The problem with USENET was the signal-to-noise ratio got worse as the number of users grew.
The first solution was moderation, but this placed too much of a burden on a single volunteer for all but the narrowest topic groups and the most dedicated volunteers.
The brilliant concept that Slashdot introduced (as far as I've been able to determine) was distributed moderation-- a mechanism to distribute this moderation load among more than one person. An approach that was hard to conceive of under NNTP made a lot more sense with a database-backed website.
If you compare the number of postings made to the top 3 most-posted-to newsgroups from the 1995 USENET statistics (which have not, to my knowledge, been updated since), to the size of discussions held on Slashdot, the number of posts per day absorbed by Slashdot had eclipsed anything on 1995 USENET back in 2000 when I last looked into this issue.
I consider "distributed moderation" a huge advance in online community development.
Aw crud, you are quite right. I didn't quite get this when reading the article. Someone mod the parent comment to this one up.
From the patent which I'm scanning through now: "As such, the rotary dial 44 can be continuously rotated by a simple swirling motion of a finger, i.e., the disc can be rotated through 360 degrees of rotation without stopping. Furthermore, the user can rotate the rotary dial 44 tangentially from all sides thus giving it more range of finger positions than that of a traditional scroll wheel as shown in FIG. 1"
It's not radial like Intellivision, it's rotating like an old phone (sorta).
The #2 competitor to the Atari 2600, the Intellivision had a controller with a disc very similar to that described on this patent application (see the picture shown at the above link). The radial dial controller (along with a phone-like keypad and a couple 'action' buttons) was used rather than a joystick or a mouse.
The Intellivision controller is described at the bottom of this page, and the problems with it are aluded to in this video game history, notably that: Unfortunately, the control discs are not a huge hit with players, along with the fact that their flimsy design leads to frequent controller breakdowns. Hardwired right into the system, this becomes a big problem for owners who have to slog the whole machine back to the dealer for repair.
I'd imagine Apple will avoid these mistakes; mice aren't integrated and I don't see why they can't insure higher quality. Personally, I found the disc an acceptable substitute for a joystick after playing with it a bit at a friend's house.
So I think there's a fair bit of prior art. I searched for 5 minutes for Intellivision and Coleco patents and found it described in Patent 4,486,629, 4,470,012, 4,462,594, and 4,439,648. I didn't see that prior art cited in the Apple patent.
That said, the new patent does A) control scrolling actions rather than main-locus-of-control actions, and B) as the patent application says, "pressing down on the disc for clicking does not cause the disc to rotate" which seems like an advance to me over the Intellivision controller.
I guess the question comes down to: how well is the usability testing going?
--LP
P.S. For a Slash-based forum on post-PC UI issues, see Nooface.
Sorry this is only on-topic due to Theo's "oil grab" comment, and I'm know its usually a bad idea to respond to anonymous trolls. But still, I think it needs to be said, if for nothing else than for others to correct me...
If it was an oil grab, an 'informed person' would have to articulate:
why the US would spend $100+ billion to control Iraqi oil revenues that are a twentieth of that annually... surely one could get a higher return elsewhere?
what evidence there is that the U.S. will actually *take* (grab) the oil, rather than leave it for the Iraqis to own and control
explain why the US would rather take oil than just buy it on the open market
under related but alternate theories, acknowledge (or explain why not) why one should be suspicious that US is doing this for oil company contracts, but why that same logic would not apply to French and Russian rationales for opposing the war
explain why the US would act in such an insecure or greedy way when only 10-15% of its current energy usage comes from persian gulf oil (~50% energy usage is oil, 25% of US oil comes from persian gulf)
An 'informed' and fair person would also be willing to acknowledge he was wrong if, 5 years (or whatever) out, the Iraqi's had a functioning democracy and controlled their own oil. Right?
I don't claim to be 'informed'. I don't *know* why the war happened, but the stated reason is pretty decent: old theories of 'containment' don't work when a nuclear-capable state can just slip a nuke to a terrorist and get away with killing millions of people, destroying economies, etc. with a decent chance of not-getting-caught and counter-nuked. With 9/11, it became crystal clear that existing terrorists have the will and the doctrine to do participate in such actions. Nation-states clearly have the will and doctrine to develop nukes. Whether they have the will to pass such material on to terrorists is unclear, but in Iraq's case, the history of invading neighbors, using weapons of mass destruction on Iranian enemies and local Kurds, and a reasonably successful history of deceiving the UN, suggested that the will to proliferate might also be there. That possibility must be stopped.
Theo's anti-war comments in The Globe and Mail can be found here. Theo wasn't told why funding was pulled but he suspects his comments there did it.
I don't think it was Theo's comments to ZDNet on "security through beer drinking" which can be found here.
The "oil grab" comment does strike me as a bit uninformed and polemic, but I'll leave that debate for another time. As an OpenBSD user, I'm sad to see the funding pulled and not happy that someone in the U.S. gov't is being petty. (Or perhaps they're just paranoid?)
If your own brain and the "Weekly World News" origination weren't enough...
Do a search on news.google.com for "Andrew Carlssin" to see if it's "legit news" and you'll get 4 hits, one of which is this rebuttal from the Times Online, a UK paper.
Snopes also looked into this and explains it's false here. When something is too much to believe, wouldn't you check there first?
