough to decode ogg, but one could probably expect a slightly shorter battery runtime (which could theoretically be offset by using a lower bitrate to reduce drive access, at least with hdd players). but of course you are right about the cpu not pushing all the weight of something like mpeg4 itself. the question is wether the specialist hardware is just some black box where the general purpose cpu sends in a shorter bitstream and receives a longer bitstream or if it's just some kind of offload engine for certain typical subtasks. the latter could help vorbis decoding as well as any other decoding.
but the most telling thing is the high number of ogg-aware no-name (or less-than-ipod-brand) players out there, these days nearly every piece of plastic offered in the mid price region seems to have ogg somewhere in the codec list.
i don't trust the cowon M3 because of it's reliance on a cable remote. as much as i love cable remotes i know how much those cables suffer from wear and tear.
the X5/M5 has this mini joystick, if it is as unusable as the one on my w800i then i would want to avoid that too.
in all other aspects of course, the cowons blow the ipod away. if only they were not considerably more expensive than an ipod with the same capacity, at least where i live:(
for me the core bonus of ogg support would lie in avoiding having to encode a given piece of music more than once. this can be very interesting for people who often create audio content themselves for uploading somewhere on the web and want to avoid close quarters warfare with the thomson lawyer squad.
admittedly, this issue used to be more pressing when cpus were slower at encoding audio - but at the same time the "voirbis heavy on cpu" argument is also shrinking in the face of mpeg4 video capable mobile players.
the availability of the rockbox port makes me really considering buying an ipod, even if only because it happens to be by far the cheapest offering in music centered (i don't even watch tv, why would i need a mobile video player?) players with >20 gb hdds
guess i should wait a while though for the port to get more polished (maybe i could get a cheaper 5th gen when the new ones come out? but iirc apple was always very good at not having many leftover predecessors in stores when the new gen comes out)
isn't "tile based" just a special case for what the article calls "beams"?
anyways, if i understand you right then the tile based approach hardly applies to anything but first-generation backwards rays for viewport pixels, which are only a small fraction of the rays used in a nontrivial raytracer.
(if you are talking about bounding box hierarchies and not viewport tiles, then sorry for the misunderstanding)
what you described is pretty much what the paper which would be "TFA" in "RTFA" is about. scaling with hyperthreading and multicore, and the good hit rate with common cpu cache hierarchies. (where the "hyperthreading" part tells us something about when the underlying experiments where done)
looks like that article was mostly about that same paper.
the whole idea of "reviving the ray tracer" seems to be to be less bound by triangle counts and being able to replace the bump maps with actual bumps instead.
> A colleague of mine recently had a brief paper
> (restricted to a maximum of two pages) rejected
> because it was too short - at exactly two pages.
> I kid you not.
How could this surprise anyone?
I don't think anyone would ever claim that the current system was any good at letting the good papers in - but it's job is less to identify all the good papers than it is to identify all the bad ones.
it's like the exact opposite of an email spam filter: with email, a few "nigerians" in the inbox do much less harm than a false positive in the junk folder, while one bad paper more in a conference or a journal would certainly be seen as worse than one good paper less.
having said that, i have to realize that my arguments are not always in line with empiricism.
while it's quite obvious that siggraph is monopolising the conference "market" of it's field much more than the top conferences of other fields do, it's easy to understand why:
the top conferences of other fields might be the most important thing in their field, but they still not known to the general public. most of the people who know that conference will also have heard of the smaller ones. with siggraph things are very different, it's so famous that the number of people who roughly know what siggraph is is much higher than the number of people who have heard of any other cgi science related conference. of course the relevant group, i.e. the scientists will all know the other conferences too, but their relative assessent will undoubtedly be biased by that public visibility of siggraph.
so, how qualified would the average computer graphics scientist be for ethnomusicology? the "broader range of subjects" of academia is off-topic, simply put.
a switch of fields within computer science would be tough enough in itself, but it seems to be much more feasible than to drop into a discipline where he would completely start at zero. in fact, transferring the CGI people's deep understanding of mind boggingly optimized data structures into other fields of computer science could even prove to be a very valuable contribution.
