Exit polls would not have caught Florida, as the vote was within the range of possible results. If not for the indeterminate vote (hanging chad) issue, no one would have noticed that thousands of votes intended for Gore were cast for Buchanan. Even if there were a mechanism for a new vote in case of vote fraud, would it be enough to only be able to be sure of the landslides?
E-voting: Fair elections in all landslides. Close elections to the friends of the voting machine makers.
You are assuming that the developer is available to ask. This can also appear on resumes (or by management further up the ladder that does not deal with developers directly). Someone puts down that they have experience with Java 1.4. Hiring agent (not the potential supervisor--someone in HR) puts them as being out of touch relative to someone who writes Java 2 Version 4...even though there is no Java 2 Version 4 (unless they go back and redo the old version numbers). Thus, those putting inaccurate info on their resumes are more likely to get past the screening process than someone who does not. Not to mention those who start claiming version 5 experience immediately, even though it's not out yet. Without honest competition, they may make it through (normally this would get caught in the interview process, but in this case, they might interview better than those who only thought to claim version 4 experience).
Or the hiring agent is told to look for people with Version 5 experience and so dumps all the Java 2 people. It makes the whole screening process much more difficult. This is especially bad since techies can email out resumes by the bushel.
Wasn't SunOS at something like 5.3, then Solaris 2.4 through 2.7? Now it's Solaris 8 and 9 (I guess the next version would be X...assuming there's a next version).
Nothing new here. Just Sun being its normal confusing self. Next week, it will be called Dark Roast 6 (Java's due for a name change as well).
Yes, because communism was such a fountain of innovation.
It's not the "brilliant idea" that needs patent protection. It's the 999 wrong ideas that one tries before finding the one brilliant idea. This is why patents are necessary for pharmaceutical research. Pharmaceutical research is basically a brute force search (start with compound; check how it interacts with other substances until you find a useful result; test side effects in live subjects). Reverse engineering is relatively easy in comparison. Without patent protection, there would be no new pharmaceuticals. Even a truly dedicated individual could not do the research on one drug in a lifetime without help.
Outside of pharmaceuticals, patents are less justified. Implementation is a much larger expense and research is smaller. Most patent innovation would still get done, since companies need to implement new features anyway. However, it is possible (albeit arguable) that patent protection is generally useful (i.e. that it produces more innovation and long term sharing in return for the short term limitations that it imposes).
Software patents are not justified. One, they are always idea patents (as opposed to implementation patents), since the implementation details are protected by copyright (copyright is better than patents from the IP holder's point of view as it lasts longer). Two, implementation costs far outweigh research costs. Three, true reverse engineering is impossible, since the template (the source code) disappears during compilation. At best, someone can duplicate the end functionality (the idea of it).
You're missing the point. It's not a matter of doing "good." It's a matter of avoiding bad being done *to* you. You are focusing on the bad things that could be done *by* Microsoft but ignoring those that could be done *to* Microsoft. Under the current system, they must patent stupid things to avoid being targets of stupid patent claims.
They can of course try to use these patents against their competitors if they like. At that point, it would make sense to castigate Microsoft for seeking stupid patents. Until then, I will focus on the problems with the system. There are enough real reasons to be critical of Microsoft...why pick on them for something they *might* choose to do.
This reminds me of the original SCO situation. SCO accuses IBM of violating a contract by putting IBM IP into Linux that IBM had agreed to only put in Unix. Then a bunch of people point out that if SCO claimed that SCO IP was in Linux (which SCO initially denied), they could try to force people to pay licensing. All of a sudden, SCO's claims blossom from a breach of contract (in search of a buy out) to the predicted rant against Linux. Further, since they were already getting the bad press for it, they had no reason not to go forward (and thanks to Baystar, a hundred million reasons to proceed).
Or Microsoft just got tired of being sued by morons who patented the obvious and decided to start patenting the obvious...first. I'm not annoyed with MS for getting stupid patents; I will save my annoyance for when they start trying to enforce stupid patents (like IBM does).
When a company is as ridiculously profitable as MS, it makes sense to avoid potential problems by taking the safe route (apply for a patent on something that may not be patentable; if not patentable, great, you're only out a little money; if it is patentable, then still OK, since you got it first). If they did not, they might suddenly find themselves paying various people a dollar for every XP install. Or facing a SCO of their own that tries for maximum annoyance in order to get bought out.
A company could pay rent in stock...if it could find someone who would take it. In general, you would have to pay a premium (which will increase as the receiver's stock holdings increase) to get them to take stock in lieu of cash; thus, normally it would be simpler (and cheaper!) to sell the stock and pay with money. Paying expenses in stock usually only makes sense if your stock is not publicly traded and you have no cash reserves.