First, the existence of filesharing technology does not imply that the developer of that technology (Microsoft) knows about a specific act of infringement using said technology (e.g. you copying files over Microsoft's SMB protocol using MS servers and clients). The RIAA case suggests that Daniel Peng does know about such specific acts, in part because he participated in them.
Second, I believe that Xerox or Microsoft would be immune from contributory copyright infringement due for the reasons outlined in the Betamax case, notably the "substantial non-infringing uses" defense. Now in theory, this might apply to Mr. Peng, but as my original post tried to point out, given his own (alleged) direct copyright infringement, he doesn't quite meet the test outlined in, say, this EFF analysis:
The recent court interpretations of the "Betamax defense" have at least two important implications for P2P developers. First, it underscores the threat of vicarious liability--at least in the Ninth Circuit, a court will not be interested in hearing about your "substantial noninfringing uses" if you are accused of vicarious infringement. Accordingly, "control" and "direct financial benefit," as described above, should be given a wide berth.
Likewise, I doubt a court will be much interested in hearing about the "substantial noninfringing uses" of Mr. Peng's SMB indexes if he himself is involved in substantial direct (not even vicarious) infringement.
Third, depending on the type of use Mr. Peng's SMB indexing service received according to whatever evidence would be at trial, it might be possible for the RIAA to show that his indexing did not, in practice, have substantial non-infringing uses. This is why the RIAA's ability to show the "top 20 searches" may be relevant; if all of them were copyrighted songs, it would be hard to successfully argue that Mr. Peng's network had 'substantial' non-infringing uses.
The more I read about this case (Findlaw has the legal papers), the more it looks to me like the RIAA carefully scoped out exactly which 4 campus network cases offered them the strongest cases. But then maybe I'm just being paranoid?
--LinuxParanoid (disclaimer: I'm a coder, not a lawyer)
The analysis points out a number of weaknesses in the RIAA argument.
But there are two strengths of the RIAA charges that the analysis glosses over:
1) Peng is charged with a substantial amount of direct copyright infringement on his own system(s). The liability he may have for these issues could easily force him to settle with the RIAA without ever bringing up the more-questionable "contributory infringement" copyright issues to court.
(That said, the RIAA charges are largely missing details of these charges, focusing instead on the sexier napster-like namecalling. It's not clear whether this implies the direct-infringement evidence is weak, or merely not useful for PR purposes.)
2) It seems to me that the 3 legal requirements the analysis outlines to declare something as 'contributory copyright infringement' should not be too hard for the RIAA to demonstrate: "that an act of direct infringement took place, that the alleged contributor knew about the act, and that the alleged contributor facilitated the act". If all three of these are demonstrated, the question whether the DMCA safe harbor provision comes into play is at issue, but I don't see that as providing Mr. Peng much protection if he knew about specific infringing incidents and did nothing but continued to facilitate them, particularly if he joined in and participated in them. This would clearly be up to a court to rule upon, but it is not an easy win for Mr. Peng.
I think the Barillari analysis demonstrates that someone in Mr. Peng's position might conceivably have a good case for operating a Samba indexing service on campus and not being guilty of contributory copyright infringement. I don't find the analysis particularly compelling that Mr. Peng won't get the axe. That said, I wish Mr. Peng the best.
Princeton eh? Could it be that snarky Ed Felten getting busted there for doing "filesharing research"?
No! He points the finger elsewhere in his Freedom to Tinker blog back in November 2002: the campus paper, the Daily Princetonian then had a nice article with some details on how filesharing works (and is policed, not tightly enough for the RIAA apparently) at Princeton these days.
At least until today.
(I had to laugh at the frankness of the music professor quoted in that article.)
Slashdot Working Group Request for Comments: 3514b Category: Informational
The Duplicate Article Flag in the Posting Queue
Status of this Memo
This memo provides information for the Slashdot community. It does not specify an Slashdot standard of any kind- we don't have standards around here. Distribution of this memo is unlimited.
1. Introduction
Slashdot authors [CBR03], lameness filters, troll detection systems, and the like often have difficulty distinguishing between articles that have are duplicates and those that are merely redundant or stale. The problem is that making such determinations is hard. To solve this problem, we define a duplicate flag, known as the "dupe" bit, in the Slashdot article posting [RFC T4C0] queue. Genuinely novel articles have this bit set to 0; those that are used for boring the reader with redundancy will have the bit set to 1.
The world I've lived in pre-Internet allowed me to, if I found a great newspaper article (or TV show episode) or song, to make a copy of it to pass along to one or a handful of friends to check out.
It certainly seems like this will be technologically unfeasible if/when sufficient copy protection becomes embedded in content-viewing technology in the mid-term future.
I know you aren't a lawyer, but you have parsed these laws carefully and talked to more lawyers than I. Did the type of usage I described above ever fall under 'fair use'? Is it 'fair use' today, and if not, is there a particular piece of legislation that changed the legality of this?
--LP
Re:The real story is tech progress, not Venter...
on
Life Made to Order
·
· Score: 1
Because "inherent complexity" (your term) is not necessarily a sign of design. Maybe, maybe not. Snowflakes are not designed by hand, they are designed by stochastic processes.
That said, a broader answer to your question would be that physicists have found that there are plenty of explanations that one can discover if one avoids presuming a designer is involved in particular phenomena. One day this success may fail, but it seems a little early in our understanding of DNA to conclude it must have been a designer behind. You know, Occam's Razor, named after the theologian whose point was "never multiply the number of miracles required in an explanation beyond what is required." Still, nature is pretty marvelous, snowflakes or DNA, isn't it?