oh, and about that general problem of academia: sure, every field has it's top conferences that are incredibly hard to get into without knowing the right people, but with siggraph it's much more drastic, because of the huge public visibility of it. every apprentice geek knows siggraph, and even your friend the photoshop monkey who thinks unix has been invented by steve jobs when he wrote OS X.
with a famous conference like that, the difference between the one top conference and "the rest" is much bigger than in other fields. for how many of the fields mentioned in your post (cs and non-cs) can you name the top conferences that are hard to get into?
i guess my smac match count is quite high above average, and i would even be surprised if i played any of the other civs as often as i played smac (well, the original civ might be an exception).
if you take out the different weapon and armor itearations - how many classes of units do you really use?
ignoring the civil units (clean formers are a nice touch, but will they really average out to survive long enough to make the investment pay off? oh, and airborne colony pods may be fine, but can't really shine next to icbm colony pods) and the end-game stuff like orbital dropping i rarely get much more than a handful different lines of units, each with numerous iterations going through the tech levels.
which leads us to the major drawback: the techlevel treadmill, upgrading all your lines to the latest developments in the armor/gun range.
the other drawback is that the game can't provide really "fleshed out" names for the units, they end up being just anonymous combinations of attack/defense/chassis/special(/special2).
> And hell! Origin could release Ultima 7 for PC's in 1993.
How many people worked on that game?
We still love those old pixel grids more than the shiny new HDR graphics, but only because they are our childhood memories. give any person is used to todays visual quality and instant gratification a game made like those old ones with which he does not connect any memories of "the good old times" and he would never endure the pains of long text passages, crude menus etc.
today we see the old games as warm and strong on story, but try to remember: back then we did care little about story, we mostly exposed us to those things because they had those really marvellous graphics, computers were the future and generally amazing miracles, they could even play games!
you can't bring that back, even linux can't do that
as you said, he was really good at making early 90ies games. as many people were.
i don't see any clue at all that would indicate that he is good at making post 2k games.
he did have the will to try new things and this is a good thing. but in trial and error you have to accept the errors and that's the problem with molyneux.
in the end i definitely prefer an average shooter over what b&w turned out to be. that one quickly becomes repetive in a boring way (sure the shooter has many repetive elements, but it kind of entertains me while repeating) if you don't accidentally get on the track of that one strategy that will bring you through the level without going through a sweat-mill of epic dimensions.
yeah, multiple competing coop squads operating in a world filled with enough NPCs to make it interesting even if the two (or more) teams would maybe never meet directly in some matches.
why? because it would make the direct meeting experiences feel special and not routine.
and indirect strategies (like blocking the supply of civilian cars to teams needing them or something like that) could make for a very interesting gameplay. gta3-like engine (otoh: a fresh isometric would certainly rock for us oldschoolers), with much smaller worlds to allow for persistent NPCs during matches. sounds like an idea?
I think the unit builder was the worst of alpha centauri (but the rest was so good, i still love it most of all civs, which, uhm, means quite a bit because the others are so good. i especially like the way the factions are presented, the feel so much more real than those civ nations which barely differ more than in the face that presents them)
the problem i had with unit builder: the units do not really evolve over long periods of time, it's still "unarmoured buggy, n attack", with only n changing, same goes for the planes etc. in the end game there are all those really funky extras like drop pods and the new chassis, but imho they come way too late in the game.
btw: civ4 has "learned" the good parts of the unit builder, the promotions don't just make your units stronger but allow for special abilities which give even more interesting combinations than in SMAC, but at the instance level, not at the class level, if you allow the analogy. at the same time it is even less administrative work, because there are no bulk upgrades as in SMAC.
Slashdot Mirror
Configurability is not necessarily a good thing. A few months ago i was visiting a couple who are true mac users (i think they even met working at the local mac store), we were planning a vacation with the help of their two somethingbooks running a version of OS X. From time to time everybody seemed to browse this, calculate that, on each of the systems.
