synaptic runs just fine under fedora as well. But I don't want a graphical tool - I mostly use my linux boxes over ssh anyway. Yum also supports apt repos, but for as long as fedora core stays rpm-based, switching is more trouble than worth for me.
Also, for some reason I don't like apt-get or apt-cache.. aptitude is pretty much the only apt-based management tool I really like.
As a long-time redhat user I've come to like rpm a lot. I like spec-files, I like rpm, I like yum.
But, deb's dependency management is a little finer grained. And aptitude is a great tool.
So, if I could get my familiar rpm and yum commandlines, aptitude, and deb dependencies I guess it'd be the best of both worlds for me.
One huge problem for rpm-based distroes is of course that each of them has different packages and dependency trees. Ever tried using five yum repositories and hoping they just somehow manage to get along? As debian has larger number of packages from single source than any of the rpm-based distroes, and debian maintainers seem to handle dependencies pretty well, the resulting repository-hell might be a little easier to handle. But that's not a problem with the package format - it's a problem with repositories and maintainers.
Open source hackers may well be numerous, but how many of those is good in software development? Remember that the process of creating something for proof of concept, or demonstration, or to solve a specific single problem (which three cover most OSS projects), and even developing without external pressures of budget, manpower and schedule, is very different from disciplined, goal-oriented professional software development. While you can fix many problems by adjusting development process to counter them, the most important issue becomes getting the developers to follow the process (and having a working process in the first place).
I'd say most hackers are better suited in more R&D oriented projects where just getting something to work is the goal, but few of them are so well suited to actual product development.
I'm just ordering a Tyan Thunder K8S Pro. It has PCI-X, dual GbE (+100Mbps one), onboard SATA RAID (although I'm not at all sure if I'm going to use the very limited RAID capabilities). There's a SCSI version as well, but I'm rather taking SATA and upgrading later with a separate SATA RAID host adapter.
I admit that on Intel side there would be more server motherboards to choose from, but it seems to me that Tyan is offering pretty much everything on Opteron boards as well.
3ware 9508S with 5 Maxtor DM+9 200GB disks in RAID5 configuration under WinXP has typical read-speed of about 40MB/s when you have several apps accessing the disk at the same time, and clocks a little over 100MB/s when you close the apps and benchmark the system. Don't know if that's fast or not - just that it's enough for me.
Depending on where in Europe you're moving back, the step down might not be THAT drastic. At least Sweden has 10 & 100Mbps connections, and Finland has 10Mbps (not everywhere, but I doubt trees need it - maybe the bears would). Still, now that I've had 10Mbps (symmetric, ethernet) for three years, I've come to yearn for more.. 100Mbps would be nice.
What do you mean with HPLJ4m+ having been "the last of the *REALLY* beefy HP printers"? I've been very happy with my HPLJ4050TN, which is the next generation of LJ4-series afaik..
No, I don't have any intention of buying an inkjet printer of any kind, brand or model. I'm happy with b&w.
Unless horde has been redone (which would break compatibility with original horde), it's ridden by massive design flaws. Technically it could be redone - while the existing design flaws are fundamental, the concept itself doesn't have such technical problems that it couldn't be fixed by better design.
Whether it has social design flaws is another matter. I think that it's design diverges too much from ed2k design that it should be separated completely.
The particular bill talked about in the article would've required the source to be delivered with any software built specifically for govt. It wouldn't have required source for cots components.
You can get source for commercial software as well as open source software. Just require it in the contract. Probably will cost more than a binary-only version, and there are likely restrictions different from restrictions in open source licenses, but the source should be available still.
I think that kind of a bill is fairly good. The govt will last longer than the company that created that particular piece of software, and it may be in use long after the company has fired those participating in development of that specific software. Having source is insurance. If license terms allow for it, being able to contract third parties to fix and/or improve the software later on is even better, but just having the source is a lot.
If the french do not like the content, why don't *they* block it, or enforce it through *their* internet providers?!
Because the carriers that bring bits to France are not liable for the content. Like telcos are not liable for the content of phone converstations. And ISPs are not liable for content of some other website they don't control.
For as long as a company does business in a country, it's liable. Now, if Yahoo decided to NOT do business in France, it could do so by blocking French IP address ranges.
However, those issues are not very interesting - they're business as usual. What's interesting is that the French can prosecute in France, and if Yahoo or its subsidiaries have presence there, the French authorities can seize their assets if Yahoo doesn't pay the fines; And that the US courts will judge by US laws, thus not forcing Yahoo to pay fines to France by order of a French court.
