The Samba project is concerned that if they look at MS API documentation (under NDA) then they will be unable to write any code that depends on it. The reason is that such source code would inevitably contain information from the API documents, and hence break the NDA. For example if a list of error codes in the SMB protocol documentation is rendered as a list of definitions in an include file in Samba, it must either be a duplicate of the documented one or be wrong.
Does anyone know if this is actually true? Any case law or legal opinions? So far I've seen sensible-looking concerns from non-lawyers but no real legal evidence. The judge seems to imply that disclosure according to the Judgement will allow any development model, but maybe she missed this point.
I was one of the people who sent in comments under the Tunny act. You can find them in the list of 47 selected comments. In it I wrote:
Microsoft must not be allowed to pretend that these interface descriptions are trade secrets, as it tried to do with its extension to Kerberos. Because OSS packages include the full source code they inevitabley reveal the full details of their operation to any programmer who downloads them. If Microsoft can claim trade secret status on an interface it can effectively block any OSS package from using that interface, since to do so would reveal the "secret" of its operation. This appears to have been the objective of the click-through license on the Kerberos extensions (see above). The "Samba" project (www.samba.org) has reverse-engineered the Microsoft file and printer sharing protocols, allowing non-Microsoft systems to gain access to resources on Microsoft systems. An updated version of Samba for Windows 2000 is being prepared which will need to inter-operate with the Windows 2000 Kerberos extensions. If these extensions are considered trade secrets then it would be impossible for the Samba project to work with these extensions, and a key component in any mix of Microsoft and non-Microsoft computers would be crippled.
I also tackled the issues of cost (e.g. subscription fees) and protocol patents.
The amount of homework required depends on the field.
In pharmacuticals this approach is feasible. A medical patent would have to mention both the disease and the treatment, and both of these have fairly well defined keywords to describe them.
At the other extreme, software patents are essentially unsearchable. A patentable idea can often be expressed in a few lines of code, so a 100,000 line program has many thousands of ideas in it, any one of which might already be patented. Worse yet there is no system for categorising and naming these ideas, so if I want to do a patent search on even one of these ideas I have only the vaguest guess as to what keywords to look for.
So the only thing to do is to put your product on the market and pray that someone doesn't pop up and say "You stepped on my patent: cease and desist selling your product immediately".
Intel faces a very similar problem: just substitute "simple logic circuits" for "lines of code" in the above. In fact chip designers these days do actually design chips in a programming language called VHDL, so the cases are exactly parallel.
Maybe a serious attempt to sort out the patent system could bring Intel on board at this point, and possibly a lot of other companies too.
Intel no doubt has a lot of patents that it regards as valuable, so it won't support a proposal to do away with patents all together. But a proposal to reduce the "landmine" effect of unknown patents might win support in the technology industry, because landmine patents are a nightmare to everyone there.
So, how do you prevent landmine patents but still keep the basic concept of a patent? Maybe the time has come for a "sweat of the brow" basis for patents. At present patents are granted on the "lightbulb moment" theory: the inventor has a flash of inspiration and the invention springs fully formed from his brow. But this leads to silly stuff like the infamous XOR patent (which patented the use of XOR for screen cursors in GUIs). However the justification for patents lies in the investment required to bring an idea to market. So maybe patents should be granted based on evidence of the hard work required to generate the invention.
You may examine the software in order to understand its operation, interfaces, and any interaction with users. However you may not incorporate information gained from such examination into any software unless that software is also licensed under the GPL.
Of course this assumes that such a clause would be legally enforceable. The reverse engineering clauses in licenses are based on the theory that the operation of the software is a trade secret and can therefore be protected. Free software eschews the idea of trade secrets. But under the theory that "freedom of contract" is sovreign, and supposing that the GPL is a contract freely entered into, then I'd guess that such a term would be as legally enforceable as any other.
Hmmm. What about open source licenses that are not GPL. This would prohibit, say, BSD picking up a feature from Linux. The plurality of open source licenses would become a major obstacle. Maybe the OSF is going to have to formally recognise other Open Source licenses as fellow travellers. OTOH maybe RMS will stay with his usual doctrinaire position on the subject.
Reading through this, I've seen some responses along the lines of "Build it and we will come". Don't believe it.
I spent some years of my life prosetylising Eiffel. It was an OO language done right, far better than C++, and considerably better than Java. Everybody listened politely, but the replies always started "Yes, but here in the real world...", and then they'd explain why nobody is ever going to adopt a minority language.
Then Python happened. Why Python and not Eiffel? I'm not sure. But I can get hired to program in Python. I never could for Eiffel. Hmmm. Build it and they might come.
So you need to talk to the marketeers. I've done courses on marketing. Thats not selling, thats marketing: the two are different. And I have to tell you that the hacker disdain for marketroids is misplaced. These guys do know what they are talking about, and they have a number of really useful tools for working out just what is going to sell your product and what is irrelevant chaff. What they don't generally understand is the hacker mindset. Thats where you come in. Talk to your marketeers. Help them understand the target market and how its members think. Put the two together and you will have something.