The point of paying employees in stock is that you hope they keep the stock. Holders of stock have a vested interest in making sure that the company increases in value. Thus, it is worth paying the premium in that case.
Note that the reluctance to accept stock in lieu of cash causes inflation immediately rather than requiring a large amount of transactions to be processed. I.e. it is not necessary to issue enough stock to dilute it to get an inflationary effect with that receiver. Further, note that the advantage of paying wages with stock is not that stock is easier to obtain, but that it is better for the employees to have stock.
Options and grants should both be expensed the same. Under the current rules, they are not. If options are expensed, then they have the same status as grants. Further, options are currently expensed for tax purposes. What this does is make the same expenses that give tax breaks included in the accounting. It may not be perfect, but it is better than the current system.
The natural corollary of your argument is that we should do things the other way: rather than mark them on the accounting, companies should not get a tax break for an option (or grant). It should just announce the stock dilution (twice for options: once at grant, once at exercise; stock grants are granted and exercised at the same time, so this is not necessary).
The primary reason that I see to expense stock options is to eliminate their current preferential treatment relative to stock *grants* and wages. Under the current system, stock grants and wages are expensed but options are not. Further, stock earnings are taxed as capital gains (thus lower than wages). As a result, companies replace wages with stock options at ridiculously low valuations for top executives.
Since stock options are riskless (if the stock price falls, one simply loses the ability to exercise the option; with granted stock, the executive notices both increases and decreases), they do not serve the claimed purpose of encouraging executives to care about company performance. By contrast, a stock grant that can only be sold after a certain measure of time encourages long term performance. Options encourage the taking of risks to increase stock price without penalizing the possibility of decreases in stock price.
This is not to say that your objection is not without merit, so here is my proposal: instead of using Black Scholes to value the options, use the current stock price (minus any exercise cost). If the option is not exercised, mark it as income in the new accounting period.
In regards to startups, I'm not sure what your worry is. Startups are not publicly traded, so those who invest in them have the direct contact necessary to properly evaluate the accounting. Further, startups are perfect places to pay in stock grants rather than options. Options are just extra paperwork. Unless the company takes off, both grants and options are just paper.
1. The division is irrelevant. The absolute level of income of the poor is the issue (not how they compare to the rich). Further, people who are in the bottom quintile in income now are more likely to be in the top quintile (if not rich, at least upper middle class) than in the bottom quintile in ten years. I.e. "the poor" is not a constant group.
2. Err...compared to their values in previous years? Again, why should I care what my income is relative to someone else? What is important to me is how I'm doing. It's not a score in a game. I don't need to be first.
The incomes of the bottom 10% have risen as well. I don't have numbers to demonstrate it on a year to year basis, but a hundred years ago, the typical job was six twelve-hour days a week (either on a farm or in a factory); now five eight-hour days in a week is typical. Further, "wife" is no longer a full time job of cooking and cleaning in a typical household (a hundred years ago, a spouse needed to stay home to do the labor that freezers, microwaves, dishwashers, and clothes washers have automated). Now, both spouses can earn external income.
Also, the majority of stocks in the US are owned by pension plans. The rich don't have pension plans. Those are primarily a middle class option. Thus, the majority of the "owners" are middle class.
Transferring income from the rich to the poor will *not* increase the median income unless you bring the poor's income above the median income. To increase the median income, you would transfer income to those *at* the median income. It will cause a decrease in GDP growth (which might *lower* the median income...a falling tide lowers all boats), as it reduces the income incentives of both the rich and the poor. The rich because they keep less of their income. The poor because they have less need of additional income...why work when you can live all right without it? Also, assuming that the poor get less transferred as they get more income, that's an effective increase in their marginal tax rate (the rate at which each additional dollar in income is taxed).
TMDA would handle the auto reply and verification system, which is sufficient. The stamp doesn't really add anything effective. It is better for the client app to pick up the verification, because putting it on the server burdens legitimate busy email servers.
It's also worth noting that this proposal doesn't actually do anything against spam. What this actually does is help alleviate the problem of false positives. Of course, it does this at the expense of allowing false negatives (every time a spammer stamps an email they are essentially marking it as not spam). In fact, it has much the same effect as increasing the spam threshold (i.e. making it harder for a message to be counted as spam) in one's filter. One still gets false positives from those who don't stamp.
Note that spammers currently send hundreds of emails to get a few through the filters. Under this system, they are guaranteed to get an email through with one email and some CPU time. This is a marginal improvement, as it does reduce the bandwidth used. However, in return, it makes mail servers use an increased amount of CPU and makes it harder to operate mailing lists.
Apparently everyone has forgotten all this since Microsoft suggested the same system. This is not to say that there aren't good points to the system: the best is the idea of opening up a new port on which challenges can be sent.
I should also point out that RAID-10 also means an entirely different kind of RAID as well as being sometimes used to refer to RAID-0+1. My previous post didn't make that clear.