This seems to establish that a company that is willing to do business worldwide is considered to have established business presence in all countries simply by having a website. But, unless the company has assets in other countries, they're pretty safe. A company selling stuff could easily say they don't do business worldwide by refusing to ship to some countries. But with auctions, you can't do that, so you must block the business by other means.
Of course if Yahoo were to block French IP range, French could use proxy servers in other countries to bypass that block. But, at that point Yahoo (IMO) could reasonably deny doing business in France: they blocked access to (parts of) the website from French IP ranges.
This time, I've started keeping my CD rips in a lossless format. Next time (which will put me around 0.75TB) I will probably start keeping raw DVD rips. After that, I don't know what else I might keep that could use so much room. Until now, audio and small video clips have taken the bulk of the space.
At about 300MB per CD, I can fit my collection to below 100GB. And it grows slowly enough that HD sizes will easily keep up. But, my DVD collection would already take some 5TB - more than three times my current total diskspace. It'll take 2-3 years before HD sizes grow faster than my DVD collection does, and then some more before I have enough that I can keep them all on disks.
Of course in the 5-10 years it takes for my diskspace to grow beyond my dvd collection (the amount of other data at that point would be somewhere between a rounding error and a hot spare), I'll have some HD-DVD / Bluray / verylargecapacityopticaldisc format videos already, and I yet again think it'll be just some 2-5 years for disk capacity to grow to that size..
Although I know everyone who has ever said this has later eaten their words, at the moment, I really don't think any home computer needs more than a few TB of storage.
Yup. As noted, I'd already need a few TBs to keep my non-hollywood (well, maybe 1-2% Hollywood afterall) commercial video entertainment material on a "Home Entertainment System". And the need for space will grow in the future - I can see it up to at least 100TB with just next gen HD video. And when we're there, there's bound to be something even larger that I want to keep online.
10 years would mean experience with Linux since release of 1.0. I know many who would qualify for that (myself included). 15, OTOH, is longer than Linux has existed..
Just halve the years of experience required (5-8 years) and you should find enough people easily. Add a few years of other *nix OS's and it's just fine.
The article states volunteer work worth about 10-15M, so the entry isn't really much less expensive than Rutan's. Could be that Scaled hadn't burned that much before they started test flights, in which case Scaled Composites would come out cheaper.
Not counting personnel costs isn't really an option unless you want to bet future flights on volunteer work.
HD manufacturers have no excuse for their confusing use of 10's power prefixes as the basic access unit of a disk is a 2's power of a byte.
Networking, OTOH, does have valid excuses. Eg. 64kbps for ISDN means 64000bps, which comes from 8kHz 8b width, and 10Mbps ethernet is indeed 10 million bits per second, not roughly 4.9% more than that - again because the electrical engineers didn't think in 2's power when assigning bandwidth, but rather in conventional 10's powers.
Depends. If I end up with a piece of junk that is partially functional (good mobo & CPU, bad RAM) but can't be used anymore, and will have to pay for all the new parts (or rather part combination), then yes.
If the parts were welded together, where the sticks of RAM have lifetime warranty, I'd of course demand lifetime warranty for the RAM still. So, if it means the shop will replace the whole due to lifetime warranty of RAM, then it's OK for me.
Still, considering I buy the combination together, if there are problems I'll take the whole combination to the shop. Even when I know it's bad RAM - I still take mobo & CPU to the shop with the RAM.
As it happened recently that something failed in a fairly new computer of mine, and even while I diagnosed it as bad RAM but took the whole computer to the shop, they did replace mobo, CPU, RAM, and the PSU - all within the warranty, for no cost at all. PSU was the strange one in the group, but as the report noted that old PSU was giving spikes, I guess it was all for the better. If I had taken just a stick of RAM to the shop, might be I'd have an unstable computer now, causing lots of frustration but no obvious diagnosis.
Nope. It's part of the yearly checkup at the repair shop. My yearly kilometers are low enough that a year passes sooner than the kilometer count between required checkups.
I do KNOW how to check / add / change oil. But why would I do that when it's one of the items on the "yearly checkup" list? I do actually know a lot more about cars than I care for. I could probably do the routine maintenance on an old car if I really had to, and had some reference material to refresh my memory with. I just don't want to.
Similarly, some years back I did a lot of computer maintenance. But these days I design software, and the computers are a tool (as opposed to target of the work), so I'd rather have them "just work". I've got better things to use my time on than to repair a PC when I can have the PC fixed at a fair price by people whose jobs' target is the PC.
As I pay for someone else to fix my PCs (where I could do that myself too, thank you), I can spend the time eg. with family, friends, or even earning money with which to pay for the repairs.