You cannot [measure] the polarization of a wave of light with out changing it.
Thats the theory behind quantum encryption, in which single photons are used to create a shared key by playing tricks with polarisation. The important point is the words "single photon".
However QE cannot work over long distances because photons get lost (i.e. attenuation). General purpose signalling sends a lot of photons so that at least a few get through (I think the detection level for general purpose detectors without special cooling is around 70 photons). They also get amplified. I'm not sure if fibre amplifiers maintain polarisation. If not then this technique is just an interesting novelty.
So tapping would be easy. Just put the signal through a splitter (e.g. a bend in the fibre) and route your half of the signal to a decoder that works in the same way as the official one. The other end sees a 3dB drop in signal, but thats probably too small to be noticed.
Where this might be important is increased bandwidth. At the moment fibre transmission uses binary keying: send photons for 1, no photons for 0. Polarisation modulation means that you could use several different angles, and hence encode more than one bit per light pulse.
But don't get too excited about the bandwidth either. The limiting factor on bandwidth at the moment is the routers at the end of the fibre. We can pump terabits down a fibre in the lab, and 100 Gbit is pretty straighforward to do in the field. But put ten 100Gbit links into a router and you have to have a machine that can switch 1 Tbit. If the average packet is 1.5kbytes (Ethernet frame) then thats around 83 million packets per second. Even with hardware assist thats an awful lot of address table lookups per second.
It's now possible to make reasonable quality diamonds in presses. They are having some problems with yellow inclusions, but these will probably be overcome in the future (and anyway its probably possible to cut engagement-ring sized stones to avoid these inclusions).
De-Beers and co are trying to aim for trademarks to identify "real" diamonds (i.e. dug out of the dirt). But I don't think it will really play in the long run.
Try asking if she would prefer an expensive honeymoon instead. That will also be a good reply to obnoxious women shoving huge rocks under her nose and saying "where's yours then?". You/she can just say "we decided to spend the money on a 5 star hotel in [Distant place]" and then gush about how wonderful it all was. Going to interesting places beats shiny rocks every time.
This won't happen. Market forces have already sorted out the way that ISPs pay each other, and the Indian ISPs are swimming against the tide.
The food-chain in ISPs looks something like this: Customer -> Tier 3 -> Tier 2 -> Tier 1, with each level paying the layer above for access. Tier 1 ISPs are people like UUNet with global reach. Tier 2 are national or "regional" (e.g. EU, Americas, Asia-Pacific). Tier 3 are local ISPs, and customers are both individual users and hosting companies.
Actually there is nothing stopping a customer or Tier 3 ISP from signing up with a Tier 1 ISP, and many do. But the principle is the same.
There are two kinds of link an ISP can have to other ISPs: Transit and Peering. In a transit link an ISP pays a larger ISP for access to "the Internet". In other words the smaller ISP can route packets through the link to any destination and expect to receive replies via the same route. In a peering relationship two ISPs, usually in the same Tier, agree to exchange traffic, usually without payment, but with the proviso that only traffic for customers of the other ISP is to be routed through that link. You can't send traffic to B through your peering link with A (although there are sometimes mutual backup link terms in the agreement).
You can think of this in your own terms quite easily. You have a transit link with your ISP that you pay for. But if you and your neighbour exchange a lot of traffic you might string an Ethernet cable between your houses and create your own peer link. But it would be very bad manners to use that link to pinch bandwidth off your neighbour.
The market forces that created this system are very straightforward. Originally the Internet worked with free transit links, but then the people investing in global networks realised that all the smaller ISPs were getting a free ride, and so they started demanding payment. This happened around 1996-7, and you can find lots of discussion papers from that time worrying about "the balkanisation of the Internet". In fact nothing of the sort happened. Metcalfe's law saw to it that everyone found more value from being connected to an unbalkanised Internet, and the net effect (sorry) was that money flowed from you and me up to Worldcom, and much good it did them. Meanwhile the smaller ISPs found that peering arrangements helped them to cut their costs because peer traffic avoided the expensive transit routes.
Thats not to say that things are so simple in real life. Peering arrangements in particular are fraught with difficulty because it usually means negotiating with your direct competitors, and you can play all sorts of dirty tricks like "hot potato" routing (routing packets to your nearest exit point instead of the globally most efficient one). But thats the general idea.
Incidentally the economics work like this regardless of the direction of most of the bits. People who tried to analyse the Internet using telephone economics got this wrong, because with the phone its usually the caller who pays. On the Internet the "caller" is hard to identify and the rules for doing so keep changing. And in any case the issue is irrelevant. You have content providers who want to reach readers and readers who want to access content. (Peer to peer changes the numbers and locations, but not the fundamentals). Both pay ISPs to provide this service, and those ISPs then pay the next tier up, and so on.