"Any halfway decent RAID controller, or software RAID will do large reads by using both drives at once, greatly increasing the performance."
If it does, it is not RAID-1, but RAID-0+1 or RAID-10 (I've seen both names and others used). Using both drives for reads is called striping and is characteristic of RAID-0. Otherwise you are correct. It's just your terminology that is loose. However, the previous poster was also correct in a literal reading of the statement: RAID-1 is for data backups but is slower; RAID-0 is speed. The two can be combined to get both benefits simultaneously (without sacrificing read speed or the data backup).
I suspect that the confusion is mainly due to the fact that there is no point in pure RAID-1. The combination is strictly better, since there is no benefit from doing all the reads from one drive (except perhaps that the driver would be slightly simpler). I.e. what you call RAID-1 is actually the combination, because you've never used actual RAID-1.
There is a big difference between monitoring *traffic* and monitoring content. It's much the same as the difference between a police officer pulling you over for speeding or searching your trunk. One is based on public behavior; the other is checking something that is private.
I'm not one of those who complained about GMail though. As far as I am concerned, Google can offer whatever deal they would like. So long as they are up front and honest, I can then make an informed decision to accept or not.
"I use an external host provider and email service, so blocking port 25 indiscriminately would be a problem for me."
It shouldn't be. That's what port 587 does (allows authenticated connections to external mail servers). 25 is for server to server connections. Client/server connections should be made across port 587 (and authenticated!) for sending.
Yes, that's what OJ (at least in the criminal trial) and the Menendez brothers keep telling us. Hurricane Carter, et. al. are probably also ecstatic about the effectiveness of juries.
There is no evidence that the current patent system fails in anything but two areas:
1. Allowing process/software patents at all. This is an issue of law, not something under the patent office's control.
2. Allowing enough time per application for the examiners to properly review the patent. It costs as little as $355 for the initial patent examination. Elsewhere in this discussion, someone posted that a reexamination costs at least $2500. Is there any surprise that the initial examination is comparatively cursory (particularly with a long patent claim)? Plus, part of the $355 is taken from the patent office and used to fund other government projects. With the funds remaining, there simply isn't time to do a complete and thorough search for prior art.
Federal district attorneys are appointed by the President. State, county, and municipality district attorneys may or may not be elected (for example, in Pennsylvania, the state Attorney General is an elected position).
Show me a razor that does not require replacement blades. The original analogy had open source software as being the razor and the support as being blades. One can often get the razor (open source software) for free, either with blades (support, particularly installation support) or through an organization (when I first attended college, my dorm room came with a pack that included a razor; open source software can often be downloaded from places like sourceforge).
Razors and blades are no better than software, it's just that one expects to have to buy support (blades) for the razor. By contrast, software is advertised as not needing support, even though it does. Software requires support because it either:
1. Doesn't do the required feature in the required way by default and needs customizing (a common open source model), or
2. Has so many options that figuring out how to configure it properly is too complex to be done by a novice (a common proprietary software model).
Note that 1 means that average users *can't* customize the software to their needs while 2 means that average users *shouldn't* try to customize the software.
One example of this is remote graphical desktops. Both Linux (open source) and MS Windows (proprietary) can do this. In Linux, it requires a moderately complex process of editing configuration files (complex both in knowing what entries to edit and which files to edit). By the time one figures out how to do it, one has also read about various other security measures as well (for example, not logging on remotely as root). In MS Windows, one uses a GUI which has a yes/no option that changes all options everywhere at once. All one needs to do is find out where to make that change.
The proprietary model is inherently less secure, as it requires much more complex software (increased complexity offers more places where failure can occur). To be secure, proprietary software must anticipate all ways that someone might use the software and secure all of them. Unfortunately, this is not possible in an environment where one can trade chocolate for passwords (i.e. one cannot simultaneously have full access to everything and be willing to skimp on security precautions).
The open source model creates added security through complexity. Since it takes someone with experience to downgrade the security options, only people with the experience to know better than to do so usually change the security options. Further, since open source software is generally maintained by someone other than the end user, it is easier to keep access from being compromised (i.e. hopefully sysadmins are less willing to give up passwords for chocolate than average users).
The proprietary model may seem to require less support, as it is easier to modify. However, it is still difficult to modify *properly*. In fact, due to the way one GUI configuration change can make multiple backend changes, it is often *more* difficult to configure properly (for example, I just noticed that my WinXP box has file and print sharing turned on, but I don't know why; it only has networking for WAN purposes; my working hypothesis is that it was either on by default or that it got turned on when I switched to complex file permissions so that I could access files from my old hard drive).
The workload problem is why you stop doing patent examinations. Instead of charging a one size fits all fee and relying on the examiner to find prior art to invalidate the patent, move the examiners over to handling the reexamination requests and charge for those. If the reexamination shows that the patent is invalid, charge the patent filer. If it is valid, charge the reexamination request filer.
