The actual number of write cycles is not changing.
That would be ideal, but if the designers of the algorithm could predict with perfect accuracy the usage patterns of the drive, they would be more profitably in a different line of business. The "would have to be done anyway" assumes quite a bit of foreknowledge, and "properly implemented" is an assumption that cannot be tested except through time in the field.
Contemplate the following: some periodic process (log, network manager, etc) makes a few 3KB files which aren't grouped through NCQ, which only need to last 30 minutes each. As a hard disk, you don't know that second part, so when is the disk "idle" enough for this cleaning to take place? Active GC on a soon to be vacated block is at least pointless write when the block can just be marked empty. At what point *should* a sector be reused in this situation?
Now, think about the multitude of ways in which even relatively simple things like a nightly append and tar of a column store (or even things like volume shadow copies) can fall outside assumptions about what is, or is not, data that needs to be consolidated.
I would imagine that drives with a "server profile" GC could sell for much more than drives with a "desktop profile" GC...
Not to be cynical, but these new algorithms, if implemented poorly, have the potential to run down the limited number of write cycles on the cells. Not that this could be strategically manipulated in any way...
Eh? Any economical DoS attack on Fidonet would have had to have been distributed, since dialing between and within zones would have been rather expensive (unless you were also a fan of JR or AC). Also, given that all of the LD sites were clearly marked in the nodelist, and that they numbered in the single to double digits for most cities in the world, and that inter-city traffic was sent mostly at local night, you would need 10 friends or operator on any corporate PBX to deny outbound non-local traffic (a slightly modified attack would simultaneously be able to deny all non-local inbound traffic as well).
To take out communication between all the nodes of a city, you would need at most the same number of phone lines as there were listed for that city (typically dozens to low hundreds at most). With some creativity around DID spoofing, abuse of the callback verification systems, faxback systems, etc., an effective attack could be mounted with access to somewhat fewer lines.
(As a side note, anybody know how capitalization works when a sentence begins with a reference to a command? Changing the first letter to a capital actually changes the name of the command, at least in Unix, so it seems like the wrong thing to do.)
Make a house style or pick from something already established? If you follow the rule that chkdsk is not a native English word, it would be italicised to indicate so, and wouldn't change capitalization for the sake of being in our language. However, it's commonly the case that proper names beginning with non-capital letters are routinely both capitalized and not (but never in the same publication). So, at the beginning of a sentence, Mr. Foo de Bar can be referred to on subsequent mentions as "de Bar supports Baz" or "De Bar supports Baz" or "DeBar supports Baz" depending on the publication. In North America, entertainment media run into this for artists with unusual renderings of their names, rather than for other kinds of newsmakers with non-capital first letters in their names as would be more frequent in Europe. For the disk utility known as "chkdsk", my instinct would be to preserve the case but otherwise indicate through typography (bold, underline, typeface) that it's a command to be typed literally, or that it's a proper noun through quotation marks on first mention, depending on context.
Also consider looking at some style manuals and the rules adopted by the Klingon Language Institute for their English language publications since upper and lower case letters can appear almost anywhere in a word or sentence in the latin alphabet representation of that language. Similarly, look at how proper names of First Nations objects, events, processes and locations are rendered in latin print since those more than occasionally lead with punctuation marks.
Don't you have a "Press F12 to select boot device" prompt somewhere between the memory test and the bootloader that would save a bit of time? Are you doing more in the BIOS than just change the boot order to prevent the Windowses from finding each others' drives through hardware enumeration at runtime? Are you elsewise abusing the way that XP and Vista bless secondary hard drives?
Why should I have to prepartition my drive just to keep Windows from messing with other drives?
Because installing multiple operating systems falls into the "expert" category of usage of the products involved? And because it reasonable for Microsoft to assume that individuals who attempt to operate their computing tools in an above-power-user manner by customizing disk partition layout are capable of making intelligent choices about such details inside or outside of the installer tool (as an "expert")?