So now we look at India, where a bunch of Tier 3 and 2 ISPs are demanding payment from Tier 1 ISPs. The Tier 1 ISPs will rightly tell them to get lost.
I suppose that the Indian ISPs (who are mostly consumer ISPs) might demand payment from content providers such as Yahoo, Slashdot and co, on the grounds that the content providers want to reach Indian eyeballs. But I don't see this flying either. Those Indian eyeballs want the content just as much as the providers want to provide it, which is why you get no-payment peering arrangements between content providers and consumers: its the flow of value that counts, not the flow of bits.
These days people can't even trust audited accounts, and BSA piracy numbers are just unaudited guesswork.
Politicians see a lot of pressure groups with axes to grind and corporate backing to pay their bills, and they know exactly how much to trust their numbers.
Of course the BSA doesn't exactly rely on its numbers to make its case. They rely on other numbers in campaign contributions as well. But thats a whole 'nother ball game.
The "Claims" section of a patent is the most important bit. It lays out what the patent covers. It consists of a series of numbered items, each of which covers one idea or variation. So far so simple.
The key thing is that a court might in future decide that some claims are valid but others are not. So the first couple of claims in a patent might well lay claim to the entire state of the art, and might only be there as a kind of #define macro for subsequent claims. I once read an encryption patent (ISTR it was for a DVD system that didn't get used) where Claim 1 was for XORing the output of a random number generator with the cleartext. This was followed by a series of claims that started "A system as in Claim 1 where the random number generator is...".
So when you see a patent that seems to claim the whole of some technology, don't panic. There is going to be tons of prior art. You just have to work out where the prior art ends and the real invention starts. This is going to be a bit grey on the boundary (thats where patent lawyers make their money), but you can still get a fairly clear idea pretty quickly. You can also get a fair idea just by looking at the claims and thinking about the technology they represent. Once you get to precise descriptions of obscure algorithms then you are into the meat of the patent.
Incidentally, don't be scared of legalese. Just think of it as an unusually verbose and unstructured programming language.
I have seen someone take a hammer to a disk platter. This was a drive that had super secret military stuff on it, and procedures said that the disk had to be physically destroyed. The IT department settled for taking the platters out of the drive and then making a half dozen dents on each side with a hammer and centrepunch.
The platters rang like a bell when hit, even when sitting on the carpet. Get a set of platters from different drives together and you could do the Anvil Chorus.
Careful here. I once hooked up an NTP server to a GPS receiver (makers name omitted because I don't recall who made it) and found a 13 second error. Obviously the makers test suite hadn't included a check that the time it gave was right.
I note that on the page you are pointing to, the microsecond accuracy is described as "nominal". In other words, it isn't actually that accurate.
Where the hell do you think the criminal element is going to go? They're going to invest their now illegal drug profits in the legal drugs of the future.
No they won't, for the simple reason that making and distributing legal drugs is not their core competence. Smuggling and distributing contraband is their core competence. Once their product is no longer illegal there are lots of companies with the knowledge and infrastructure. Purity requirements alone are enough to put the illegal dealers out of business because only drugs made to sufficient purity will be legal to sell, and underground labs can't come anywhere near meeting those requirements.
If drugs are made legal then the illegal dealers will be out of business. They have a huge infrastructure dedicated to shipping those drugs into the country, and it will no longer have any value. I don't know of any product other than illgeal drugs that could use that infrastructure, so they can't switch to any other product.
There are two kinds of relationship in the ISP world: peering and transit.
In a transit relationship a larger ISP provides a smaller one with long-distance connectivity for a fee, in exactly the same way that you pay your ISP. So local-ISP pays, say, UUNET for a connection that gives it access to the rest of the Internet. Such connections have a maximum bandwidth and typically include a per-megabyte charge element as well.
Peering relationships are used between ISPs of similar size to reduce the costs of transit. Two ISPs will agree to exchange traffic, but only where the source and destination are within those two ISPs. A cannot send traffic to C via its peering arrangement with B.
However all of this is irrelevant. When service is stolen the cost of the service is taken as its retail cost, not its incremental cost to the provider. Otherwise people stealing cable TV could argue that they have done nothing wrong because they didn't increase anyone's incremental costs.
The Hacker Crackdown by Bruce Sterling contains a description of the general process, although he was writing about the Secret Service rather than the FBI. In particluar,
Standard computer crime search warrants, which date back to the early 80s, use a sweeping language that targets computers, most anything attached to a computer, most anything used to operate a computer - most anything that remotely resembles a computer - plus most any and all written documents surrounding it. Computer crime investigators have strongly urged agents to seize the works.
Elsewhere it talks about the actual process of seizure. They do take photographs of the configuration before disturbing anything. However this is for their benefit, not yours. You are assumed to know how to reassemble your own kit.