Note that to make this really work, one should send the reexamination request to the patent filer (with prior art) and let them determine for themselves if they want to defend the patent. If they concede the prior art, they can just release their patent filing (no charge for this).
This removes the burden of finding prior art from patent examiners and puts it on the shoulders of the reexamination filers (who presumably are interested parties who are aware of the general state of the industry). Patent examiners (now more properly called arbiters) can then just rule on the basis of the prior art claim (i.e. is there prior art that makes this obvious). Further, there is no longer any need to rush the examination. Instead of a flat fee, patent examiners (arbiters) can charge by the hour.
If you are in the right and provably so, you should have the better lawyers, as you are the one whose lawyers are going to get paid. To get to a possibility of millions of dollars, you would have to spend millions of dollars on the trial. It's not like you're in danger of having your $2000 trial turn into a million dollar trial because they ran up their legal bill; the judge won't award a legal fee that much higher than was needed.
It's not like the system has never been tried. We don't need to theorize on how the law might act under it. Loser pays is the default in most countries (and exists even in the US to a limited degree). If you want to convince me that it will damage the legal system, you can't just point out ways that it might possibly not work; you must actually use the existing data from places where it is being used to show that it doesn't work.
Just for kicks, I'll post my recommended patent system again:
1. Replace patent examination fees with patent filing fees; eliminate the initial examination (which is obviously not extensive enough) and just file the application automatically.
2. Move the current patent examiners to patent arbiter positions. If two parties disagree on a patent (presumably because one is trying to enforce the patent on the other), then they go before one of the new patent arbiters and state their position. Now paid by the hour rather than the case, the patent arbiter can sift through all the evidence (prior art basically) before ruling. No more four hour rubber stamps. The loser pays for the arbiter's time.
3. The loser may appeal to the regular court system if they feel the arbiter's decision was incorrect (this is just like the current system). If they do so, the arbiter can be called to testify as to why the original decision was made.
Since Microsoft doesn't provide the e-mail accounts, I don't see why it should be responsible for blocking spam at the account level. Would it be wise for MS to put some intelligent spam filtering in Outlook Express? Sure."
It seemed pretty clear to me that what the author meant was that MS Exchange does not include spam blocking by default (although Hotmail does). That is what he explicitly mentioned elsewhere in the article.
Saying "not an OS issue" is definitely missing the point. Email, web browsing, et. al. are not OS issues either (certainly much less so than *backups*), but MS still includes them with its OS.
In regards to Passport, Microsoft remembered step 2 but forget the Futurama step 1:
1. Give product away for free until users (web sites in this case) are addicted.
2. Jack up rates.
3. Profit!
Microsoft went straight to 2 without 1; thus, Passport never took off.
The only spam that has gotten past my SpamAssassin recently would have been stopped if my MTA was SPF compatible. It shows up as coming from the address to which it was sent.
It is very difficult to forge DNS records (one needs access to the legitimate name server). What is not difficult is creating legitimate DNS records. However, if spammers have to buy legitimate domains, then we can easily fix this by blacklisting those domains (and possibly revoking them). This can actually be done quite agressively with honeypot addresses (such addresses are set up such that no one has a legitimate reason to send to them and advertised as such on the web; thus, anyone who sends to them can be assumed to be a spammer and blacklisted).
Most ISPs do properly list their MX records (which list incoming mail servers). The problem comes with PTR records which should exist but are frequently outside the purview of small domains (requiring them stops over 90% of all spam, but blocks a small but significant number of legitimate senders who can't change the PTR records for their own IPs). SPF is simply a more flexible version. One can use SPF to specify that email is restricted to senders who have PTR records (if available) or list one's IPs or make all outgoing mail servers have MX records or whatever.
I agree that querying Spamhaus (etc.) and letting them interpret the SPF records is better than querying the DNS system direct. Spamhaus can block queries on spammer throwaway domains as they discover them (merging SPF and RBL into one query).
SPF is not a Microsoft technology. Caller ID is the Microsoft solution (similar but different). SPF was designed by pobox.com. Microsoft and pobox.com recently agreed to make SPF and Caller ID compatible, but they are still different methods:
1. SPF is text based; Caller ID is XML based (even though no other email header or DNS record is).
2. SPF verifies the envelope sender; Caller ID verifies the From header of the email. While both will be the same in many cases, they do not have to be.
Slashdot Mirror
Exit polls would not have caught Florida, as the vote was within the range of possible results. If not for the indeterminate vote (hanging chad) issue, no one would have noticed that thousands of votes intended for Gore were cast for Buchanan. Even if there were a mechanism for a new vote in case of vote fraud, would it be enough to only be able to be sure of the landslides?