"how do i switch apps with this thing?" - "press {some key combination}"
of course i tried to memorize that, it's good to know the basics of operation of as many systems as possible. a few minutes later i had the other somethingbook in my hands, pressed those keys and and, you probably know it, i did not, nothing happened. her mate used a different shortcut than her for such a basic thing.
i enjoy customizing my system as much as anybody else (beware of touching anything but the center of the synaptics' on this dell), but if it's driven as far as (or if the system even forces you to go as far as) making it impossible for you to fluidly operate a vanilla system it turns evil.
besides, the shortcuts in windows are not half as bad as most switchers think who, simply did not bother to learn about them back in their windows days. it has happened often enough that i routinely pressed one of the few combos i know (many more than the alt-tab, alt-f4 and the notorious, often misunderstood ctrl-alt-del, but still less than 30% of what's there) in front of a switcher and to my complete surprise the switcher looked at me as if i was a magician who made windows do impossible things.
granted, microsoft does an incredibly bad job at advertizing those things and they seem to be so low-priority for the marketing fools at redmond that i would not be surprised if MS was stupid enough to leave most of the more useful ones out in vista ("keyboard shortcuts do not fit into the experience we have envisioned for vista" or some other crap like that - or just forgetting) but they are there, waiting for me on every windows machine i encounter (ok, i would not find them on cyrillic or asian keyboards). i don't demand that they work on macs, or blame macs for their shortcuts not working on windows, but macs don't even seem to get "on other macs" properly done.
true electric cars have a very big part of the money in their batteries. good luck at DIYing huge lipolys.
and don't even think of the cost of replacing them every few years.
older cars would not be that bad for a project like this, since they used put way less steel into tunring the car into a crash-proof tank 20 years ago than they are doing today (at least speaking of european cars like the smaller ranges of VW, a 50ies US cruiser with cast-iron tail fins would probably be a different story). "passive security" might be nice for people who are even scared of walking, but it poses a bad hit on fuel economy.
but that aside, i question the whole idea of converting cars to electricity. as long as most of our energy comes from fossile sources, it is the most environmentally friendly thing to allocate that fossile fuel to mobile applicatoins, where their high energy density makes a real difference in overall energy consumption.
hybrid is a different story, it still sources from the dense fossile fuel but uses some electric components to kind of "low pass filter" the overall energy demand over time, leading to more efficient burning of the primary energy carrier.
that all being said, there are electric cars that are very friendly to the environment (like this), but they are friendly to the environment because they use less total energy thanks to being extremely lightweight and low-powered and not because of being electrically powered. i would really love to see a hybrid built on a base like that, with a low-powered, high-efficiency primary power source based on fossile/bio high density liquid fuel. could give amazing range of operationat extremely low cost, but i guess there is not enough money in developing a sufficiently small engine far enough to reach the efficiency of modern diesel engines.
i can't wait to keep all those tiny batteries maintained. i wired PAN might be a different thing, but it would make you look very stupid.
no chance, since the trend to smaller mobile phones has reversed a few years ago, when the started to converge away the markets of all the other gadgets, one at a time. personally i like this, because it is easier to carry one big phone than to carry a (writs)phone, an electric organizer, an mp3 player and a cheap clickpicthing. the volume of all those things mainly consist of a display and a battery, why would anyone not want to multiplex as much functions as possible through those bottlenecks?
bluetooth is good enough at that by itself, i guess this could be an application for zigbee, but we won't see that on phones, ever.