(Aside: I recall a case in Guernsey where a new sports car was bought in to the island. The Customs disassembled it in a search for contraband and didn't find anything. Then they told the owner that he could take it away. Not only did he have to pay to have it put back together, but the warranty was now void. Neither of these two things was considered "damage" worthy of compensation).
On passwords, I'm not sure about the US. I suspect your Fifth Amendment protects you. In the UK the Regulation of Investigative Practices Act authorises the police to demand your passwords and encryption keys on pain of two years imprisonment for failure to comply, and if you tell anyone other than your lawyer about it then you can be put in prison for five years.
The Hacker Crackdown is probably the best book on computer cracking I've read, even though it was written over 10 years ago. It looks at the subject from the POV of the crackers, the cops and the civil libertarians. If you are interested in the subject then read it.
I don't have a problem with the idea that uncapping your cable modem is theft of service, or with the idea that the perpetrators should be prosecuted.
Where I do have a problem is with what actually happened and the lack of due process associated. Look at the sequence of events:
The ISP notices the uncapped modem (I gather they use SNMP to ask the modem what its set to: nothing sophisticated).
The ISP calls the FBI and alleges that this crime has cost it over $10,000. Hmmm. Where did that number come from? I'm on a 512kbit service for £25/month. Suppose I uncapped my modem to get the theoretical maximum of 64Mbits (the full channel bandwidth that is shared between all users on a spur). That is in theory a 128-fold increase in service, so I should be paying £3,200 per month, or around $5,000. So that may be two months service at 64Mbits. Maybe not too unreasonable, although I don't know how they estimated the time.
The FBI get a search warrant based on the ISP's complaint and seize computers. This is perfectly legal: the authorities are permitted to seize the "instrumentality of the crime". If a PC was used to uncap the modem then it is an instrumentality of the crime. Also, if the case came to court then the defence could ask what evidence the prosecution had that the supposed perpetrators were actually responsible. Maybe it was a prankster thinking to do a "favour". Any prosecution is going to need smoking-gun scripts found on the suspect's PCs.
No charges are filed. Despite what I said just now, the whole thing is never tested in court. Confiscation of the computers (and any private data thereon) is considered enough of a punishment, and doesn't require the expense of a trial.
All of this is perfectly reasonable and legal, but it is never the less an end-run around the due-process principle. Based on a complaint and a search warrant your property can be effectively confiscated, and you have almost no come-back. Of course in theory you can sue for the return of your property, but all the police have to do is claim an "ongoing investigation" to make the suit fail.
A key to good performance is balancing the amount you spend on different parts of the system. There is no point in buying the latest 2GHz CPU if you then team it up with 64Mb of PC100 RAM and a hard disk drive with a 33MHz IDE interface.
However many PC manufacturers do that kind of thing, especially at the budget end. Then they can claim the machine is "fast" without having to spend lots of money.
DIY lets you balance the performance of the various components to meet your requirements instead of the marketing dept.
Biotech is increasingly an IT-intensive industry. Some of the biggest iron in the world exists to crunch biotech problems, mostly related to protein folding and drug interactions. They also generate huge amounts of experimental data that has to be managed and mined. Finally there is a lot of automated lab equipment for parallelising those bits of it that still involve real chemicals and real biology.
The field here is wide open. Lots of university biology departments are spinning off companies to make innovative new sensors, so you can get involved there. Or you can go and manage a Beowulf cluster for a big drug firm. Or anything in between.
So what should I be earning in my current job? I can check the salary offers on Jobserve, and assuming that they are not too inflated by fictional jobs I can get some idea. But that just says what the going rate for my job title is. It doesn't say what I personally am worth to the company.
Company personnel departments can subscribe to salary surveys that are a slightly more accurate version of what I can find on Jobserve. Same problem.
From the point of view of the company, the value of an employee is the money they will earn if they hire him/er minus the money they will make if they don't, and also minus the costs of employing that person (about 50% of which is the salary, and the other 50% is desk, office space, electricity bills, computer, 10% of the cost of a manager, etc).
Any sum involving a big approximate number minus another big approximate number is going to have a huge error bar on the answer. For a software engineer the error bar probably exceeds the salary.
So managers, who perforce must make decisions regardless of the absence of evidence, fall back on rules of thumb. The first rule is, pay the going rate for that job title, hence those salary surveys. The second rule of thumb is that staff turnover is expensive. Every time you hire someone you commit money to teaching them the ropes until they become productive (whenever that is, see above), and also take a risk that this investment will be written off if they turn out to be bozos, or just up and leave in six months.
So, are you worth another 50% to your current employer? Possibly. Your management guesses so, but they might be wrong.
The hardest bit has already been done for you thanks to the disaster: management support. This (believe it or not) is what managers are for.
First, write a policy for users to follow (not more than a page), get your boss to sign it, and then distribute copies. This tells everyone that the boss is behind this. If your boss does not have line authority over the people in question then get someone who does.