E-voting: Fair elections in all landslides. Close elections to the friends of the voting machine makers.
You are assuming that the developer is available to ask. This can also appear on resumes (or by management further up the ladder that does not deal with developers directly). Someone puts down that they have experience with Java 1.4. Hiring agent (not the potential supervisor--someone in HR) puts them as being out of touch relative to someone who writes Java 2 Version 4...even though there is no Java 2 Version 4 (unless they go back and redo the old version numbers). Thus, those putting inaccurate info on their resumes are more likely to get past the screening process than someone who does not. Not to mention those who start claiming version 5 experience immediately, even though it's not out yet. Without honest competition, they may make it through (normally this would get caught in the interview process, but in this case, they might interview better than those who only thought to claim version 4 experience).
Or the hiring agent is told to look for people with Version 5 experience and so dumps all the Java 2 people. It makes the whole screening process much more difficult. This is especially bad since techies can email out resumes by the bushel.
Wasn't SunOS at something like 5.3, then Solaris 2.4 through 2.7? Now it's Solaris 8 and 9 (I guess the next version would be X...assuming there's a next version).
Nothing new here. Just Sun being its normal confusing self. Next week, it will be called Dark Roast 6 (Java's due for a name change as well).
Yes, because communism was such a fountain of innovation.
It's not the "brilliant idea" that needs patent protection. It's the 999 wrong ideas that one tries before finding the one brilliant idea. This is why patents are necessary for pharmaceutical research. Pharmaceutical research is basically a brute force search (start with compound; check how it interacts with other substances until you find a useful result; test side effects in live subjects). Reverse engineering is relatively easy in comparison. Without patent protection, there would be no new pharmaceuticals. Even a truly dedicated individual could not do the research on one drug in a lifetime without help.
Outside of pharmaceuticals, patents are less justified. Implementation is a much larger expense and research is smaller. Most patent innovation would still get done, since companies need to implement new features anyway. However, it is possible (albeit arguable) that patent protection is generally useful (i.e. that it produces more innovation and long term sharing in return for the short term limitations that it imposes).
Software patents are not justified. One, they are always idea patents (as opposed to implementation patents), since the implementation details are protected by copyright (copyright is better than patents from the IP holder's point of view as it lasts longer). Two, implementation costs far outweigh research costs. Three, true reverse engineering is impossible, since the template (the source code) disappears during compilation. At best, someone can duplicate the end functionality (the idea of it).
You're missing the point. It's not a matter of doing "good." It's a matter of avoiding bad being done *to* you. You are focusing on the bad things that could be done *by* Microsoft but ignoring those that could be done *to* Microsoft. Under the current system, they must patent stupid things to avoid being targets of stupid patent claims.
They can of course try to use these patents against their competitors if they like. At that point, it would make sense to castigate Microsoft for seeking stupid patents. Until then, I will focus on the problems with the system. There are enough real reasons to be critical of Microsoft...why pick on them for something they *might* choose to do.
This reminds me of the original SCO situation. SCO accuses IBM of violating a contract by putting IBM IP into Linux that IBM had agreed to only put in Unix. Then a bunch of people point out that if SCO claimed that SCO IP was in Linux (which SCO initially denied), they could try to force people to pay licensing. All of a sudden, SCO's claims blossom from a breach of contract (in search of a buy out) to the predicted rant against Linux. Further, since they were already getting the bad press for it, they had no reason not to go forward (and thanks to Baystar, a hundred million reasons to proceed).
Please stop giving them ideas.
Or Microsoft just got tired of being sued by morons who patented the obvious and decided to start patenting the obvious...first. I'm not annoyed with MS for getting stupid patents; I will save my annoyance for when they start trying to enforce stupid patents (like IBM does).
When a company is as ridiculously profitable as MS, it makes sense to avoid potential problems by taking the safe route (apply for a patent on something that may not be patentable; if not patentable, great, you're only out a little money; if it is patentable, then still OK, since you got it first). If they did not, they might suddenly find themselves paying various people a dollar for every XP install. Or facing a SCO of their own that tries for maximum annoyance in order to get bought out.
A company could pay rent in stock...if it could find someone who would take it. In general, you would have to pay a premium (which will increase as the receiver's stock holdings increase) to get them to take stock in lieu of cash; thus, normally it would be simpler (and cheaper!) to sell the stock and pay with money. Paying expenses in stock usually only makes sense if your stock is not publicly traded and you have no cash reserves.
The point of paying employees in stock is that you hope they keep the stock. Holders of stock have a vested interest in making sure that the company increases in value. Thus, it is worth paying the premium in that case.
Note that the reluctance to accept stock in lieu of cash causes inflation immediately rather than requiring a large amount of transactions to be processed. I.e. it is not necessary to issue enough stock to dilute it to get an inflationary effect with that receiver. Further, note that the advantage of paying wages with stock is not that stock is easier to obtain, but that it is better for the employees to have stock.