--LP
The real story is tech progress, not Venter...
on
Life Made to Order
·
· Score: 4, Informative
Craig Venter is propounding the vision. But the real science/engineering described in that article seems to be the following:
In mid-2002, researchers at SUNY-Stony Brook synthesized a 7,500-letter long Polio DNA sequence, converted it to RNA, then "combined that RNA with enzymes and other molecules in a test tube, and watched as whole polio viruses assembled spontaneously."
The complicated chemical steps used to synthesize the DNA are error-prone; errors grow linearly with the number of steps "so researchers typically limit fragments to fewer than 80 letters."
The Stony-brook researchers thus took two years.
A company called Egea Biosciences has a prototype machine, the device makes a mistake only once for every 10,000 DNA letters, or bases, a 100-fold improvement over conventional techniques that typically have an error rate of one in 100.
The CEO of that company "says the technology could be extended to yield in a matter of weeks highly accurate strands 100,000 bases in length--long enough to make a very simple bacterial genome."
That's what I got out of the article. And a recognition that there is a loose analogy to semiconductor manufacturing in there. The Venter name is useful mostly for hype as far as I can tell. Actually, setting a vision is really important so I should cut him some slack, but I more appreciated the tech details above which were buried in the middle of the article.
Exactly- putting together your own CD of what you thought were the best Beatles songs would be totally permissible under the "mix rights" concept, and the whole point... with the following restrictions (I'd be happy with either set of restrictions; more happy with A but totally willing to settle for B):
A) you could only distribute the resulting 'best of beetles' to a few people (under 5, or say, under 10), and you could not legally sell the results (a flavor or expansion of 'fair use' rights under copyright)
or
B) you would have to pay for each redistributed mix CD (so I would have to pay RIAA or some entity $17 to distribute one mix CD (or $2/song or whatever) to a friend, but still it could be a mix of *my* choosing) and you could not redistribute more than X amount of material (X CDs or tracks, before whatever you are doing becomes a commercial proposition which requires a special license with the companies involved, and not just generic 'mix rights' for sharing among friends).
It's a fantasy I admit. But no-holds-barred-P2P isn't what I want; "mix sharing" with close friends is what I want. I'm willing to pay for it. But nobody in the music industry is figuring out how to give it to me the flexibility that I (and I suspect millions of consumers) want to share music with my friends.
I like the idea of labeling DRM material. Great idea.
The best kind of choice is 'informed choice', and I hate the thought of those recording labels pulling a fast one on the consumer.
That said, what I really want as 'joe-average-consumer' isn't some infinite-right-to-Napster-share, but what I call "mix rights": the right to compile an album of favorite songs for my own use or for distribution to a few of my friends (I'd even pay for the the redistribution rights.) In the old day, this sort of thing would be considered probably fair use (like clipping and making a few copies of a good news article), but I sense that those freedoms are getting squeezed, legally and technologically, to my great annoyance.
That, and as a consumer I want to be able to hear a song on the radio and have some way to immediately identify the artist/title so I can purchase that song, either later on the spot.
As a consumer, that's all I really want. I've wanted that for 20 years and it annoys the heck out of me that the content industry is too paranoiadly self-centered to deliver it.
Fair enough. While I wouldn't quite consider Alpha NT's adoption of the LP64 programming model as particularly relevant, it is worth pointing out that Microsoft is shipping a 64-bit OS for Intel hardware through OEMs (but not through the retail channel) who offer Itanium/Itanium2 systems. To wit, Windows XP 64-bit for workstations and Windows Advanced Server, Limited Edition 1.2 for servers.
Given that, my comment "how soon your OS supports more than 4 GB of RAM" really ends up being: how soon does Microsoft decide to roll it's 64-bit kernel down into it's mainstream product line? It's probably a bit of a chicken-and-egg argument between Microsoft and Intel... combined with the wildcard of how good/bad the 32-bit application performance is on Itanium2 and how crippling the lack of 16-bit support is.
That said, the technical barriers for Microsoft to extend the already-working 64-bit functionality into consumer-class OSes probably aren't that great compared to the hurdles they've already passed- so I retract my comment about them 'getting their act together' on this score. Making the OS compatible enough for consumers may be a large challenge, but given Microsoft's ability to shift customers to Win95 and later WinXP, I imagine they'll get that right.
Your view of 'when are we ready for 64-bits' largely depends A) on how much money you are willing to spend on RAM and B) how soon your OS supports more than 4 GB of RAM on potential 64-bit hardware (PAE hacks notwithstanding).
If you're willing to spend $200 for RAM in your system, then when 4 GB of RAM is cheaper than $200, you'll basically be wanting a 64-bit system (PAE hacks notwithstanding).
With pricewatch.com showing 1 GB of PC133 SDRAM going for as little as $120, I'd guess that another 4x drop in RAM prices would lead to substantial consumer demand for 64-bit hardware.** And that doesn't even include the demand for 4+GB RAM now in database applications. Whatever the case, this would seem to be earlier than 2007. Unless Microsoft doesn't get its act together (they were pretty late with 32-bit 386 support, IIRC)... which wouldn't be such a bad thing, for Linux at least. But I wouldn't count on that.
--LP
** Yes yes, technically you probably need to spend a bit more to get higher density RAM so that you can fill or exceed 4 GB given the limited number of memory slots available in your system.