Now, do you do your hair / beard / moustache yourself? Cook from real raw materials (as opposed to fastfood, restaurants, or cook using almost ready ingredients)? Sew your own suits? And so on. Some people eat out almost always, some cook themselves. Some buy their clothes, some sew themselves. Some fix their own cars or computers, some have others fix them. It's just a choice about what you do yourself and what not. Nobody can do everything these days. Really. And I like cooking more than fixing a car. I like designing software more than diagnosing broken PCs.
I already treat mobo + CPU + RAM as a single package. Soldering them together wouldn't make much of a difference to me - assuming combinations I want would be available for sale.
If I want a new CPU (at leats 50% speed increase, otherwise it just isn't worth the trouble), I need to get a new mobo because the old mobo won't support the new CPU. And then I need to get new RAM as the old RAM isn't compatible with the new mobo. Or is just so slow that I wouldn't realize the speed increase from CPU change really. At that point I think about moving the old comp as is to some supportive role (replace old firewall, fileserver, mailserver, or something). So, I'll look at the oldest comp to replace that with the one being replaced currently, and notice that the 7 year old case won't take the new mobo formfactor anymore, and would need new PSU anyway, and so on.
So, I end up buying pretty much a whole new computer unless I'm willing to ditch mobo + CPU + RAM that's at least twice as fast as the oldest one still in use.
For me, a car is a way to get from place A to place B comfortably, economically, and in reasonable time.
In last year I've opened the hood a few times: to add water (and some liquid stuff they sell at gas stops) to windshield cleaning thingie. If there's anything except adding fuel or the cleaning solution, the car goes to a repair shop. Well, changing tires is done in another shop as I know the people working there and know they're going to sell me good tires for reasonable price.
I don't want to open the hood. I don't want to do any repairs. I want to move between locations comfortably, economically, when I want to (as opposed to when the busses/trains go, although I do choose the train quite often).
For these reasons, I'm not going to buy a 20 year old "real car" but rather a new one that will work without trouble for a few years again.
Whose collection of video and music are you talking about? At least not mine..
Last year I spent over 5kEUR on DVDs and CDs. While in the meantime I downloaded similar content by gigabytes and uploaded by terabytes. Oh yes, I do have some Hollywood content, too, but that's all bought, not downloaded, because that I see on TV, too. Non-hollywood content I won't be seeing on TV or hearing on radio, so I have to download, watch/listen, and then buy.
30$ per disc? It's SRP is 19.98, and I paid perhaps some 14$ per disc (plus 14$ for the movie). It was cheap for a disc, expensive per minute of content. I'm used to paying about 24$ per disc, nowadays often with 3 episodes (of ~23 minutes each). Used to be four episodes, even six, per disc, and the price per disc certainly hasn't come down..
Even this assumes that people try to vote optimally instead of honestly, which I don't think is the case. Also, the scenario assumes a few rounds of polls with correct results, and that people would, based on what the polls give as the expected result, approve of more candidates after each poll, and their new approvals would be reflected (correctly) in the next poll. Also, would really all the Tsongas supporters approve of Kerrey any more than they approve of Clinton? And so on. A small number of people would still "waste" their votes by not approving of any of the top candidates, just approving of what they really would approve of. Wasting a vote by voting for a candidate that has no chance of winning is a valid strategy. Hey, we have Donald Duck often getting a significant approval (thousands of votes - OK, it's not THAT significant, but still). There are empty votes, votes that are given to entities not on the list, and so on.
Note that I also required approval of majority before anything else is considered. In that case, Tsongas has no chance, whatsoever, as he's got ~33% approval. Also, I would think it quite unlikely that each and every person would approve the same number of candidates on their list.. I can easily think of cases where I could approve of two candidates, really hate two, and consider the rest of them as "lesser evils". Which brings me to a problem in approval voting: I want to express the concepts of "approval", "disapproval", and "indifference" instead of having to lump those I'm indifferent towards to either "approved" or "disapproved" category. However, approval would be better than standard plurality vote.
On Condorcet, I didn't see a "tie" vote anywhere, it looked like it required ranking (no ties) of candidates with the option of not ranking them all (truncated list). If there is the option of a "tie" vote, would it be possible to rank one candidate as the best, not vote (disapprove of) one, and tie the rest (indifference)?
Voting in Finland. The only governmental "single winner" vote in Finland is Presidential election. We have the parliament (200 members), city/county councils (I think 441 total in Finland) with a few to tens of members depending on the size of the city, and the president. Those are the three common publicly elected governmental bodies.