Options and grants should both be expensed the same. Under the current rules, they are not. If options are expensed, then they have the same status as grants. Further, options are currently expensed for tax purposes. What this does is make the same expenses that give tax breaks included in the accounting. It may not be perfect, but it is better than the current system.
The natural corollary of your argument is that we should do things the other way: rather than mark them on the accounting, companies should not get a tax break for an option (or grant). It should just announce the stock dilution (twice for options: once at grant, once at exercise; stock grants are granted and exercised at the same time, so this is not necessary).
The primary reason that I see to expense stock options is to eliminate their current preferential treatment relative to stock *grants* and wages. Under the current system, stock grants and wages are expensed but options are not. Further, stock earnings are taxed as capital gains (thus lower than wages). As a result, companies replace wages with stock options at ridiculously low valuations for top executives.
Since stock options are riskless (if the stock price falls, one simply loses the ability to exercise the option; with granted stock, the executive notices both increases and decreases), they do not serve the claimed purpose of encouraging executives to care about company performance. By contrast, a stock grant that can only be sold after a certain measure of time encourages long term performance. Options encourage the taking of risks to increase stock price without penalizing the possibility of decreases in stock price.
This is not to say that your objection is not without merit, so here is my proposal: instead of using Black Scholes to value the options, use the current stock price (minus any exercise cost). If the option is not exercised, mark it as income in the new accounting period.
In regards to startups, I'm not sure what your worry is. Startups are not publicly traded, so those who invest in them have the direct contact necessary to properly evaluate the accounting. Further, startups are perfect places to pay in stock grants rather than options. Options are just extra paperwork. Unless the company takes off, both grants and options are just paper.
Linspire's primary target audience has always been the home market. Lycoris, Mandrake, etc. also have OEM programs.
1. The division is irrelevant. The absolute level of income of the poor is the issue (not how they compare to the rich). Further, people who are in the bottom quintile in income now are more likely to be in the top quintile (if not rich, at least upper middle class) than in the bottom quintile in ten years. I.e. "the poor" is not a constant group.
2. Err...compared to their values in previous years? Again, why should I care what my income is relative to someone else? What is important to me is how I'm doing. It's not a score in a game. I don't need to be first.
The incomes of the bottom 10% have risen as well. I don't have numbers to demonstrate it on a year to year basis, but a hundred years ago, the typical job was six twelve-hour days a week (either on a farm or in a factory); now five eight-hour days in a week is typical. Further, "wife" is no longer a full time job of cooking and cleaning in a typical household (a hundred years ago, a spouse needed to stay home to do the labor that freezers, microwaves, dishwashers, and clothes washers have automated). Now, both spouses can earn external income.
Also, the majority of stocks in the US are owned by pension plans. The rich don't have pension plans. Those are primarily a middle class option. Thus, the majority of the "owners" are middle class.
Transferring income from the rich to the poor will *not* increase the median income unless you bring the poor's income above the median income. To increase the median income, you would transfer income to those *at* the median income. It will cause a decrease in GDP growth (which might *lower* the median income...a falling tide lowers all boats), as it reduces the income incentives of both the rich and the poor. The rich because they keep less of their income. The poor because they have less need of additional income...why work when you can live all right without it? Also, assuming that the poor get less transferred as they get more income, that's an effective increase in their marginal tax rate (the rate at which each additional dollar in income is taxed).
TMDA would handle the auto reply and verification system, which is sufficient. The stamp doesn't really add anything effective. It is better for the client app to pick up the verification, because putting it on the server burdens legitimate busy email servers.
It's also worth noting that this proposal doesn't actually do anything against spam. What this actually does is help alleviate the problem of false positives. Of course, it does this at the expense of allowing false negatives (every time a spammer stamps an email they are essentially marking it as not spam). In fact, it has much the same effect as increasing the spam threshold (i.e. making it harder for a message to be counted as spam) in one's filter. One still gets false positives from those who don't stamp.
Note that spammers currently send hundreds of emails to get a few through the filters. Under this system, they are guaranteed to get an email through with one email and some CPU time. This is a marginal improvement, as it does reduce the bandwidth used. However, in return, it makes mail servers use an increased amount of CPU and makes it harder to operate mailing lists.
Apparently everyone has forgotten all this since Microsoft suggested the same system. This is not to say that there aren't good points to the system: the best is the idea of opening up a new port on which challenges can be sent.
I should also point out that RAID-10 also means an entirely different kind of RAID as well as being sometimes used to refer to RAID-0+1. My previous post didn't make that clear.
"Any halfway decent RAID controller, or software RAID will do large reads by using both drives at once, greatly increasing the performance."