Ah, stupid me. The IDC figures were servers only. And I can't find workstation figures, other than these IDC 3Q2002 unit shipment figures for workstations. Still, if the average selling price for "Personal Workstations" (ie Intel-based) is $8k, 70K units for HP would be another $560 million. If Linux is half of that (which would be surprising, an upper bound), that's an extra $280 million which still leaves my conclusions intact, with a missing $1 billion or so.
Would service contracts + openview agents really double the cost of the hardware? Seems unlikely. With Dell, a next-day service contract adds maybe 10-20% or so for a 3-year term in my expereience. But my attempts to reverse-engineer the figures here are getting a bit flimsy, I admit.
Good question; I didn't have a sense of it post-merger, so let's check the 2002 annual report. Skimming that, company wide they had $72 billion in gross revenues, $56.6 billion in net revenues. Broken down, that was:
The PC division of the combined HP/Compaq had net revenues of $22 billion.
The HP Services revenue was $9.1 billion.
HP's Imaging and printing was $20.3 billion.
The Enterprise group (which is all the non-PC business I think: workstation+server+storage) net revenues were $11.4 billion.
$2 billion out of $11.4 billion is 17% of HP's enterprise sales. Not ludicrous, but somehow I didn't think Linux had quite grown that big yet. I'd love for it to be $2 billion but that still seems quite high to me in relation to other figures floating in my memory.
Let me cross-check Google. Ah, here we go. A recent IDC batch of figures (add salt) in TechExtreme say that Linux revenue, industry-wide, was $607 million in the fourth quarter 2002, with conventional UNIX totalling $5 billion. Which extrapolating badly would mean Linux system sales have hit $2.4 billion (its actually less because prior quarters were less- Linux is growing rapidly). HP's share of that isn't outlined there, but it does say IBM is #2 with 20.5% and Dell has 19.5%. Assuming HP is getting, say, 25% of the market, HP's Linux sales would be $600 million. If they were like twice that of IBMs (which seems quite unlikely or HP'd be bragging about it), that's still only $1.2 billion.
So I still don't understand what the heck that $2 billion dollar figure represents.
An HP/Red Hat support partnership is sort of no big deal. It's great to see, but not a surprise.
What left me semi-stunned (until I regained my natural skepticism) was the following sentence:
Today's announcement builds on our $2 billion in Linux-based revenue in 2002 and our decade of commitment to the open source and Linux communities," said Peter Blackmore, executive vice president, HP Enterprise Systems Group. (emphasis mine.)
Where the heck does HP get this figure from? (And if VA Linux couldn't make it in the Linux hardware biz, how come HP is making $2 billion revenues just a couple years later?)
"Sniff test" problems here... but I wouldn't mind being enlightened by someone from HP.
My earlier comments: [Linux is still missing] things like hot-swap memory, hot-swap CPUs, memory failure resiliency (OS quits using memory if recoverable but warning-sign single-bit ECC memory errors get too great), kernel hot-patching, multipath IO, workload management stuff, and ever-more SMP/NUMA scalability.... Sun/HP/IBM were calling their Unix offerings five years ago 'enterprise' without having any of those features (even though the mainframe mostly did.)
Response: I'm afraid you're mistaken. Sun, HP, and IBM have had all of those features for much longer than 5 years
OK, I respect you a lot but I have to differ. I stand by my statement, although I might qualify it just a tad by turning my "without having any of those features" into a "without having nearly any of those features." I went and looked stuff up to check my memory. According to my papers, 6 years ago was AIX 4.2, HP-UX 10.10, Solaris 2.51.
Hot swap CPUs/memory were not out then for any of those vendors (one potential caveat below). 4.5 years ago we had AIX 4.3, HP-UX 11, Solaris 2.6. The first Unix vendor I remember starting to hype that stuff (and they would hype it if they had it, right?) was HP, with HP-UX's memory resilience which showed up in HP-UX 11 in September 1997. HP-UX 11 could also detect certain types CPU failures and notify the admin to shut down the processor, but this was not quite hot swap; it was swap-at-next-reboot. Likewise, IBM had some processor 'resilience' at that time. I think Sun was the first major UNIX vendor to have hot-swap CPUs, with Solaris 2.6 on its Ultra Enterprise 10000-only servers in 1997. I don't remember if the UE10000 ran Solaris 2.5 or not (perhaps you were closer to that as a Solaris admin?) but I really doubt hot-plug CPUs was in there any earlier, "much longer than 5 years" like back in the Solaris 2.4 and earlier days.
Sun also introduced multipathing IO to the high-end UNIX environment with the E10000s, and the other vendors followed within a couple years. HP had a tool called Process Resource Manager for workload management for a long time, maybe even back in 1995; I don't recall when a comparable offering from Sun came out, and I don't think IBM's workload manager came out until 1998 or 1999. In terms of SMP scalability, 5 years ago in late 1996/early 1997 UNIX vendors were showing off 8 and 12-way TPC-C SMP benchmarks and maybe one of them had a 16-way benchmark out, but you didn't see 20, 32 and 64-way benchmarking till later after they had worked a bunch of kinks out.
Anyway, I think saying Sun, IBM and HP Unix offerings had "all those features" for "much longer than 5 years" is just misremembering. I'm open to correction however.
Slashdot Mirror
Nor has it solved the problem of back-row smartasses. But at least now we only see their funny/insightful gags rather than the lame ones, if we want.