Second, get your boss's approval for a half hour tutorial for all staff on the subject, attendance mandatory for all users including him/herself. Get the boss to start with a brief repeat of the disaster story, then hand over to you (write the boss a script if necessary). Tell people why they need to do it and what it is they need to do, but obviously don't go into techie detail. Also emphasise that unsafe behaviour is letting the team down: its not just your work at risk, its everyones.
Its your responsibility to determine policy, configure machines, tell people what to do, monitor progress, and report to your boss. This can and should include saying that certain users are refusing to following departmental policy. Its then his/her job to take things further, upt to and including disciplinary action if necessary. Its not likely to be necessary: few people are that boneheaded.
Good luck. Culture change is hard, but its one of the most valuable things you can do.
The problem with a binding promise not to pursue open source projects is the risk of someone using the promise to evade the patent protection.
Suppose that Red Hat gains a patent on a new voice compression algorithm, and United Megacorp produces a voice conferencing product that uses this algorithm. UM can release its product with modular plugins for the codecs, and then just release the patented codec as OSS. Hardly a big win for the open source side.
Having said that, not all patents could be subverted so easily. The codec patent could be evaded, but a patent on using the codec within a voice conferencing application would not. At that point (AIUI) UM would have to release the entire conferencing system as OSS in order to fit inside the RH promise conditions.
Sounds like Farnsworth suffered from some mild form of Autism. See this article for more details, but "idiot savant" talents like perfect drawings in childhood and a later inability to understand what the world is doing sound very much like it.
If so then Farnsworth is a bad example to use for the main point of this essay. Farnsworth's problems are as likely to stem from autism as anything else. Other lone geniuses have managed to create major inventions and use the patent system to do so. The Bell telephone is the obvious example.
Slashdot Mirror
Does anyone know if this is actually true? Any case law or legal opinions? So far I've seen sensible-looking concerns from non-lawyers but no real legal evidence. The judge seems to imply that disclosure according to the Judgement will allow any development model, but maybe she missed this point.
Paul.
I also tackled the issues of cost (e.g. subscription fees) and protocol patents.
Hmmm. It seems that I was right to be worried.
Paul.
I've seen it demonstrated by Jack Cohen (science writer and reproductive biologist). Its cool.
Do a google search to find a number of recipies.
Paul.
In pharmacuticals this approach is feasible. A medical patent would have to mention both the disease and the treatment, and both of these have fairly well defined keywords to describe them.
At the other extreme, software patents are essentially unsearchable. A patentable idea can often be expressed in a few lines of code, so a 100,000 line program has many thousands of ideas in it, any one of which might already be patented. Worse yet there is no system for categorising and naming these ideas, so if I want to do a patent search on even one of these ideas I have only the vaguest guess as to what keywords to look for.
So the only thing to do is to put your product on the market and pray that someone doesn't pop up and say "You stepped on my patent: cease and desist selling your product immediately".
Intel faces a very similar problem: just substitute "simple logic circuits" for "lines of code" in the above. In fact chip designers these days do actually design chips in a programming language called VHDL, so the cases are exactly parallel.
Paul.
Intel no doubt has a lot of patents that it regards as valuable, so it won't support a proposal to do away with patents all together. But a proposal to reduce the "landmine" effect of unknown patents might win support in the technology industry, because landmine patents are a nightmare to everyone there.
So, how do you prevent landmine patents but still keep the basic concept of a patent? Maybe the time has come for a "sweat of the brow" basis for patents. At present patents are granted on the "lightbulb moment" theory: the inventor has a flash of inspiration and the invention springs fully formed from his brow. But this leads to silly stuff like the infamous XOR patent (which patented the use of XOR for screen cursors in GUIs). However the justification for patents lies in the investment required to bring an idea to market. So maybe patents should be granted based on evidence of the hard work required to generate the invention.
What do you think?
Paul.
You may examine the software in order to understand its operation, interfaces, and any interaction with users. However you may not incorporate information gained from such examination into any software unless that software is also licensed under the GPL.
Of course this assumes that such a clause would be legally enforceable. The reverse engineering clauses in licenses are based on the theory that the operation of the software is a trade secret and can therefore be protected. Free software eschews the idea of trade secrets. But under the theory that "freedom of contract" is sovreign, and supposing that the GPL is a contract freely entered into, then I'd guess that such a term would be as legally enforceable as any other.
Hmmm. What about open source licenses that are not GPL. This would prohibit, say, BSD picking up a feature from Linux. The plurality of open source licenses would become a major obstacle. Maybe the OSF is going to have to formally recognise other Open Source licenses as fellow travellers. OTOH maybe RMS will stay with his usual doctrinaire position on the subject.
Paul.
I spent some years of my life prosetylising Eiffel. It was an OO language done right, far better than C++, and considerably better than Java. Everybody listened politely, but the replies always started "Yes, but here in the real world...", and then they'd explain why nobody is ever going to adopt a minority language.