If it does, it is not RAID-1, but RAID-0+1 or RAID-10 (I've seen both names and others used). Using both drives for reads is called striping and is characteristic of RAID-0. Otherwise you are correct. It's just your terminology that is loose. However, the previous poster was also correct in a literal reading of the statement: RAID-1 is for data backups but is slower; RAID-0 is speed. The two can be combined to get both benefits simultaneously (without sacrificing read speed or the data backup).
I suspect that the confusion is mainly due to the fact that there is no point in pure RAID-1. The combination is strictly better, since there is no benefit from doing all the reads from one drive (except perhaps that the driver would be slightly simpler). I.e. what you call RAID-1 is actually the combination, because you've never used actual RAID-1.
There is a big difference between monitoring *traffic* and monitoring content. It's much the same as the difference between a police officer pulling you over for speeding or searching your trunk. One is based on public behavior; the other is checking something that is private.
I'm not one of those who complained about GMail though. As far as I am concerned, Google can offer whatever deal they would like. So long as they are up front and honest, I can then make an informed decision to accept or not.
"I use an external host provider and email service, so blocking port 25 indiscriminately would be a problem for me."
It shouldn't be. That's what port 587 does (allows authenticated connections to external mail servers). 25 is for server to server connections. Client/server connections should be made across port 587 (and authenticated!) for sending.
Perhaps they got it confused with a different patent, and that one should say "applying for a credit card online." It's also possible that they meant this claim: "2. The system of claim 1, wherein said terminal further comprises means for generating a personal qualification report including means for determining the type and amount of goods or services which can be provided to an entity wherein said means for determining comprises means for mathematically processing said information entered through said means for entering." As I read it, that could be construed as representing the authorization process.
"The jury system works reasonably well"
Yes, that's what OJ (at least in the criminal trial) and the Menendez brothers keep telling us. Hurricane Carter, et. al. are probably also ecstatic about the effectiveness of juries.
There is no evidence that the current patent system fails in anything but two areas:
1. Allowing process/software patents at all. This is an issue of law, not something under the patent office's control.
2. Allowing enough time per application for the examiners to properly review the patent. It costs as little as $355 for the initial patent examination. Elsewhere in this discussion, someone posted that a reexamination costs at least $2500. Is there any surprise that the initial examination is comparatively cursory (particularly with a long patent claim)? Plus, part of the $355 is taken from the patent office and used to fund other government projects. With the funds remaining, there simply isn't time to do a complete and thorough search for prior art.
Federal district attorneys are appointed by the President. State, county, and municipality district attorneys may or may not be elected (for example, in Pennsylvania, the state Attorney General is an elected position).
Show me a razor that does not require replacement blades. The original analogy had open source software as being the razor and the support as being blades. One can often get the razor (open source software) for free, either with blades (support, particularly installation support) or through an organization (when I first attended college, my dorm room came with a pack that included a razor; open source software can often be downloaded from places like sourceforge).
Razors and blades are no better than software, it's just that one expects to have to buy support (blades) for the razor. By contrast, software is advertised as not needing support, even though it does. Software requires support because it either:
1. Doesn't do the required feature in the required way by default and needs customizing (a common open source model), or
2. Has so many options that figuring out how to configure it properly is too complex to be done by a novice (a common proprietary software model).
Note that 1 means that average users *can't* customize the software to their needs while 2 means that average users *shouldn't* try to customize the software.
One example of this is remote graphical desktops. Both Linux (open source) and MS Windows (proprietary) can do this. In Linux, it requires a moderately complex process of editing configuration files (complex both in knowing what entries to edit and which files to edit). By the time one figures out how to do it, one has also read about various other security measures as well (for example, not logging on remotely as root). In MS Windows, one uses a GUI which has a yes/no option that changes all options everywhere at once. All one needs to do is find out where to make that change.
The proprietary model is inherently less secure, as it requires much more complex software (increased complexity offers more places where failure can occur). To be secure, proprietary software must anticipate all ways that someone might use the software and secure all of them. Unfortunately, this is not possible in an environment where one can trade chocolate for passwords (i.e. one cannot simultaneously have full access to everything and be willing to skimp on security precautions).
The open source model creates added security through complexity. Since it takes someone with experience to downgrade the security options, only people with the experience to know better than to do so usually change the security options. Further, since open source software is generally maintained by someone other than the end user, it is easier to keep access from being compromised (i.e. hopefully sysadmins are less willing to give up passwords for chocolate than average users).
The proprietary model may seem to require less support, as it is easier to modify. However, it is still difficult to modify *properly*. In fact, due to the way one GUI configuration change can make multiple backend changes, it is often *more* difficult to configure properly (for example, I just noticed that my WinXP box has file and print sharing turned on, but I don't know why; it only has networking for WAN purposes; my working hypothesis is that it was either on by default or that it got turned on when I switched to complex file permissions so that I could access files from my old hard drive).