The application logic here isn't far removed from what RHEL-derived installers do by default on similar hardware configurations, that is, make enough data/swap/root partitions to hold the important stuff, and put boot and GRUB where it will work. In both cases, the vast majority of the target customers do not "have to bloddy worry about it" because the defaults are sensible, and are available to the user during the installation. Anyone who has gone out of their way to do custom things to their disks and partitions should have already critically thought about and understood what their desired final partition and system configuration will look like before embarking on an OS install.
If Microsoft has any fault here, it's for not putting documentation about the behaviour of one of their pre-release products in a sufficiently blindingly obvious location in front of the user so that even self-proclaimed experts are forced to read and understand how the product's characteristics may vary from each of an unlimited number of different users' expectations.
Of course, you knew that some malware will patch their host to retain exclusive access by preventing infection by other malware, right? Depending on what the "few petty IRC-bot infections" consisted of, you may have had a reasonably well inoculated machine protected by someone with an active interest in preventing further infections, especially against well-publicized vectors as were contained in conficker.
BBEdit and UltraEdit are good GUI text processors in that they provide a variety of useful functions for transforming text systematically. They can do regexp, document comparison, advanced templates, whitespace conversion, etc. but do not force the user to go there. gEdit could be in that group when it matures a bit.
To make pretty customer-facing documents (PDFs) that rarely change in substantive content, decent layout tools (applies j/k rules, supports ligatures, can position text and graphic elements through coordinate specification and not just nudging with the arrow keys, non-broken interface for using more than six text styles in one document, tables that breathe correctly, understands color management, supports document bleed without gross frame hacks) include Adobe's InDesign and Illustrator which support file pointers to Word and other documents so that content and presentation can be handled by different people. Recent Corel Draw versions with passable multi-page document support are also good, but not cross platform.
In my experience, Word and sxw are both wastes of time in this area since both novice and expert users spend more time fighting the interface than doing work to produce their desired formatting and layout. They're great tools for generating body copy across organizational boundaries, but poor for making it look good.
Latest vanilla FF3 on OS X here. _Closing_ the tab with the above link increased CPU usage by 4%, which is persistent after 3 minutes. Every time I close and re-open that link in a tab, FF claims and hangs on to another 88 MB without letting go.
What needs to be in a bug report for me to report this without it being binned into WONTFIX or WORKSFORME?
This doesn't need 1000x in variation, just the 10x which is fairly common, but on all sides.
With hearing, we know that humans can be trained to use echolocation without external instrumentation. With sight, we've spent more Homo spp.-hours successfully hunting in pitch black than not. With smell and taste, there are countless food and wine experts who seem to be able to do basic analytical chemistry without kit. So sensitivity is trainable within the hardware limits.
It's unclear if the article refers to 1000x less than average light sensitivity (random human) or maximum recorded conscious light sensitivity (exceptional human), or the maximum theoretical sensitivity based on the hardware. Since it's a fairly mainstream science article, average is as good a guess as any.
Just for kicks, let's say that a trained human could be usefully "10x more sensitive" to light than the average. Further, let's assume that being 3 s.d. out in one direction yields 10x more natural sensitivity (this is not unusual, given the not tiny number of super tasters, etc. who work professionally, and the logarithmic responses of the sense hardware). Let's also toss some inhibition-lowering herbs in there for some uber-sensitive good times. On the other side, let's assume that the scientists detected light at the basal rate, and not at some stressed rate. We know that stressed or determined humans can manifest unusually high strength, loud vocalizations, concentrated emissions of various gases, etc. which span more than four orders of magnitude (e.g. speaking at 30 db versus yelling at 110 db). We know that the ability to generate these signals at various intensities is subject to the usual genetic variation, and also that physical illness and mental state may enhance the generation of (sometimes warm, sometimes smelly) signals. Light emissions may, or may not follow that pattern, but in the absence of information to the contrary, I'll assume we don't need to create a new biology for this case.