So, in Presidential elections, there is the first round (one vote per person, each vote to one and only one candidate). If someone gets >50%, we're done. Doesn't happen, though.. So, there is a second round where the choice is between the top two candidates. Not a perfect system, but works. My first candidate has never been elected, but I still go in the first round for my primary candidate. I know that s/he (yes, we've had women as candidates, and I've voted for one in the first round) might have a chance since s/he's usually in the "high-middle" in the polls, close enough to the top candidates.
So, I always vote on the second round, and my vote has always been rather "against" than "for", as the second round candidates up to date have been one I'm indifferent towards and one I strongly disapprove of.
The major difference between US and Finland in presidential elections is thus that in US, there are two candidates that have any chance at all to be elected, and the rest are protest votes, while in Finland, we almost always have at least three candidates with a good realistic chance of getting elected. So, it's not as likely that people would approve of two of the top3 as it is to approve of just one of the top2. In Your example, I would guess eg. that some of the Tsongas supporters are casting protest votes, and wouldn't care one way or the other about the rest of the candidates, thus casting their votes only for Tsongas. This would eliminate some of the apparent Kerrey support. Also, there is nothing about Harkin that would make him more than harmless. He's someone the people know not to fuck the country, not to do anything that would change things, not to disrupt the system. Someone who's good for just waiting for the next elections to get a real president. While that's a valid option, I don't think it's good - it's pretty much the same as having no president at all:) People who vote for the harmless option as their primary vote don't seem like ones to cast a second vote.
Still, trying to get an order between the candidates in the example would require sociological study instead of mathematical study. From the patterns presented alone I can't gather what would be the actual votes, as I don't really believe people could rank all candidates. Poeple are likely to have an opinion (strong yes or no) about two or three candidates, and no opinion about the rest.
Oh yes, how can people in Finland vote for candidates not on the list in the first place (note: the votes for invalid candidates are tallied as "invalid" and don't thus effect the vote)? We write the number of our chosen candidate on a piece of paper. So, people can write whatever they want there.. The votes are first optically read, and then the votes that couldn't be OCR'd are read and input by the vote counters.
One point that isn't addressed in this (or other) mathematical discussions is approval voting. I think whoever gets elected should have approval from the majority of the voters, so perhaps using any method after eliminating all candidates (and votes for) who don't have majority approval would probably be better than any single method..
OK, taking the example You provided: For one, it's obviously a "mathematical puzzle" in that it's created so that changing the voting method changes the winner, and each candidate wins with one system. An unlikely scenario. In most cases, there are strong and weak candidates, and we should focus on how to choose between the strong ones. Tsongas looks like an extremist. He's got a loyal following, but most people just hate him. So, should not win. I'll just rewrite the table as if he'd been eliminated: 18 of them prefer Kerrey to Clinton to Harkin to Brown 12 of them prefer Clinton to Harkin to Kerrey to Brown 10 of them prefer Brown to Clinton to Harkin to Kerrey 9 of them prefer Kerrey to Brown to Harkin to Clinton 4 of them prefer Harkin to Clinton to Kerrey to Brown 2 of them prefer Harkin to Brown to Kerrey to Clinton
Looks to me that the winner should be either Kerrey or Clinton, as: - Kerrey has nearly half the first place votes (very strong primary support), no other candidate comes close - Clinton has high overall rating - neither Brown nor Harkin scores high Actually, Harkin looks like a "harmless" candidate that is approved of but doesn't get strong support. If Kerrey were eliminated, Clinton would get over half the first places. OTOH, for Clinton to get more first places than Kerrey, both Brown and Harkin would have to be eliminated.
Clinton and Kerrey both have high enough score that either would probably be OK by most people, where Brown and Harkin both lack a large, strong following and/or approval.
Also note that Clinton would win the "Least of the evils" -vote where each round the most last places is eliminated:)
Perhaps a combination of approval voting with instant runoff: people get to rank from one down as many candidates as they approve of, then candidates not approved by at least half the voters are dropped and instant runoff count performed for the rest, possibly resulting in some votes becoming "empty" in the process (no candidate left on the count approved by the vote). Kerrey would win that one.
Why combine approval with another method? Obvious idiots, whatever their following, are removed as if they never were choices anyway. I would rather call this "approval based elimination" combined with whatever favors primary support.
Oh yes, I can't say how I'd like Condorcet. I do know that I don't like the person who wrote the website You referred to, which biases me against Condorcet (can't be good if it's strong support consists of singleminded "Mine is best, the rest are nothing" idiots).