The workload problem is why you stop doing patent examinations. Instead of charging a one size fits all fee and relying on the examiner to find prior art to invalidate the patent, move the examiners over to handling the reexamination requests and charge for those. If the reexamination shows that the patent is invalid, charge the patent filer. If it is valid, charge the reexamination request filer.
Note that to make this really work, one should send the reexamination request to the patent filer (with prior art) and let them determine for themselves if they want to defend the patent. If they concede the prior art, they can just release their patent filing (no charge for this).
This removes the burden of finding prior art from patent examiners and puts it on the shoulders of the reexamination filers (who presumably are interested parties who are aware of the general state of the industry). Patent examiners (now more properly called arbiters) can then just rule on the basis of the prior art claim (i.e. is there prior art that makes this obvious). Further, there is no longer any need to rush the examination. Instead of a flat fee, patent examiners (arbiters) can charge by the hour.
If you are in the right and provably so, you should have the better lawyers, as you are the one whose lawyers are going to get paid. To get to a possibility of millions of dollars, you would have to spend millions of dollars on the trial. It's not like you're in danger of having your $2000 trial turn into a million dollar trial because they ran up their legal bill; the judge won't award a legal fee that much higher than was needed.
It's not like the system has never been tried. We don't need to theorize on how the law might act under it. Loser pays is the default in most countries (and exists even in the US to a limited degree). If you want to convince me that it will damage the legal system, you can't just point out ways that it might possibly not work; you must actually use the existing data from places where it is being used to show that it doesn't work.
Just for kicks, I'll post my recommended patent system again:
1. Replace patent examination fees with patent filing fees; eliminate the initial examination (which is obviously not extensive enough) and just file the application automatically.
2. Move the current patent examiners to patent arbiter positions. If two parties disagree on a patent (presumably because one is trying to enforce the patent on the other), then they go before one of the new patent arbiters and state their position. Now paid by the hour rather than the case, the patent arbiter can sift through all the evidence (prior art basically) before ruling. No more four hour rubber stamps. The loser pays for the arbiter's time.
3. The loser may appeal to the regular court system if they feel the arbiter's decision was incorrect (this is just like the current system). If they do so, the arbiter can be called to testify as to why the original decision was made.
"Spam blocking for our e-mail accounts.
Since Microsoft doesn't provide the e-mail accounts, I don't see why it should be responsible for blocking spam at the account level. Would it be wise for MS to put some intelligent spam filtering in Outlook Express? Sure."
It seemed pretty clear to me that what the author meant was that MS Exchange does not include spam blocking by default (although Hotmail does). That is what he explicitly mentioned elsewhere in the article.
Saying "not an OS issue" is definitely missing the point. Email, web browsing, et. al. are not OS issues either (certainly much less so than *backups*), but MS still includes them with its OS.
In regards to Passport, Microsoft remembered step 2 but forget the Futurama step 1:
1. Give product away for free until users (web sites in this case) are addicted.
2. Jack up rates.
3. Profit!
Microsoft went straight to 2 without 1; thus, Passport never took off.
The only spam that has gotten past my SpamAssassin recently would have been stopped if my MTA was SPF compatible. It shows up as coming from the address to which it was sent.
It is very difficult to forge DNS records (one needs access to the legitimate name server). What is not difficult is creating legitimate DNS records. However, if spammers have to buy legitimate domains, then we can easily fix this by blacklisting those domains (and possibly revoking them). This can actually be done quite agressively with honeypot addresses (such addresses are set up such that no one has a legitimate reason to send to them and advertised as such on the web; thus, anyone who sends to them can be assumed to be a spammer and blacklisted).
Most ISPs do properly list their MX records (which list incoming mail servers). The problem comes with PTR records which should exist but are frequently outside the purview of small domains (requiring them stops over 90% of all spam, but blocks a small but significant number of legitimate senders who can't change the PTR records for their own IPs). SPF is simply a more flexible version. One can use SPF to specify that email is restricted to senders who have PTR records (if available) or list one's IPs or make all outgoing mail servers have MX records or whatever.
I agree that querying Spamhaus (etc.) and letting them interpret the SPF records is better than querying the DNS system direct. Spamhaus can block queries on spammer throwaway domains as they discover them (merging SPF and RBL into one query).
SPF is not a Microsoft technology. Caller ID is the Microsoft solution (similar but different). SPF was designed by pobox.com. Microsoft and pobox.com recently agreed to make SPF and Caller ID compatible, but they are still different methods:
1. SPF is text based; Caller ID is XML based (even though no other email header or DNS record is).
2. SPF verifies the envelope sender; Caller ID verifies the From header of the email. While both will be the same in many cases, they do not have to be.