--LP
The problem with USENET was the signal-to-noise ratio got worse as the number of users grew.
The first solution was moderation, but this placed too much of a burden on a single volunteer for all but the narrowest topic groups and the most dedicated volunteers.
The brilliant concept that Slashdot introduced (as far as I've been able to determine) was distributed moderation-- a mechanism to distribute this moderation load among more than one person. An approach that was hard to conceive of under NNTP made a lot more sense with a database-backed website.
If you compare the number of postings made to the top 3 most-posted-to newsgroups from the 1995 USENET statistics (which have not, to my knowledge, been updated since), to the size of discussions held on Slashdot, the number of posts per day absorbed by Slashdot had eclipsed anything on 1995 USENET back in 2000 when I last looked into this issue.
I consider "distributed moderation" a huge advance in online community development.
Corrections to my notion of history are welcome.
--LP
Aw crud, you are quite right. I didn't quite get this when reading the article. Someone mod the parent comment to this one up.
From the patent which I'm scanning through now: "As such, the rotary dial 44 can be continuously rotated by a simple swirling motion of a finger, i.e., the disc can be rotated through 360 degrees of rotation without stopping. Furthermore, the user can rotate the rotary dial 44 tangentially from all sides thus giving it more range of finger positions than that of a traditional scroll wheel as shown in FIG. 1"
It's not radial like Intellivision, it's rotating like an old phone (sorta).
--LP
Radial discs for user input are not exactly a new idea; that said, they didn't take off in earlier incarnations.
Anybody remember Intellivision?
The #2 competitor to the Atari 2600, the Intellivision had a controller with a disc very similar to that described on this patent application (see the picture shown at the above link). The radial dial controller (along with a phone-like keypad and a couple 'action' buttons) was used rather than a joystick or a mouse.
The Intellivision controller is described at the bottom of this page, and the problems with it are aluded to in this video game history, notably that:
Unfortunately, the control discs are not a huge hit with players, along with the fact that their flimsy design leads to frequent controller breakdowns. Hardwired right into the system, this becomes a big problem for owners who have to slog the whole machine back to the dealer for repair.
I'd imagine Apple will avoid these mistakes; mice aren't integrated and I don't see why they can't insure higher quality. Personally, I found the disc an acceptable substitute for a joystick after playing with it a bit at a friend's house.
So I think there's a fair bit of prior art. I searched for 5 minutes for Intellivision and Coleco patents and found it described in
Patent 4,486,629, 4,470,012, 4,462,594, and 4,439,648. I didn't see that prior art cited in the Apple patent.
That said, the new patent does A) control scrolling actions rather than main-locus-of-control actions, and B) as the patent application says, "pressing down on the disc for clicking does not cause the disc to rotate" which seems like an advance to me over the Intellivision controller.
I guess the question comes down to: how well is the usability testing going?
--LP
P.S. For a Slash-based forum on post-PC UI issues, see Nooface.
If it was an oil grab, an 'informed person' would have to articulate:
why the US would spend $100+ billion to control Iraqi oil revenues that are a twentieth of that annually... surely one could get a higher return elsewhere?
what evidence there is that the U.S. will actually *take* (grab) the oil, rather than leave it for the Iraqis to own and control
explain why the US would rather take oil than just buy it on the open market
under related but alternate theories, acknowledge (or explain why not) why one should be suspicious that US is doing this for oil company contracts, but why that same logic would not apply to French and Russian rationales for opposing the war
explain why the US would act in such an insecure or greedy way when only 10-15% of its current energy usage comes from persian gulf oil (~50% energy usage is oil, 25% of US oil comes from persian gulf)
An 'informed' and fair person would also be willing to acknowledge he was wrong if, 5 years (or whatever) out, the Iraqi's had a functioning democracy and controlled their own oil. Right?
I don't claim to be 'informed'. I don't *know* why the war happened, but the stated reason is pretty decent: old theories of 'containment' don't work when a nuclear-capable state can just slip a nuke to a terrorist and get away with killing millions of people, destroying economies, etc. with a decent chance of not-getting-caught and counter-nuked. With 9/11, it became crystal clear that existing terrorists have the will and the doctrine to do participate in such actions. Nation-states clearly have the will and doctrine to develop nukes. Whether they have the will to pass such material on to terrorists is unclear, but in Iraq's case, the history of invading neighbors, using weapons of mass destruction on Iranian enemies and local Kurds, and a reasonably successful history of deceiving the UN, suggested that the will to proliferate might also be there. That possibility must be stopped.
--LP
Theo's anti-war comments in The Globe and Mail can be found here. Theo wasn't told why funding was pulled but he suspects his comments there did it.
I don't think it was Theo's comments to ZDNet on "security through beer drinking" which can be found here.
The "oil grab" comment does strike me as a bit uninformed and polemic, but I'll leave that debate for another time. As an OpenBSD user, I'm sad to see the funding pulled and not happy that someone in the U.S. gov't is being petty. (Or perhaps they're just paranoid?)
--LP
If your own brain and the "Weekly World News" origination weren't enough...
Do a search on news.google.com for "Andrew Carlssin" to see if it's "legit news" and you'll get 4 hits, one of which is this rebuttal from the Times Online, a UK paper.
Snopes also looked into this and explains it's false here. When something is too much to believe, wouldn't you check there first?