Also, I'm biased by the fact that we have three strong parties (instead of two) in Finland. Makes a world of difference.
Slashdot Mirror
Macromedia doesn't provide a 64bit version of flash plugin, therefore I can't have it. Web looks quite different without flash...
synaptic runs just fine under fedora as well. But I don't want a graphical tool - I mostly use my linux boxes over ssh anyway. Yum also supports apt repos, but for as long as fedora core stays rpm-based, switching is more trouble than worth for me.
Also, for some reason I don't like apt-get or apt-cache.. aptitude is pretty much the only apt-based management tool I really like.
As a long-time redhat user I've come to like rpm a lot. I like spec-files, I like rpm, I like yum.
But, deb's dependency management is a little finer grained. And aptitude is a great tool.
So, if I could get my familiar rpm and yum commandlines, aptitude, and deb dependencies I guess it'd be the best of both worlds for me.
One huge problem for rpm-based distroes is of course that each of them has different packages and dependency trees. Ever tried using five yum repositories and hoping they just somehow manage to get along? As debian has larger number of packages from single source than any of the rpm-based distroes, and debian maintainers seem to handle dependencies pretty well, the resulting repository-hell might be a little easier to handle. But that's not a problem with the package format - it's a problem with repositories and maintainers.
Open source hackers may well be numerous, but how many of those is good in software development? Remember that the process of creating something for proof of concept, or demonstration, or to solve a specific single problem (which three cover most OSS projects), and even developing without external pressures of budget, manpower and schedule, is very different from disciplined, goal-oriented professional software development.
While you can fix many problems by adjusting development process to counter them, the most important issue becomes getting the developers to follow the process (and having a working process in the first place).
I'd say most hackers are better suited in more R&D oriented projects where just getting something to work is the goal, but few of them are so well suited to actual product development.
I'm just ordering a Tyan Thunder K8S Pro. It has PCI-X, dual GbE (+100Mbps one), onboard SATA RAID (although I'm not at all sure if I'm going to use the very limited RAID capabilities). There's a SCSI version as well, but I'm rather taking SATA and upgrading later with a separate SATA RAID host adapter.
I admit that on Intel side there would be more server motherboards to choose from, but it seems to me that Tyan is offering pretty much everything on Opteron boards as well.
one word: selinux
3ware 9508S with 5 Maxtor DM+9 200GB disks in RAID5 configuration under WinXP has typical read-speed of about 40MB/s when you have several apps accessing the disk at the same time, and clocks a little over 100MB/s when you close the apps and benchmark the system.
Don't know if that's fast or not - just that it's enough for me.
Depending on where in Europe you're moving back, the step down might not be THAT drastic. At least Sweden has 10 & 100Mbps connections, and Finland has 10Mbps (not everywhere, but I doubt trees need it - maybe the bears would).
Still, now that I've had 10Mbps (symmetric, ethernet) for three years, I've come to yearn for more.. 100Mbps would be nice.
What do you mean with HPLJ4m+ having been "the last of the *REALLY* beefy HP printers"?
I've been very happy with my HPLJ4050TN, which is the next generation of LJ4-series afaik..
No, I don't have any intention of buying an inkjet printer of any kind, brand or model. I'm happy with b&w.
Unless horde has been redone (which would break compatibility with original horde), it's ridden by massive design flaws.
Technically it could be redone - while the existing design flaws are fundamental, the concept itself doesn't have such technical problems that it couldn't be fixed by better design.
Whether it has social design flaws is another matter. I think that it's design diverges too much from ed2k design that it should be separated completely.
The particular bill talked about in the article would've required the source to be delivered with any software built specifically for govt. It wouldn't have required source for cots components.
You can get source for commercial software as well as open source software. Just require it in the contract. Probably will cost more than a binary-only version, and there are likely restrictions different from restrictions in open source licenses, but the source should be available still.
I think that kind of a bill is fairly good. The govt will last longer than the company that created that particular piece of software, and it may be in use long after the company has fired those participating in development of that specific software.
Having source is insurance. If license terms allow for it, being able to contract third parties to fix and/or improve the software later on is even better, but just having the source is a lot.
Because the carriers that bring bits to France are not liable for the content. Like telcos are not liable for the content of phone converstations. And ISPs are not liable for content of some other website they don't control.
For as long as a company does business in a country, it's liable. Now, if Yahoo decided to NOT do business in France, it could do so by blocking French IP address ranges.