Then Python happened. Why Python and not Eiffel? I'm not sure. But I can get hired to program in Python. I never could for Eiffel. Hmmm. Build it and they might come.
So you need to talk to the marketeers. I've done courses on marketing. Thats not selling, thats marketing: the two are different. And I have to tell you that the hacker disdain for marketroids is misplaced. These guys do know what they are talking about, and they have a number of really useful tools for working out just what is going to sell your product and what is irrelevant chaff. What they don't generally understand is the hacker mindset. Thats where you come in. Talk to your marketeers. Help them understand the target market and how its members think. Put the two together and you will have something.
Good luck.
Paul.
Thats the theory behind quantum encryption, in which single photons are used to create a shared key by playing tricks with polarisation. The important point is the words "single photon".
However QE cannot work over long distances because photons get lost (i.e. attenuation). General purpose signalling sends a lot of photons so that at least a few get through (I think the detection level for general purpose detectors without special cooling is around 70 photons). They also get amplified. I'm not sure if fibre amplifiers maintain polarisation. If not then this technique is just an interesting novelty.
So tapping would be easy. Just put the signal through a splitter (e.g. a bend in the fibre) and route your half of the signal to a decoder that works in the same way as the official one. The other end sees a 3dB drop in signal, but thats probably too small to be noticed.
Where this might be important is increased bandwidth. At the moment fibre transmission uses binary keying: send photons for 1, no photons for 0. Polarisation modulation means that you could use several different angles, and hence encode more than one bit per light pulse.
But don't get too excited about the bandwidth either. The limiting factor on bandwidth at the moment is the routers at the end of the fibre. We can pump terabits down a fibre in the lab, and 100 Gbit is pretty straighforward to do in the field. But put ten 100Gbit links into a router and you have to have a machine that can switch 1 Tbit. If the average packet is 1.5kbytes (Ethernet frame) then thats around 83 million packets per second. Even with hardware assist thats an awful lot of address table lookups per second.
Paul.
De-Beers and co are trying to aim for trademarks to identify "real" diamonds (i.e. dug out of the dirt). But I don't think it will really play in the long run.
Try asking if she would prefer an expensive honeymoon instead. That will also be a good reply to obnoxious women shoving huge rocks under her nose and saying "where's yours then?". You/she can just say "we decided to spend the money on a 5 star hotel in [Distant place]" and then gush about how wonderful it all was. Going to interesting places beats shiny rocks every time.
Paul.
The food-chain in ISPs looks something like this: Customer -> Tier 3 -> Tier 2 -> Tier 1, with each level paying the layer above for access. Tier 1 ISPs are people like UUNet with global reach. Tier 2 are national or "regional" (e.g. EU, Americas, Asia-Pacific). Tier 3 are local ISPs, and customers are both individual users and hosting companies.
Actually there is nothing stopping a customer or Tier 3 ISP from signing up with a Tier 1 ISP, and many do. But the principle is the same.
There are two kinds of link an ISP can have to other ISPs: Transit and Peering. In a transit link an ISP pays a larger ISP for access to "the Internet". In other words the smaller ISP can route packets through the link to any destination and expect to receive replies via the same route. In a peering relationship two ISPs, usually in the same Tier, agree to exchange traffic, usually without payment, but with the proviso that only traffic for customers of the other ISP is to be routed through that link. You can't send traffic to B through your peering link with A (although there are sometimes mutual backup link terms in the agreement).
You can think of this in your own terms quite easily. You have a transit link with your ISP that you pay for. But if you and your neighbour exchange a lot of traffic you might string an Ethernet cable between your houses and create your own peer link. But it would be very bad manners to use that link to pinch bandwidth off your neighbour.
The market forces that created this system are very straightforward. Originally the Internet worked with free transit links, but then the people investing in global networks realised that all the smaller ISPs were getting a free ride, and so they started demanding payment. This happened around 1996-7, and you can find lots of discussion papers from that time worrying about "the balkanisation of the Internet". In fact nothing of the sort happened. Metcalfe's law saw to it that everyone found more value from being connected to an unbalkanised Internet, and the net effect (sorry) was that money flowed from you and me up to Worldcom, and much good it did them. Meanwhile the smaller ISPs found that peering arrangements helped them to cut their costs because peer traffic avoided the expensive transit routes.
Thats not to say that things are so simple in real life. Peering arrangements in particular are fraught with difficulty because it usually means negotiating with your direct competitors, and you can play all sorts of dirty tricks like "hot potato" routing (routing packets to your nearest exit point instead of the globally most efficient one). But thats the general idea.
Incidentally the economics work like this regardless of the direction of most of the bits. People who tried to analyse the Internet using telephone economics got this wrong, because with the phone its usually the caller who pays. On the Internet the "caller" is hard to identify and the rules for doing so keep changing. And in any case the issue is irrelevant. You have content providers who want to reach readers and readers who want to access content. (Peer to peer changes the numbers and locations, but not the fundamentals). Both pay ISPs to provide this service, and those ISPs then pay the next tier up, and so on.