"as every other people of the world the iranian people have the undeniable right to use violent video games for peaceful purposes!" heh, i want him to say that in some hyped up speech in front of the UN
yeah, the description of that tanker having exactly two "bomb-spots" did ring a bell.
well, rockbox shows us that many of those boxes are fast enough to decode ogg, but one could probably expect a slightly shorter battery runtime (which could theoretically be offset by using a lower bitrate to reduce drive access, at least with hdd players). but of course you are right about the cpu not pushing all the weight of something like mpeg4 itself. the question is wether the specialist hardware is just some black box where the general purpose cpu sends in a shorter bitstream and receives a longer bitstream or if it's just some kind of offload engine for certain typical subtasks. the latter could help vorbis decoding as well as any other decoding.
but the most telling thing is the high number of ogg-aware no-name (or less-than-ipod-brand) players out there, these days nearly every piece of plastic offered in the mid price region seems to have ogg somewhere in the codec list.
i don't trust the cowon M3 because of it's reliance on a cable remote. as much as i love cable remotes i know how much those cables suffer from wear and tear.
:(
the X5/M5 has this mini joystick, if it is as unusable as the one on my w800i then i would want to avoid that too.
in all other aspects of course, the cowons blow the ipod away. if only they were not considerably more expensive than an ipod with the same capacity, at least where i live
for me the core bonus of ogg support would lie in avoiding having to encode a given piece of music more than once. this can be very interesting for people who often create audio content themselves for uploading somewhere on the web and want to avoid close quarters warfare with the thomson lawyer squad.
admittedly, this issue used to be more pressing when cpus were slower at encoding audio - but at the same time the "voirbis heavy on cpu" argument is also shrinking in the face of mpeg4 video capable mobile players.
the availability of the rockbox port makes me really considering buying an ipod, even if only because it happens to be by far the cheapest offering in music centered (i don't even watch tv, why would i need a mobile video player?) players with >20 gb hdds
guess i should wait a while though for the port to get more polished (maybe i could get a cheaper 5th gen when the new ones come out? but iirc apple was always very good at not having many leftover predecessors in stores when the new gen comes out)
isn't "tile based" just a special case for what the article calls "beams"?
anyways, if i understand you right then the tile based approach hardly applies to anything but first-generation backwards rays for viewport pixels, which are only a small fraction of the rays used in a nontrivial raytracer.
(if you are talking about bounding box hierarchies and not viewport tiles, then sorry for the misunderstanding)
what you described is pretty much what the paper which would be "TFA" in "RTFA" is about. scaling with hyperthreading and multicore, and the good hit rate with common cpu cache hierarchies. (where the "hyperthreading" part tells us something about when the underlying experiments where done)
looks like that article was mostly about that same paper.
bump mapping...
the whole idea of "reviving the ray tracer" seems to be to be less bound by triangle counts and being able to replace the bump maps with actual bumps instead.
> A colleague of mine recently had a brief paper
> (restricted to a maximum of two pages) rejected
> because it was too short - at exactly two pages.
> I kid you not.
How could this surprise anyone?
I don't think anyone would ever claim that the current system was any good at letting the good papers in - but it's job is less to identify all the good papers than it is to identify all the bad ones.
it's like the exact opposite of an email spam filter: with email, a few "nigerians" in the inbox do much less harm than a false positive in the junk folder, while one bad paper more in a conference or a journal would certainly be seen as worse than one good paper less.
having said that, i have to realize that my arguments are not always in line with empiricism.
while it's quite obvious that siggraph is monopolising the conference "market" of it's field much more than the top conferences of other fields do, it's easy to understand why:
the top conferences of other fields might be the most important thing in their field, but they still not known to the general public. most of the people who know that conference will also have heard of the smaller ones. with siggraph things are very different, it's so famous that the number of people who roughly know what siggraph is is much higher than the number of people who have heard of any other cgi science related conference. of course the relevant group, i.e. the scientists will all know the other conferences too, but their relative assessent will undoubtedly be biased by that public visibility of siggraph.
so, how qualified would the average computer graphics scientist be for ethnomusicology? the "broader range of subjects" of academia is off-topic, simply put.
a switch of fields within computer science would be tough enough in itself, but it seems to be much more feasible than to drop into a discipline where he would completely start at zero. in fact, transferring the CGI people's deep understanding of mind boggingly optimized data structures into other fields of computer science could even prove to be a very valuable contribution.