However, those issues are not very interesting - they're business as usual. What's interesting is that the French can prosecute in France, and if Yahoo or its subsidiaries have presence there, the French authorities can seize their assets if Yahoo doesn't pay the fines; And that the US courts will judge by US laws, thus not forcing Yahoo to pay fines to France by order of a French court.
This seems to establish that a company that is willing to do business worldwide is considered to have established business presence in all countries simply by having a website. But, unless the company has assets in other countries, they're pretty safe. A company selling stuff could easily say they don't do business worldwide by refusing to ship to some countries. But with auctions, you can't do that, so you must block the business by other means.
Of course if Yahoo were to block French IP range, French could use proxy servers in other countries to bypass that block. But, at that point Yahoo (IMO) could reasonably deny doing business in France: they blocked access to (parts of) the website from French IP ranges.
[IANAL and so on]
VideoLan is not a codec but an application. Dirac is a codec. You could stream dirac-encoded video with VideoLan I presume.
But, my DVD collection would already take some 5TB - more than three times my current total diskspace. It'll take 2-3 years before HD sizes grow faster than my DVD collection does, and then some more before I have enough that I can keep them all on disks.
Of course in the 5-10 years it takes for my diskspace to grow beyond my dvd collection (the amount of other data at that point would be somewhere between a rounding error and a hot spare), I'll have some HD-DVD / Bluray / verylargecapacityopticaldisc format videos already, and I yet again think it'll be just some 2-5 years for disk capacity to grow to that size..
Yup. As noted, I'd already need a few TBs to keep my non-hollywood (well, maybe 1-2% Hollywood afterall) commercial video entertainment material on a "Home Entertainment System". And the need for space will grow in the future - I can see it up to at least 100TB with just next gen HD video. And when we're there, there's bound to be something even larger that I want to keep online.
10 years would mean experience with Linux since release of 1.0. I know many who would qualify for that (myself included). 15, OTOH, is longer than Linux has existed..
Just halve the years of experience required (5-8 years) and you should find enough people easily. Add a few years of other *nix OS's and it's just fine.
The article states volunteer work worth about 10-15M, so the entry isn't really much less expensive than Rutan's. Could be that Scaled hadn't burned that much before they started test flights, in which case Scaled Composites would come out cheaper.
Not counting personnel costs isn't really an option unless you want to bet future flights on volunteer work.
HD manufacturers have no excuse for their confusing use of 10's power prefixes as the basic access unit of a disk is a 2's power of a byte.
Networking, OTOH, does have valid excuses. Eg. 64kbps for ISDN means 64000bps, which comes from 8kHz 8b width, and 10Mbps ethernet is indeed 10 million bits per second, not roughly 4.9% more than that - again because the electrical engineers didn't think in 2's power when assigning bandwidth, but rather in conventional 10's powers.
Depends. If I end up with a piece of junk that is partially functional (good mobo & CPU, bad RAM) but can't be used anymore, and will have to pay for all the new parts (or rather part combination), then yes.
If the parts were welded together, where the sticks of RAM have lifetime warranty, I'd of course demand lifetime warranty for the RAM still. So, if it means the shop will replace the whole due to lifetime warranty of RAM, then it's OK for me.
Still, considering I buy the combination together, if there are problems I'll take the whole combination to the shop. Even when I know it's bad RAM - I still take mobo & CPU to the shop with the RAM.
As it happened recently that something failed in a fairly new computer of mine, and even while I diagnosed it as bad RAM but took the whole computer to the shop, they did replace mobo, CPU, RAM, and the PSU - all within the warranty, for no cost at all. PSU was the strange one in the group, but as the report noted that old PSU was giving spikes, I guess it was all for the better. If I had taken just a stick of RAM to the shop, might be I'd have an unstable computer now, causing lots of frustration but no obvious diagnosis.
Nope. It's part of the yearly checkup at the repair shop. My yearly kilometers are low enough that a year passes sooner than the kilometer count between required checkups.
I do KNOW how to check / add / change oil. But why would I do that when it's one of the items on the "yearly checkup" list? I do actually know a lot more about cars than I care for. I could probably do the routine maintenance on an old car if I really had to, and had some reference material to refresh my memory with. I just don't want to.
Similarly, some years back I did a lot of computer maintenance. But these days I design software, and the computers are a tool (as opposed to target of the work), so I'd rather have them "just work". I've got better things to use my time on than to repair a PC when I can have the PC fixed at a fair price by people whose jobs' target is the PC.
As I pay for someone else to fix my PCs (where I could do that myself too, thank you), I can spend the time eg. with family, friends, or even earning money with which to pay for the repairs.