--LP
First, the existence of filesharing technology does not imply that the developer of that technology (Microsoft) knows about a specific act of infringement using said technology (e.g. you copying files over Microsoft's SMB protocol using MS servers and clients). The RIAA case suggests that Daniel Peng does know about such specific acts, in part because he participated in them.
Second, I believe that Xerox or Microsoft would be immune from contributory copyright infringement due for the reasons outlined in the Betamax case, notably the "substantial non-infringing uses" defense. Now in theory, this might apply to Mr. Peng, but as my original post tried to point out, given his own (alleged) direct copyright infringement, he doesn't quite meet the test outlined in, say, this EFF analysis:
The recent court interpretations of the "Betamax defense" have at least two important implications for P2P developers. First, it underscores the threat of vicarious liability--at least in the Ninth Circuit, a court will not be interested in hearing about your "substantial noninfringing uses" if you are accused of vicarious infringement. Accordingly, "control" and "direct financial benefit," as described above, should be given a wide berth.
Likewise, I doubt a court will be much interested in hearing about the "substantial noninfringing uses" of Mr. Peng's SMB indexes if he himself is involved in substantial direct (not even vicarious) infringement.
Third, depending on the type of use Mr. Peng's SMB indexing service received according to whatever evidence would be at trial, it might be possible for the RIAA to show that his indexing did not, in practice, have substantial non-infringing uses. This is why the RIAA's ability to show the "top 20 searches" may be relevant; if all of them were copyrighted songs, it would be hard to successfully argue that Mr. Peng's network had 'substantial' non-infringing uses.
The more I read about this case (Findlaw has the legal papers), the more it looks to me like the RIAA carefully scoped out exactly which 4 campus network cases offered them the strongest cases. But then maybe I'm just being paranoid?
--LinuxParanoid (disclaimer: I'm a coder, not a lawyer)
The analysis points out a number of weaknesses in the RIAA argument.
But there are two strengths of the RIAA charges that the analysis glosses over:
1) Peng is charged with a substantial amount of direct copyright infringement on his own system(s). The liability he may have for these issues could easily force him to settle with the RIAA without ever bringing up the more-questionable "contributory infringement" copyright issues to court.
(That said, the RIAA charges are largely missing details of these charges, focusing instead on the sexier napster-like namecalling. It's not clear whether this implies the direct-infringement evidence is weak, or merely not useful for PR purposes.)
2) It seems to me that the 3 legal requirements the analysis outlines to declare something as 'contributory copyright infringement' should not be too hard for the RIAA to demonstrate: "that an act of direct infringement took place, that the alleged contributor knew about the act, and that the alleged contributor facilitated the act". If all three of these are demonstrated, the question whether the DMCA safe harbor provision comes into play is at issue, but I don't see that as providing Mr. Peng much protection if he knew about specific infringing incidents and did nothing but continued to facilitate them, particularly if he joined in and participated in them. This would clearly be up to a court to rule upon, but it is not an easy win for Mr. Peng.
I think the Barillari analysis demonstrates that someone in Mr. Peng's position might conceivably have a good case for operating a Samba indexing service on campus and not being guilty of contributory copyright infringement. I don't find the analysis particularly compelling that Mr. Peng won't get the axe. That said, I wish Mr. Peng the best.
--LP
You gotta love the (in)consistency of logic employed by the RIAA.
Is it "These systems are best described as'local area Napster networks,' said Cary Sherman, President, RIAA"
or "This is a particularly flagrant way to illegally distribute millions of copyrighted works over the Internet." ???
Inquiring minds want to know. (hint: "Internet != local area network")
The emphasis on "napster=bad" is so similar to that "Metallica good! Napster Bad!" cartoon I have to chuckle.
--LP
Princeton eh? Could it be that snarky Ed Felten getting busted there for doing "filesharing research"?
No! He points the finger elsewhere in his Freedom to Tinker blog back in November 2002: the campus paper, the Daily Princetonian then had a nice article with some details on how filesharing works (and is policed, not tightly enough for the RIAA apparently) at Princeton these days.
At least until today.
(I had to laugh at the frankness of the music professor quoted in that article.)
--LP
--LP
Professor Felten,
The world I've lived in pre-Internet allowed me to, if I found a great newspaper article (or TV show episode) or song, to make a copy of it to pass along to one or a handful of friends to check out.
It certainly seems like this will be technologically unfeasible if/when sufficient copy protection becomes embedded in content-viewing technology in the mid-term future.
I know you aren't a lawyer, but you have parsed these laws carefully and talked to more lawyers than I. Did the type of usage I described above ever fall under 'fair use'? Is it 'fair use' today, and if not, is there a particular piece of legislation that changed the legality of this?
--LP
Because "inherent complexity" (your term) is not necessarily a sign of design. Maybe, maybe not. Snowflakes are not designed by hand, they are designed by stochastic processes.
That said, a broader answer to your question would be that physicists have found that there are plenty of explanations that one can discover if one avoids presuming a designer is involved in particular phenomena. One day this success may fail, but it seems a little early in our understanding of DNA to conclude it must have been a designer behind. You know, Occam's Razor, named after the theologian whose point was "never multiply the number of miracles required in an explanation beyond what is required." Still, nature is pretty marvelous, snowflakes or DNA, isn't it?