If you leave out genetic variations on both sides, training, drugs, and dynamic range of human signals at 10x each easily produce an overlap of the two points of interest which are nominally at 1000x apart.
If you're a religious trainee but have sharpened your senses through meditation for years and are slightly medicated from incense in a room with no light, a random confused stressed out person with a common illness might not seem so dark.
muridae's information from astronomy a few posts up suggests that all these things don't even need to line up to achieve sensitivity.
I'm not saying that any of this is actually the case, but that it's plausible from what we know and that supporting or refuting data would be fairly easily accessible through experimentation.
I'm still having an internal debate about whether the spiritual journey is more complete beginning with nothing to let go, or beginning with everything to let go.
Whatever the answer to that is, I'm in favour of any research or basis for dialogue that could engage individuals to take on more complete views of their lives, whether that's through organized or personal religion or more careful observation and action in other domains of practice. Some of my Eastern religious explorations point to an expansion of physics which could alleviate much unwarranted suffering (and thereby also to some new intellectual and other tools/toys). I'll give up the mystery of dragons at the edge of our physical world to have a chance to seek the dragons of our humane world.
Exactly. They are there to learn (and contribute), not to fight with their tools.
Forcing academics to use different file formats and software than almost everyone else in the community is the very definition of creating intentional incompatibilities. If you want to argue that forcing IIS users to switch to Apache on ideological grounds alone is in the public interest, a very similar argument would hold for the converse switch.
Zero tolerance on any application which does not intentionally include compatibility for every file format ever or any software which has a security issue or bug would objectively exclude almost every piece of non-trivial software from use.
At Canadian campuses, the "defective-by-design solution-in-search-of-a-problem garbage" software, whether it produces ugly things like Word tables, or ugly things like swriter tables, each works well enough for at least 30 percent of the learners, so prohibition on proprietary or open source would force significant numbers of users to fight with or relearn software, rather than applying or enhancing their expertise. This is hard to justify.
Since your argument is based on ideology rather than objective user needs, I'd like to see some arguments or evidence in support of having the IT department compete with the philosophy or political science departments, in the context of enhancing learning achievement for the whole university.
Infrastructure management strategies are important because they are usually an organization's second or third most expensive cost, and often their second or third most valuable asset, both of which link importantly into the organization's bottom line as a determinant of success or failure.
More pragmatically, if the university wants to become an Open University or a medical school or howsoever different in 5 years, and has planned for this in the overall strategic plan, in the absence of a strategic plan IT will not be able to design, deploy and test large infrastructure and wetware appropriately beforehand if it has no strategy to get pieces on the ground, and it will be disadvantaged in the future with respect to providing the new services sustainably.
Sensible strategies also deal with things like not having to buy 200 copies of MS Office from Staples at the last minute, and also not having the OC-3 choke when 3000 computers decide to simultaneously autoupdate. Infrastructure strategies also tell you how not to OMG Zerg RUSH!!!1! into new/other technologies, purchases and implementations without considering organization-wide impacts.
Although many (all?) public universities in North America have not shortage of students awaiting enrollment, xednieht might be thinking of private universities and other specialized post-secondary institutions that are less responsive to market pressures, or which have decided to implement a newspaper publishing model. But even then, the advice provided wouldn't even work well for modern newspapers, as thriving examples such as The Economist point out.
Also, be aware that missions and strategic plans must cascade in any organization of significant size and internal diversity, so one plan will not fit all users. Use good science to ascertain and respond to users' needs, or they will do it themselves.
Yes, zero tolerance on those specifications (or, really, anything not otherwise illegal) would be a mistake since that would make the university non-interoperable with students, faculty, other researchers, industry, government, and the rest of the world. University IT departments (really, all IT departments) should have at least the purpose and goal of enabling the organization to meet its business objectives (efficient output of high quality teaching, research, advocacy, other products) through the effective management and provision of ICT tools. IT departments, like accounting departments and the loading dock, are means to informing and fulfilling other broader strategic objectives, and should rarely be ends unto themselves. More sophisticated IT departments should certainly form strategy around their own operations and aspire to become more than a supporting department, but not at the expense of achieving the organization's regular business objectives.