Now, do you do your hair / beard / moustache yourself? Cook from real raw materials (as opposed to fastfood, restaurants, or cook using almost ready ingredients)? Sew your own suits? And so on.
Some people eat out almost always, some cook themselves. Some buy their clothes, some sew themselves. Some fix their own cars or computers, some have others fix them.
It's just a choice about what you do yourself and what not. Nobody can do everything these days. Really. And I like cooking more than fixing a car. I like designing software more than diagnosing broken PCs.
I already treat mobo + CPU + RAM as a single package. Soldering them together wouldn't make much of a difference to me - assuming combinations I want would be available for sale.
If I want a new CPU (at leats 50% speed increase, otherwise it just isn't worth the trouble), I need to get a new mobo because the old mobo won't support the new CPU. And then I need to get new RAM as the old RAM isn't compatible with the new mobo. Or is just so slow that I wouldn't realize the speed increase from CPU change really.
At that point I think about moving the old comp as is to some supportive role (replace old firewall, fileserver, mailserver, or something). So, I'll look at the oldest comp to replace that with the one being replaced currently, and notice that the 7 year old case won't take the new mobo formfactor anymore, and would need new PSU anyway, and so on.
So, I end up buying pretty much a whole new computer unless I'm willing to ditch mobo + CPU + RAM that's at least twice as fast as the oldest one still in use.
For me, a car is a way to get from place A to place B comfortably, economically, and in reasonable time.
In last year I've opened the hood a few times: to add water (and some liquid stuff they sell at gas stops) to windshield cleaning thingie. If there's anything except adding fuel or the cleaning solution, the car goes to a repair shop. Well, changing tires is done in another shop as I know the people working there and know they're going to sell me good tires for reasonable price.
I don't want to open the hood. I don't want to do any repairs. I want to move between locations comfortably, economically, when I want to (as opposed to when the busses/trains go, although I do choose the train quite often).
For these reasons, I'm not going to buy a 20 year old "real car" but rather a new one that will work without trouble for a few years again.
Whose collection of video and music are you talking about? At least not mine..
Last year I spent over 5kEUR on DVDs and CDs. While in the meantime I downloaded similar content by gigabytes and uploaded by terabytes. Oh yes, I do have some Hollywood content, too, but that's all bought, not downloaded, because that I see on TV, too. Non-hollywood content I won't be seeing on TV or hearing on radio, so I have to download, watch/listen, and then buy.
30$ per disc? It's SRP is 19.98, and I paid perhaps some 14$ per disc (plus 14$ for the movie).
It was cheap for a disc, expensive per minute of content. I'm used to paying about 24$ per disc, nowadays often with 3 episodes (of ~23 minutes each). Used to be four episodes, even six, per disc, and the price per disc certainly hasn't come down..
Even this assumes that people try to vote optimally instead of honestly, which I don't think is the case.
:) People who vote for the harmless option as their primary vote don't seem like ones to cast a second vote.
Also, the scenario assumes a few rounds of polls with correct results, and that people would, based on what the polls give as the expected result, approve of more candidates after each poll, and their new approvals would be reflected (correctly) in the next poll.
Also, would really all the Tsongas supporters approve of Kerrey any more than they approve of Clinton? And so on. A small number of people would still "waste" their votes by not approving of any of the top candidates, just approving of what they really would approve of. Wasting a vote by voting for a candidate that has no chance of winning is a valid strategy. Hey, we have Donald Duck often getting a significant approval (thousands of votes - OK, it's not THAT significant, but still). There are empty votes, votes that are given to entities not on the list, and so on.
Note that I also required approval of majority before anything else is considered. In that case, Tsongas has no chance, whatsoever, as he's got ~33% approval. Also, I would think it quite unlikely that each and every person would approve the same number of candidates on their list.. I can easily think of cases where I could approve of two candidates, really hate two, and consider the rest of them as "lesser evils".
Which brings me to a problem in approval voting: I want to express the concepts of "approval", "disapproval", and "indifference" instead of having to lump those I'm indifferent towards to either "approved" or "disapproved" category. However, approval would be better than standard plurality vote.
On Condorcet, I didn't see a "tie" vote anywhere, it looked like it required ranking (no ties) of candidates with the option of not ranking them all (truncated list). If there is the option of a "tie" vote, would it be possible to rank one candidate as the best, not vote (disapprove of) one, and tie the rest (indifference)?
Voting in Finland. The only governmental "single winner" vote in Finland is Presidential election. We have the parliament (200 members), city/county councils (I think 441 total in Finland) with a few to tens of members depending on the size of the city, and the president. Those are the three common publicly elected governmental bodies.