--LP
Craig Venter is propounding the vision. But the real science/engineering described in that article seems to be the following:
In mid-2002, researchers at SUNY-Stony Brook synthesized a 7,500-letter long Polio DNA sequence, converted it to RNA, then "combined that RNA with enzymes and other molecules in a test tube, and watched as whole polio viruses assembled spontaneously."
The complicated chemical steps used to synthesize the DNA are error-prone; errors grow linearly with the number of steps "so researchers typically limit fragments to fewer than 80 letters."
The Stony-brook researchers thus took two years.
A company called Egea Biosciences has a prototype machine, the device makes a mistake only once for every 10,000 DNA letters, or bases, a 100-fold improvement over conventional techniques that typically have an error rate of one in 100.
The CEO of that company "says the technology could be extended to yield in a matter of weeks highly accurate strands 100,000 bases in length--long enough to make a very simple bacterial genome."
That's what I got out of the article. And a recognition that there is a loose analogy to semiconductor manufacturing in there. The Venter name is useful mostly for hype as far as I can tell. Actually, setting a vision is really important so I should cut him some slack, but I more appreciated the tech details above which were buried in the middle of the article.
--LP
The FSF should at least offer to make the company's names on its Patron sponsor list linkable to the companys' websites. It is 2003 you know.
I hate having to go to Google to type in "OEone Corporation" to find out who the heck they are.
--LP
Exactly- putting together your own CD of what you thought were the best Beatles songs would be totally permissible under the "mix rights" concept, and the whole point... with the following restrictions (I'd be happy with either set of restrictions; more happy with A but totally willing to settle for B):
A) you could only distribute the resulting 'best of beetles' to a few people (under 5, or say, under 10), and you could not legally sell the results (a flavor or expansion of 'fair use' rights under copyright)
or
B) you would have to pay for each redistributed mix CD (so I would have to pay RIAA or some entity $17 to distribute one mix CD (or $2/song or whatever) to a friend, but still it could be a mix of *my* choosing) and you could not redistribute more than X amount of material (X CDs or tracks, before whatever you are doing becomes a commercial proposition which requires a special license with the companies involved, and not just generic 'mix rights' for sharing among friends).
It's a fantasy I admit. But no-holds-barred-P2P isn't what I want; "mix sharing" with close friends is what I want. I'm willing to pay for it. But nobody in the music industry is figuring out how to give it to me the flexibility that I (and I suspect millions of consumers) want to share music with my friends.
I like the idea of labeling DRM material. Great idea.
The best kind of choice is 'informed choice', and I hate the thought of those recording labels pulling a fast one on the consumer.
That said, what I really want as 'joe-average-consumer' isn't some infinite-right-to-Napster-share, but what I call "mix rights": the right to compile an album of favorite songs for my own use or for distribution to a few of my friends (I'd even pay for the the redistribution rights.) In the old day, this sort of thing would be considered probably fair use (like clipping and making a few copies of a good news article), but I sense that those freedoms are getting squeezed, legally and technologically, to my great annoyance.
That, and as a consumer I want to be able to hear a song on the radio and have some way to immediately identify the artist/title so I can purchase that song, either later on the spot.
As a consumer, that's all I really want. I've wanted that for 20 years and it annoys the heck out of me that the content industry is too paranoiadly self-centered to deliver it.
--LP
Fair enough. While I wouldn't quite consider Alpha NT's adoption of the LP64 programming model as particularly relevant, it is worth pointing out that Microsoft is shipping a 64-bit OS for Intel hardware through OEMs (but not through the retail channel) who offer Itanium/Itanium2 systems. To wit, Windows XP 64-bit for workstations and Windows Advanced Server, Limited Edition 1.2 for servers.
... combined with the wildcard of how good/bad the 32-bit application performance is on Itanium2 and how crippling the lack of 16-bit support is.
Given that, my comment "how soon your OS supports more than 4 GB of RAM" really ends up being: how soon does Microsoft decide to roll it's 64-bit kernel down into it's mainstream product line? It's probably a bit of a chicken-and-egg argument between Microsoft and Intel
That said, the technical barriers for Microsoft to extend the already-working 64-bit functionality into consumer-class OSes probably aren't that great compared to the hurdles they've already passed- so I retract my comment about them 'getting their act together' on this score. Making the OS compatible enough for consumers may be a large challenge, but given Microsoft's ability to shift customers to Win95 and later WinXP, I imagine they'll get that right.
--LP
Your view of 'when are we ready for 64-bits' largely depends A) on how much money you are willing to spend on RAM and B) how soon your OS supports more than 4 GB of RAM on potential 64-bit hardware (PAE hacks notwithstanding).
If you're willing to spend $200 for RAM in your system, then when 4 GB of RAM is cheaper than $200, you'll basically be wanting a 64-bit system (PAE hacks notwithstanding).
With pricewatch.com showing 1 GB of PC133 SDRAM going for as little as $120, I'd guess that another 4x drop in RAM prices would lead to substantial consumer demand for 64-bit hardware.** And that doesn't even include the demand for 4+GB RAM now in database applications. Whatever the case, this would seem to be earlier than 2007. Unless Microsoft doesn't get its act together (they were pretty late with 32-bit 386 support, IIRC)... which wouldn't be such a bad thing, for Linux at least. But I wouldn't count on that.
--LP
** Yes yes, technically you probably need to spend a bit more to get higher density RAM so that you can fill or exceed 4 GB given the limited number of memory slots available in your system.