At one university in which I participate, the CIO decided that nightly six-hour backups of 13 different legacy mainframe systems was inefficient, and increasingly costly as upkeep and replacement parts became increasingly expensive. However, the systems had co-evolved with the business operating procedures of the entire 40-year-old institution, providing great efficiencies at the human business administrator level, such that processing was swift, accurate, and responsive to the needs of the students, faculties and other customers. Everyone on campus had two credentials: the library card and the IT (AIX) login. IT signed with [that system that Oracle bought] on the understanding that the business logic and VT interfaces from all of the legacy systems would be ported over to the new web-based system in two years, such that the data would be seamlessly carried through and users would require minimal retraining on business processes. The vendor-supplied consultants, it was thought, would be able to magically absorb decades of institutional knowledge about the university's documented business processes, without any consideration for the many more legitimate but undocumented exceptions. They did no meaningful consulting with the users of the proposed new system. Oddly at the time, the management faculty applied enough pressure to largely not participate in this migration on the grounds that their system (which replicated the functions of the central system on recent platforms) would not be made more efficient by this plan.
Three years after, data from two of the silos were ported, the interfaces and business logic were completely new, and average administrative processing time in many customer-facing units rose from hours to weeks. Some of the more experienced administrators left in frustration, taking valuable intra- and extra-institutional knowledge with them. Others stayed and became "[platform] translators" who would take old paper forms completed by faculty, staff and students, and manually key them into the various parts of the new system (the system also assigned new terminology to things like "semester", "term", "quarter", "credit hour", etc. which had different meanings elsewhere in hardcopy and departmental regulations). IT deployed a new helpdesk specific to the new platform, and customer-facing staffing for faculty and students (and the queuing area at the registrar's office) had to be more than doubled to deal with the increased volume of exception handling required. This $35 million migration was already $10 million over budget, and perhaps 15 per cent complete. At this time, elsewhere in IT, people were implementing different SSO systems for students, faculty and administrative staff, and yet a different system for wireless access.
Over the next five years, they migrated the remaining silos, disappearing probably over $50 million in various kinds of internal accounts in the process, including federal funding, some of the smaller endowed chairs, etc. (For three years, our relatively small research group couldn't access around $2 million in f
Why is it a good thing to test against three different HTML5 implementations now and again later, as opposed to waiting for the standard to finalize?
I thought we've been down this road before around a decade ago and extra standard functionality implemented in funny external libraries that also couldn't agree on how the standard was implemented. SVG reminds me somewhat of how VRML was (mis-)handled, while some of bleeding edge parts of HTML5 working best/only with Google Gears feels a lot like the competing client side Java plugins and system-specific hackage required to be compatible with each of the three major JREs.
As an aside, I'm starting to dislike that Firefox's defaults are to check for updates to itself or addons at every invocation, to the point where using more than around five addons will almost guarantee a two minute penalty on each invocation. Combined with the session restoration, crash recovery and db integrity checks, it's been tempting to get coffee while waiting for Firefox to boot, rather than after pressing the power button. Some colleagues are switching from Firefox back to IE or Safari because of the drain on productivity and group meetings, and because of the recent Firefox 3 builds which creep to 80MB per tab for squirrel mail, without letting go until the process exits.
Slashdot Mirror
K-Meleon was pretty awesome on the (dated) 32MB Lappy 486 when IE3 and Opera were choking hard on DHTML.
The actual number of write cycles is not changing.
That would be ideal, but if the designers of the algorithm could predict with perfect accuracy the usage patterns of the drive, they would be more profitably in a different line of business. The "would have to be done anyway" assumes quite a bit of foreknowledge, and "properly implemented" is an assumption that cannot be tested except through time in the field.