So now we look at India, where a bunch of Tier 3 and 2 ISPs are demanding payment from Tier 1 ISPs. The Tier 1 ISPs will rightly tell them to get lost.
I suppose that the Indian ISPs (who are mostly consumer ISPs) might demand payment from content providers such as Yahoo, Slashdot and co, on the grounds that the content providers want to reach Indian eyeballs. But I don't see this flying either. Those Indian eyeballs want the content just as much as the providers want to provide it, which is why you get no-payment peering arrangements between content providers and consumers: its the flow of value that counts, not the flow of bits.
Paul.
Politicians see a lot of pressure groups with axes to grind and corporate backing to pay their bills, and they know exactly how much to trust their numbers.
Of course the BSA doesn't exactly rely on its numbers to make its case. They rely on other numbers in campaign contributions as well. But thats a whole 'nother ball game.
Paul.
The key thing is that a court might in future decide that some claims are valid but others are not. So the first couple of claims in a patent might well lay claim to the entire state of the art, and might only be there as a kind of #define macro for subsequent claims. I once read an encryption patent (ISTR it was for a DVD system that didn't get used) where Claim 1 was for XORing the output of a random number generator with the cleartext. This was followed by a series of claims that started "A system as in Claim 1 where the random number generator is...".
So when you see a patent that seems to claim the whole of some technology, don't panic. There is going to be tons of prior art. You just have to work out where the prior art ends and the real invention starts. This is going to be a bit grey on the boundary (thats where patent lawyers make their money), but you can still get a fairly clear idea pretty quickly. You can also get a fair idea just by looking at the claims and thinking about the technology they represent. Once you get to precise descriptions of obscure algorithms then you are into the meat of the patent.
Incidentally, don't be scared of legalese. Just think of it as an unusually verbose and unstructured programming language.
Paul.
The platters rang like a bell when hit, even when sitting on the carpet. Get a set of platters from different drives together and you could do the Anvil Chorus.
Paul.
I note that on the page you are pointing to, the microsecond accuracy is described as "nominal". In other words, it isn't actually that accurate.
A purpse-made GPS receiver for time server applications is available, but costs rather more.
Paul.
No they won't, for the simple reason that making and distributing legal drugs is not their core competence. Smuggling and distributing contraband is their core competence. Once their product is no longer illegal there are lots of companies with the knowledge and infrastructure. Purity requirements alone are enough to put the illegal dealers out of business because only drugs made to sufficient purity will be legal to sell, and underground labs can't come anywhere near meeting those requirements.
If drugs are made legal then the illegal dealers will be out of business. They have a huge infrastructure dedicated to shipping those drugs into the country, and it will no longer have any value. I don't know of any product other than illgeal drugs that could use that infrastructure, so they can't switch to any other product.
Paul.
There are two kinds of relationship in the ISP world: peering and transit.
In a transit relationship a larger ISP provides a smaller one with long-distance connectivity for a fee, in exactly the same way that you pay your ISP. So local-ISP pays, say, UUNET for a connection that gives it access to the rest of the Internet. Such connections have a maximum bandwidth and typically include a per-megabyte charge element as well.
Peering relationships are used between ISPs of similar size to reduce the costs of transit. Two ISPs will agree to exchange traffic, but only where the source and destination are within those two ISPs. A cannot send traffic to C via its peering arrangement with B.
However all of this is irrelevant. When service is stolen the cost of the service is taken as its retail cost, not its incremental cost to the provider. Otherwise people stealing cable TV could argue that they have done nothing wrong because they didn't increase anyone's incremental costs.
Paul.
Standard computer crime search warrants, which date back to the early 80s, use a sweeping language that targets computers, most anything attached to a computer, most anything used to operate a computer - most anything that remotely resembles a computer - plus most any and all written documents surrounding it. Computer crime investigators have strongly urged agents to seize the works.
Elsewhere it talks about the actual process of seizure. They do take photographs of the configuration before disturbing anything. However this is for their benefit, not yours. You are assumed to know how to reassemble your own kit.
(Aside: I recall a case in Guernsey where a new sports car was bought in to the island. The Customs disassembled it in a search for contraband and didn't find anything. Then they told the owner that he could take it away. Not only did he have to pay to have it put back together, but the warranty was now void. Neither of these two things was considered "damage" worthy of compensation).
On passwords, I'm not sure about the US. I suspect your Fifth Amendment protects you. In the UK the Regulation of Investigative Practices Act authorises the police to demand your passwords and encryption keys on pain of two years imprisonment for failure to comply, and if you tell anyone other than your lawyer about it then you can be put in prison for five years.