Ah, stupid me. The IDC figures were servers only. And I can't find workstation figures, other than these IDC 3Q2002 unit shipment figures for workstations. Still, if the average selling price for "Personal Workstations" (ie Intel-based) is $8k, 70K units for HP would be another $560 million. If Linux is half of that (which would be surprising, an upper bound), that's an extra $280 million which still leaves my conclusions intact, with a missing $1 billion or so.
Would service contracts + openview agents really double the cost of the hardware? Seems unlikely. With Dell, a next-day service contract adds maybe 10-20% or so for a 3-year term in my expereience. But my attempts to reverse-engineer the figures here are getting a bit flimsy, I admit.
--LP
Yemen blocks sex websites? Shocker. Now if only they could block sex spam, maybe it'd be worth moving there...
--LP (j/k)
Good question; I didn't have a sense of it post-merger, so let's check the 2002 annual report. Skimming that, company wide they had $72 billion in gross revenues, $56.6 billion in net revenues. Broken down, that was:
The PC division of the combined HP/Compaq had net revenues of $22 billion.
The HP Services revenue was $9.1 billion.
HP's Imaging and printing was $20.3 billion.
The Enterprise group (which is all the non-PC business I think: workstation+server+storage) net revenues were $11.4 billion.
$2 billion out of $11.4 billion is 17% of HP's enterprise sales. Not ludicrous, but somehow I didn't think Linux had quite grown that big yet. I'd love for it to be $2 billion but that still seems quite high to me in relation to other figures floating in my memory.
Let me cross-check Google. Ah, here we go. A recent IDC batch of figures (add salt) in TechExtreme say that Linux revenue, industry-wide, was $607 million in the fourth quarter 2002, with conventional UNIX totalling $5 billion. Which extrapolating badly would mean Linux system sales have hit $2.4 billion (its actually less because prior quarters were less- Linux is growing rapidly). HP's share of that isn't outlined there, but it does say IBM is #2 with 20.5% and Dell has 19.5%. Assuming HP is getting, say, 25% of the market, HP's Linux sales would be $600 million. If they were like twice that of IBMs (which seems quite unlikely or HP'd be bragging about it), that's still only $1.2 billion.
So I still don't understand what the heck that $2 billion dollar figure represents.
--LP
An HP/Red Hat support partnership is sort of no big deal. It's great to see, but not a surprise.
What left me semi-stunned (until I regained my natural skepticism) was the following sentence:
Today's announcement builds on our $2 billion in Linux-based revenue in 2002 and our decade of commitment to the open source and Linux communities," said Peter Blackmore, executive vice president, HP Enterprise Systems Group. (emphasis mine.)
Where the heck does HP get this figure from? (And if VA Linux couldn't make it in the Linux hardware biz, how come HP is making $2 billion revenues just a couple years later?)
"Sniff test" problems here... but I wouldn't mind being enlightened by someone from HP.
--LP
My earlier comments: [Linux is still missing] things like hot-swap memory, hot-swap CPUs, memory failure resiliency (OS quits using memory if recoverable but warning-sign single-bit ECC memory errors get too great), kernel hot-patching, multipath IO, workload management stuff, and ever-more SMP/NUMA scalability. ... Sun/HP/IBM were calling their Unix offerings five years ago 'enterprise' without having any of those features (even though the mainframe mostly did.)
Response: I'm afraid you're mistaken. Sun, HP, and IBM have had all of those features for much longer than 5 years
OK, I respect you a lot but I have to differ. I stand by my statement, although I might qualify it just a tad by turning my "without having any of those features" into a "without having nearly any of those features." I went and looked stuff up to check my memory. According to my papers, 6 years ago was AIX 4.2, HP-UX 10.10, Solaris 2.51.
Hot swap CPUs/memory were not out then for any of those vendors (one potential caveat below). 4.5 years ago we had AIX 4.3, HP-UX 11, Solaris 2.6. The first Unix vendor I remember starting to hype that stuff (and they would hype it if they had it, right?) was HP, with HP-UX's memory resilience which showed up in HP-UX 11 in September 1997. HP-UX 11 could also detect certain types CPU failures and notify the admin to shut down the processor, but this was not quite hot swap; it was swap-at-next-reboot. Likewise, IBM had some processor 'resilience' at that time. I think Sun was the first major UNIX vendor to have hot-swap CPUs, with Solaris 2.6 on its Ultra Enterprise 10000-only servers in 1997. I don't remember if the UE10000 ran Solaris 2.5 or not (perhaps you were closer to that as a Solaris admin?) but I really doubt hot-plug CPUs was in there any earlier, "much longer than 5 years" like back in the Solaris 2.4 and earlier days.
Sun also introduced multipathing IO to the high-end UNIX environment with the E10000s, and the other vendors followed within a couple years. HP had a tool called Process Resource Manager for workload management for a long time, maybe even back in 1995; I don't recall when a comparable offering from Sun came out, and I don't think IBM's workload manager came out until 1998 or 1999. In terms of SMP scalability, 5 years ago in late 1996/early 1997 UNIX vendors were showing off 8 and 12-way TPC-C SMP benchmarks and maybe one of them had a 16-way benchmark out, but you didn't see 20, 32 and 64-way benchmarking till later after they had worked a bunch of kinks out.
Anyway, I think saying Sun, IBM and HP Unix offerings had "all those features" for "much longer than 5 years" is just misremembering. I'm open to correction however.
--LP