Contemplate the following: some periodic process (log, network manager, etc) makes a few 3KB files which aren't grouped through NCQ, which only need to last 30 minutes each. As a hard disk, you don't know that second part, so when is the disk "idle" enough for this cleaning to take place? Active GC on a soon to be vacated block is at least pointless write when the block can just be marked empty. At what point *should* a sector be reused in this situation?
Now, think about the multitude of ways in which even relatively simple things like a nightly append and tar of a column store (or even things like volume shadow copies) can fall outside assumptions about what is, or is not, data that needs to be consolidated.
I would imagine that drives with a "server profile" GC could sell for much more than drives with a "desktop profile" GC...
Not to be cynical, but these new algorithms, if implemented poorly, have the potential to run down the limited number of write cycles on the cells. Not that this could be strategically manipulated in any way...
Eh? Any economical DoS attack on Fidonet would have had to have been distributed, since dialing between and within zones would have been rather expensive (unless you were also a fan of JR or AC). Also, given that all of the LD sites were clearly marked in the nodelist, and that they numbered in the single to double digits for most cities in the world, and that inter-city traffic was sent mostly at local night, you would need 10 friends or operator on any corporate PBX to deny outbound non-local traffic (a slightly modified attack would simultaneously be able to deny all non-local inbound traffic as well).
To take out communication between all the nodes of a city, you would need at most the same number of phone lines as there were listed for that city (typically dozens to low hundreds at most). With some creativity around DID spoofing, abuse of the callback verification systems, faxback systems, etc., an effective attack could be mounted with access to somewhat fewer lines.
(As a side note, anybody know how capitalization works when a sentence begins with a reference to a command? Changing the first letter to a capital actually changes the name of the command, at least in Unix, so it seems like the wrong thing to do.)
Make a house style or pick from something already established? If you follow the rule that chkdsk is not a native English word, it would be italicised to indicate so, and wouldn't change capitalization for the sake of being in our language. However, it's commonly the case that proper names beginning with non-capital letters are routinely both capitalized and not (but never in the same publication). So, at the beginning of a sentence, Mr. Foo de Bar can be referred to on subsequent mentions as "de Bar supports Baz" or "De Bar supports Baz" or "DeBar supports Baz" depending on the publication. In North America, entertainment media run into this for artists with unusual renderings of their names, rather than for other kinds of newsmakers with non-capital first letters in their names as would be more frequent in Europe. For the disk utility known as "chkdsk", my instinct would be to preserve the case but otherwise indicate through typography (bold, underline, typeface) that it's a command to be typed literally, or that it's a proper noun through quotation marks on first mention, depending on context.
Also consider looking at some style manuals and the rules adopted by the Klingon Language Institute for their English language publications since upper and lower case letters can appear almost anywhere in a word or sentence in the latin alphabet representation of that language. Similarly, look at how proper names of First Nations objects, events, processes and locations are rendered in latin print since those more than occasionally lead with punctuation marks.
Don't you have a "Press F12 to select boot device" prompt somewhere between the memory test and the bootloader that would save a bit of time? Are you doing more in the BIOS than just change the boot order to prevent the Windowses from finding each others' drives through hardware enumeration at runtime? Are you elsewise abusing the way that XP and Vista bless secondary hard drives?
Why should I have to prepartition my drive just to keep Windows from messing with other drives?
Because installing multiple operating systems falls into the "expert" category of usage of the products involved? And because it reasonable for Microsoft to assume that individuals who attempt to operate their computing tools in an above-power-user manner by customizing disk partition layout are capable of making intelligent choices about such details inside or outside of the installer tool (as an "expert")?
The application logic here isn't far removed from what RHEL-derived installers do by default on similar hardware configurations, that is, make enough data/swap/root partitions to hold the important stuff, and put boot and GRUB where it will work. In both cases, the vast majority of the target customers do not "have to bloddy worry about it" because the defaults are sensible, and are available to the user during the installation. Anyone who has gone out of their way to do custom things to their disks and partitions should have already critically thought about and understood what their desired final partition and system configuration will look like before embarking on an OS install.