The Hacker Crackdown is probably the best book on computer cracking I've read, even though it was written over 10 years ago. It looks at the subject from the POV of the crackers, the cops and the civil libertarians. If you are interested in the subject then read it.
Paul.
Where I do have a problem is with what actually happened and the lack of due process associated. Look at the sequence of events:
- The ISP notices the uncapped modem (I gather they use SNMP to ask the modem what its set to: nothing sophisticated).
- The ISP calls the FBI and alleges that this crime has cost it over $10,000. Hmmm. Where did that number come from? I'm on a 512kbit service for £25/month. Suppose I uncapped my modem to get the theoretical maximum of 64Mbits (the full channel bandwidth that is shared between all users on a spur). That is in theory a 128-fold increase in service, so I should be paying £3,200 per month, or around $5,000. So that may be two months service at 64Mbits. Maybe not too unreasonable, although I don't know how they estimated the time.
- The FBI get a search warrant based on the ISP's complaint and seize computers. This is perfectly legal: the authorities are permitted to seize the "instrumentality of the crime". If a PC was used to uncap the modem then it is an instrumentality of the crime. Also, if the case came to court then the defence could ask what evidence the prosecution had that the supposed perpetrators were actually responsible. Maybe it was a prankster thinking to do a "favour". Any prosecution is going to need smoking-gun scripts found on the suspect's PCs.
-
No charges are filed. Despite what I said just now, the whole thing is never tested in court. Confiscation of the computers (and any private data thereon) is considered enough of a punishment, and doesn't require the expense of a trial.
All of this is perfectly reasonable and legal, but it is never the less an end-run around the due-process principle. Based on a complaint and a search warrant your property can be effectively confiscated, and you have almost no come-back. Of course in theory you can sue for the return of your property, but all the police have to do is claim an "ongoing investigation" to make the suit fail.Paul.
However many PC manufacturers do that kind of thing, especially at the budget end. Then they can claim the machine is "fast" without having to spend lots of money.
DIY lets you balance the performance of the various components to meet your requirements instead of the marketing dept.
Paul.
Paul.
The field here is wide open. Lots of university biology departments are spinning off companies to make innovative new sensors, so you can get involved there. Or you can go and manage a Beowulf cluster for a big drug firm. Or anything in between.
Paul.
Company personnel departments can subscribe to salary surveys that are a slightly more accurate version of what I can find on Jobserve. Same problem.
From the point of view of the company, the value of an employee is the money they will earn if they hire him/er minus the money they will make if they don't, and also minus the costs of employing that person (about 50% of which is the salary, and the other 50% is desk, office space, electricity bills, computer, 10% of the cost of a manager, etc).
Any sum involving a big approximate number minus another big approximate number is going to have a huge error bar on the answer. For a software engineer the error bar probably exceeds the salary.
So managers, who perforce must make decisions regardless of the absence of evidence, fall back on rules of thumb. The first rule is, pay the going rate for that job title, hence those salary surveys. The second rule of thumb is that staff turnover is expensive. Every time you hire someone you commit money to teaching them the ropes until they become productive (whenever that is, see above), and also take a risk that this investment will be written off if they turn out to be bozos, or just up and leave in six months.
So, are you worth another 50% to your current employer? Possibly. Your management guesses so, but they might be wrong.
Paul.
First, write a policy for users to follow (not more than a page), get your boss to sign it, and then distribute copies. This tells everyone that the boss is behind this. If your boss does not have line authority over the people in question then get someone who does.
Second, get your boss's approval for a half hour tutorial for all staff on the subject, attendance mandatory for all users including him/herself. Get the boss to start with a brief repeat of the disaster story, then hand over to you (write the boss a script if necessary). Tell people why they need to do it and what it is they need to do, but obviously don't go into techie detail. Also emphasise that unsafe behaviour is letting the team down: its not just your work at risk, its everyones.
Its your responsibility to determine policy, configure machines, tell people what to do, monitor progress, and report to your boss. This can and should include saying that certain users are refusing to following departmental policy. Its then his/her job to take things further, upt to and including disciplinary action if necessary. Its not likely to be necessary: few people are that boneheaded.
Good luck. Culture change is hard, but its one of the most valuable things you can do.
Paul.
Suppose that Red Hat gains a patent on a new voice compression algorithm, and United Megacorp produces a voice conferencing product that uses this algorithm. UM can release its product with modular plugins for the codecs, and then just release the patented codec as OSS. Hardly a big win for the open source side.
Having said that, not all patents could be subverted so easily. The codec patent could be evaded, but a patent on using the codec within a voice conferencing application would not. At that point (AIUI) UM would have to release the entire conferencing system as OSS in order to fit inside the RH promise conditions.
Paul.
If so then Farnsworth is a bad example to use for the main point of this essay. Farnsworth's problems are as likely to stem from autism as anything else. Other lone geniuses have managed to create major inventions and use the patent system to do so. The Bell telephone is the obvious example.
Paul.