oh, and about that general problem of academia: sure, every field has it's top conferences that are incredibly hard to get into without knowing the right people, but with siggraph it's much more drastic, because of the huge public visibility of it. every apprentice geek knows siggraph, and even your friend the photoshop monkey who thinks unix has been invented by steve jobs when he wrote OS X.
with a famous conference like that, the difference between the one top conference and "the rest" is much bigger than in other fields. for how many of the fields mentioned in your post (cs and non-cs) can you name the top conferences that are hard to get into?
i guess my smac match count is quite high above average, and i would even be surprised if i played any of the other civs as often as i played smac (well, the original civ might be an exception).
if you take out the different weapon and armor itearations - how many classes of units do you really use?
ignoring the civil units (clean formers are a nice touch, but will they really average out to survive long enough to make the investment pay off? oh, and airborne colony pods may be fine, but can't really shine next to icbm colony pods) and the end-game stuff like orbital dropping i rarely get much more than a handful different lines of units, each with numerous iterations going through the tech levels.
which leads us to the major drawback: the techlevel treadmill, upgrading all your lines to the latest developments in the armor/gun range.
the other drawback is that the game can't provide really "fleshed out" names for the units, they end up being just anonymous combinations of attack/defense/chassis/special(/special2).
i want fresh ideas as much as everybody else, but molyneux has failed at least once too often.
give trust (which means funding) to someone new, maybe someone who has actually already succeeded to make anything above-average during this decade.
and in what way is an rts game with an added gimmick or two new anyways?
heh, i never realized you could rise the tax and simply reoccupy the territory, i think i played through the whole game on the minimal budget :)
(i saw the tax option just recently when i did some futile attemts with dos-box or some different version of the windows os)
> And hell! Origin could release Ultima 7 for PC's in 1993.
How many people worked on that game?
We still love those old pixel grids more than the shiny new HDR graphics, but only because they are our childhood memories. give any person is used to todays visual quality and instant gratification a game made like those old ones with which he does not connect any memories of "the good old times" and he would never endure the pains of long text passages, crude menus etc.
today we see the old games as warm and strong on story, but try to remember: back then we did care little about story, we mostly exposed us to those things because they had those really marvellous graphics, computers were the future and generally amazing miracles, they could even play games!
you can't bring that back, even linux can't do that
as you said, he was really good at making early 90ies games. as many people were.
i don't see any clue at all that would indicate that he is good at making post 2k games.
he did have the will to try new things and this is a good thing. but in trial and error you have to accept the errors and that's the problem with molyneux.
in the end i definitely prefer an average shooter over what b&w turned out to be. that one quickly becomes repetive in a boring way (sure the shooter has many repetive elements, but it kind of entertains me while repeating) if you don't accidentally get on the track of that one strategy that will bring you through the level without going through a sweat-mill of epic dimensions.
yeah, multiple competing coop squads operating in a world filled with enough NPCs to make it interesting even if the two (or more) teams would maybe never meet directly in some matches.
why? because it would make the direct meeting experiences feel special and not routine.
and indirect strategies (like blocking the supply of civilian cars to teams needing them or something like that) could make for a very interesting gameplay. gta3-like engine (otoh: a fresh isometric would certainly rock for us oldschoolers), with much smaller worlds to allow for persistent NPCs during matches. sounds like an idea?
I think the unit builder was the worst of alpha centauri (but the rest was so good, i still love it most of all civs, which, uhm, means quite a bit because the others are so good. i especially like the way the factions are presented, the feel so much more real than those civ nations which barely differ more than in the face that presents them)
the problem i had with unit builder: the units do not really evolve over long periods of time, it's still "unarmoured buggy, n attack", with only n changing, same goes for the planes etc. in the end game there are all those really funky extras like drop pods and the new chassis, but imho they come way too late in the game.
btw: civ4 has "learned" the good parts of the unit builder, the promotions don't just make your units stronger but allow for special abilities which give even more interesting combinations than in SMAC, but at the instance level, not at the class level, if you allow the analogy. at the same time it is even less administrative work, because there are no bulk upgrades as in SMAC.