If Microsoft has any fault here, it's for not putting documentation about the behaviour of one of their pre-release products in a sufficiently blindingly obvious location in front of the user so that even self-proclaimed experts are forced to read and understand how the product's characteristics may vary from each of an unlimited number of different users' expectations.
I'm pretty sure the service tech has a good way to detect fakes with high accuracy.
At least OpenDNS is clueful enough to filter out BS like the following:
http://www.domainnotfound.ca/bellassist/dnsassist/content/ErrorPage/_iceUrlFlag=15?_IceUrl=true&q=www.non-existent-domain.com-INSERT-MALWARE-HERE%3CSCRIPT%3E
Of course, you knew that some malware will patch their host to retain exclusive access by preventing infection by other malware, right? Depending on what the "few petty IRC-bot infections" consisted of, you may have had a reasonably well inoculated machine protected by someone with an active interest in preventing further infections, especially against well-publicized vectors as were contained in conficker.
> thousands of people have tried to take this beast over in the past few years
Which groups of timelines are you from? For most of us, Conficker is not even one Earth year old.
BBEdit and UltraEdit are good GUI text processors in that they provide a variety of useful functions for transforming text systematically. They can do regexp, document comparison, advanced templates, whitespace conversion, etc. but do not force the user to go there. gEdit could be in that group when it matures a bit.
To make pretty customer-facing documents (PDFs) that rarely change in substantive content, decent layout tools (applies j/k rules, supports ligatures, can position text and graphic elements through coordinate specification and not just nudging with the arrow keys, non-broken interface for using more than six text styles in one document, tables that breathe correctly, understands color management, supports document bleed without gross frame hacks) include Adobe's InDesign and Illustrator which support file pointers to Word and other documents so that content and presentation can be handled by different people. Recent Corel Draw versions with passable multi-page document support are also good, but not cross platform.
In my experience, Word and sxw are both wastes of time in this area since both novice and expert users spend more time fighting the interface than doing work to produce their desired formatting and layout. They're great tools for generating body copy across organizational boundaries, but poor for making it look good.
Latest vanilla FF3 on OS X here. _Closing_ the tab with the above link increased CPU usage by 4%, which is persistent after 3 minutes. Every time I close and re-open that link in a tab, FF claims and hangs on to another 88 MB without letting go.
What needs to be in a bug report for me to report this without it being binned into WONTFIX or WORKSFORME?
> pardon anybody that brakes [sic] the law if enough time has passed since the offence [sic] was committed?
In most countries, we've not found statutes of limitation objectionable enough to do anything meaningful about them.
> You might not like it, but having seen/heard about alternatives on such a proeminent [sic] place does make a difference
What was on the second billboard during your last OS install?
http://www.customizegoogle.com/ will filter out ExpertSexChange (and other useless sites) from your Google results.
Clippy: It looks like you're implementing a Duff's Device. Would you like help?
Get help with implementing the Duff's Device
Just type the construct without help
This doesn't need 1000x in variation, just the 10x which is fairly common, but on all sides.
With hearing, we know that humans can be trained to use echolocation without external instrumentation. With sight, we've spent more Homo spp.-hours successfully hunting in pitch black than not. With smell and taste, there are countless food and wine experts who seem to be able to do basic analytical chemistry without kit. So sensitivity is trainable within the hardware limits.
It's unclear if the article refers to 1000x less than average light sensitivity (random human) or maximum recorded conscious light sensitivity (exceptional human), or the maximum theoretical sensitivity based on the hardware. Since it's a fairly mainstream science article, average is as good a guess as any.