So, in Presidential elections, there is the first round (one vote per person, each vote to one and only one candidate). If someone gets >50%, we're done. Doesn't happen, though.. So, there is a second round where the choice is between the top two candidates.
Not a perfect system, but works. My first candidate has never been elected, but I still go in the first round for my primary candidate. I know that s/he (yes, we've had women as candidates, and I've voted for one in the first round) might have a chance since s/he's usually in the "high-middle" in the polls, close enough to the top candidates.
So, I always vote on the second round, and my vote has always been rather "against" than "for", as the second round candidates up to date have been one I'm indifferent towards and one I strongly disapprove of.
The major difference between US and Finland in presidential elections is thus that in US, there are two candidates that have any chance at all to be elected, and the rest are protest votes, while in Finland, we almost always have at least three candidates with a good realistic chance of getting elected. So, it's not as likely that people would approve of two of the top3 as it is to approve of just one of the top2.
In Your example, I would guess eg. that some of the Tsongas supporters are casting protest votes, and wouldn't care one way or the other about the rest of the candidates, thus casting their votes only for Tsongas. This would eliminate some of the apparent Kerrey support.
Also, there is nothing about Harkin that would make him more than harmless. He's someone the people know not to fuck the country, not to do anything that would change things, not to disrupt the system. Someone who's good for just waiting for the next elections to get a real president. While that's a valid option, I don't think it's good - it's pretty much the same as having no president at all
Still, trying to get an order between the candidates in the example would require sociological study instead of mathematical study. From the patterns presented alone I can't gather what would be the actual votes, as I don't really believe people could rank all candidates. Poeple are likely to have an opinion (strong yes or no) about two or three candidates, and no opinion about the rest.
Oh yes, how can people in Finland vote for candidates not on the list in the first place (note: the votes for invalid candidates are tallied as "invalid" and don't thus effect the vote)?
We write the number of our chosen candidate on a piece of paper. So, people can write whatever they want there.. The votes are first optically read, and then the votes that couldn't be OCR'd are read and input by the vote counters.
One point that isn't addressed in this (or other) mathematical discussions is approval voting.
:)
I think whoever gets elected should have approval from the majority of the voters, so perhaps using any method after eliminating all candidates (and votes for) who don't have majority approval would probably be better than any single method..
OK, taking the example You provided:
For one, it's obviously a "mathematical puzzle" in that it's created so that changing the voting method changes the winner, and each candidate wins with one system. An unlikely scenario. In most cases, there are strong and weak candidates, and we should focus on how to choose between the strong ones.
Tsongas looks like an extremist. He's got a loyal following, but most people just hate him. So, should not win. I'll just rewrite the table as if he'd been eliminated:
18 of them prefer Kerrey to Clinton to Harkin to Brown
12 of them prefer Clinton to Harkin to Kerrey to Brown
10 of them prefer Brown to Clinton to Harkin to Kerrey
9 of them prefer Kerrey to Brown to Harkin to Clinton
4 of them prefer Harkin to Clinton to Kerrey to Brown
2 of them prefer Harkin to Brown to Kerrey to Clinton
Looks to me that the winner should be either Kerrey or Clinton, as:
- Kerrey has nearly half the first place votes (very strong primary support), no other candidate comes close
- Clinton has high overall rating
- neither Brown nor Harkin scores high
Actually, Harkin looks like a "harmless" candidate that is approved of but doesn't get strong support.
If Kerrey were eliminated, Clinton would get over half the first places. OTOH, for Clinton to get more first places than Kerrey, both Brown and Harkin would have to be eliminated.
Clinton and Kerrey both have high enough score that either would probably be OK by most people, where Brown and Harkin both lack a large, strong following and/or approval.
Also note that Clinton would win the "Least of the evils" -vote where each round the most last places is eliminated
Perhaps a combination of approval voting with instant runoff: people get to rank from one down as many candidates as they approve of, then candidates not approved by at least half the voters are dropped and instant runoff count performed for the rest, possibly resulting in some votes becoming "empty" in the process (no candidate left on the count approved by the vote). Kerrey would win that one.
Why combine approval with another method? Obvious idiots, whatever their following, are removed as if they never were choices anyway. I would rather call this "approval based elimination" combined with whatever favors primary support.
Oh yes, I can't say how I'd like Condorcet. I do know that I don't like the person who wrote the website You referred to, which biases me against Condorcet (can't be good if it's strong support consists of singleminded "Mine is best, the rest are nothing" idiots).
Also, I'm biased by the fact that we have three strong parties (instead of two) in Finland. Makes a world of difference.