Just for kicks, let's say that a trained human could be usefully "10x more sensitive" to light than the average. Further, let's assume that being 3 s.d. out in one direction yields 10x more natural sensitivity (this is not unusual, given the not tiny number of super tasters, etc. who work professionally, and the logarithmic responses of the sense hardware). Let's also toss some inhibition-lowering herbs in there for some uber-sensitive good times. On the other side, let's assume that the scientists detected light at the basal rate, and not at some stressed rate. We know that stressed or determined humans can manifest unusually high strength, loud vocalizations, concentrated emissions of various gases, etc. which span more than four orders of magnitude (e.g. speaking at 30 db versus yelling at 110 db). We know that the ability to generate these signals at various intensities is subject to the usual genetic variation, and also that physical illness and mental state may enhance the generation of (sometimes warm, sometimes smelly) signals. Light emissions may, or may not follow that pattern, but in the absence of information to the contrary, I'll assume we don't need to create a new biology for this case.
If you leave out genetic variations on both sides, training, drugs, and dynamic range of human signals at 10x each easily produce an overlap of the two points of interest which are nominally at 1000x apart.
If you're a religious trainee but have sharpened your senses through meditation for years and are slightly medicated from incense in a room with no light, a random confused stressed out person with a common illness might not seem so dark.
muridae's information from astronomy a few posts up suggests that all these things don't even need to line up to achieve sensitivity.
I'm not saying that any of this is actually the case, but that it's plausible from what we know and that supporting or refuting data would be fairly easily accessible through experimentation.
I'm still having an internal debate about whether the spiritual journey is more complete beginning with nothing to let go, or beginning with everything to let go.
Whatever the answer to that is, I'm in favour of any research or basis for dialogue that could engage individuals to take on more complete views of their lives, whether that's through organized or personal religion or more careful observation and action in other domains of practice. Some of my Eastern religious explorations point to an expansion of physics which could alleviate much unwarranted suffering (and thereby also to some new intellectual and other tools/toys). I'll give up the mystery of dragons at the edge of our physical world to have a chance to seek the dragons of our humane world.
Exactly. They are there to learn (and contribute), not to fight with their tools.
Forcing academics to use different file formats and software than almost everyone else in the community is the very definition of creating intentional incompatibilities. If you want to argue that forcing IIS users to switch to Apache on ideological grounds alone is in the public interest, a very similar argument would hold for the converse switch.
Zero tolerance on any application which does not intentionally include compatibility for every file format ever or any software which has a security issue or bug would objectively exclude almost every piece of non-trivial software from use.
At Canadian campuses, the "defective-by-design solution-in-search-of-a-problem garbage" software, whether it produces ugly things like Word tables, or ugly things like swriter tables, each works well enough for at least 30 percent of the learners, so prohibition on proprietary or open source would force significant numbers of users to fight with or relearn software, rather than applying or enhancing their expertise. This is hard to justify.
Since your argument is based on ideology rather than objective user needs, I'd like to see some arguments or evidence in support of having the IT department compete with the philosophy or political science departments, in the context of enhancing learning achievement for the whole university.
Infrastructure management strategies are important because they are usually an organization's second or third most expensive cost, and often their second or third most valuable asset, both of which link importantly into the organization's bottom line as a determinant of success or failure.
More pragmatically, if the university wants to become an Open University or a medical school or howsoever different in 5 years, and has planned for this in the overall strategic plan, in the absence of a strategic plan IT will not be able to design, deploy and test large infrastructure and wetware appropriately beforehand if it has no strategy to get pieces on the ground, and it will be disadvantaged in the future with respect to providing the new services sustainably.
Sensible strategies also deal with things like not having to buy 200 copies of MS Office from Staples at the last minute, and also not having the OC-3 choke when 3000 computers decide to simultaneously autoupdate. Infrastructure strategies also tell you how not to OMG Zerg RUSH!!!1! into new/other technologies, purchases and implementations without considering organization-wide impacts.
Although many (all?) public universities in North America have not shortage of students awaiting enrollment, xednieht might be thinking of private universities and other specialized post-secondary institutions that are less responsive to market pressures, or which have decided to implement a newspaper publishing model. But even then, the advice provided wouldn't even work well for modern newspapers, as thriving examples such as The Economist point out.
Mod parent up.
Also, be aware that missions and strategic plans must cascade in any organization of significant size and internal diversity, so one plan will not fit all users. Use good science to ascertain and respond to users' needs, or they will do it themselves.
Yes, zero tolerance on those specifications (or, really, anything not otherwise illegal) would be a mistake since that would make the university non-interoperable with students, faculty, other researchers, industry, government, and the rest of the world. University IT departments (really, all IT departments) should have at least the purpose and goal of enabling the organization to meet its business objectives (efficient output of high quality teaching, research, advocacy, other products) through the effective management and provision of ICT tools. IT departments, like accounting departments and the loading dock, are means to informing and fulfilling other broader strategic objectives, and should rarely be ends unto themselves. More sophisticated IT departments should certainly form strategy around their own operations and aspire to become more than a supporting department, but not at the expense of achieving the organization's regular business objectives.
At one university in which I participate, the CIO decided that nightly six-hour backups of 13 different legacy mainframe systems was inefficient, and increasingly costly as upkeep and replacement parts became increasingly expensive. However, the systems had co-evolved with the business operating procedures of the entire 40-year-old institution, providing great efficiencies at the human business administrator level, such that processing was swift, accurate, and responsive to the needs of the students, faculties and other customers. Everyone on campus had two credentials: the library card and the IT (AIX) login. IT signed with [that system that Oracle bought] on the understanding that the business logic and VT interfaces from all of the legacy systems would be ported over to the new web-based system in two years, such that the data would be seamlessly carried through and users would require minimal retraining on business processes. The vendor-supplied consultants, it was thought, would be able to magically absorb decades of institutional knowledge about the university's documented business processes, without any consideration for the many more legitimate but undocumented exceptions. They did no meaningful consulting with the users of the proposed new system. Oddly at the time, the management faculty applied enough pressure to largely not participate in this migration on the grounds that their system (which replicated the functions of the central system on recent platforms) would not be made more efficient by this plan.
Three years after, data from two of the silos were ported, the interfaces and business logic were completely new, and average administrative processing time in many customer-facing units rose from hours to weeks. Some of the more experienced administrators left in frustration, taking valuable intra- and extra-institutional knowledge with them. Others stayed and became "[platform] translators" who would take old paper forms completed by faculty, staff and students, and manually key them into the various parts of the new system (the system also assigned new terminology to things like "semester", "term", "quarter", "credit hour", etc. which had different meanings elsewhere in hardcopy and departmental regulations). IT deployed a new helpdesk specific to the new platform, and customer-facing staffing for faculty and students (and the queuing area at the registrar's office) had to be more than doubled to deal with the increased volume of exception handling required. This $35 million migration was already $10 million over budget, and perhaps 15 per cent complete. At this time, elsewhere in IT, people were implementing different SSO systems for students, faculty and administrative staff, and yet a different system for wireless access.
Over the next five years, they migrated the remaining silos, disappearing probably over $50 million in various kinds of internal accounts in the process, including federal funding, some of the smaller endowed chairs, etc. (For three years, our relatively small research group couldn't access around $2 million in f
Why is it a good thing to test against three different HTML5 implementations now and again later, as opposed to waiting for the standard to finalize?
I thought we've been down this road before around a decade ago and extra standard functionality implemented in funny external libraries that also couldn't agree on how the standard was implemented. SVG reminds me somewhat of how VRML was (mis-)handled, while some of bleeding edge parts of HTML5 working best/only with Google Gears feels a lot like the competing client side Java plugins and system-specific hackage required to be compatible with each of the three major JREs.
As an aside, I'm starting to dislike that Firefox's defaults are to check for updates to itself or addons at every invocation, to the point where using more than around five addons will almost guarantee a two minute penalty on each invocation. Combined with the session restoration, crash recovery and db integrity checks, it's been tempting to get coffee while waiting for Firefox to boot, rather than after pressing the power button. Some colleagues are switching from Firefox back to IE or Safari because of the drain on productivity and group meetings, and because of the recent Firefox 3 builds which creep to 80MB per tab for squirrel mail, without letting go until the process exits.