Darwin was the first person to notice evolution, but he did not refine it to a science. Gould's theory of punctuated equilibrium has been disputed by Dennit, Dawkins and others. Quite simply, once you consider the core principles of evolution, it just doesn't make sense.
Evolution is not a force out of the ether that is triggered by a struggle for survival, it is a process that is built into the mechanisms that allow us to reproduce.
Here is a simple description of the modern theory of evolution:
You need to have:
Variation (one organism does not have all of the genes present in the species)
Mutation (when an organism reproduces, its offspring will have some probability of being genetically different from the parent.
Copying fidelity (genetic material must be copied from one generation to the next)
Substrate Neutrality (It doesn't care whether you're a bunch of cells, a bunch of carbon nanotubes, or a bunch of lisp code, so long as the above principles hold.
Some organisms will reproduce a lot, and some will reproduce a little or not at all. Natural selection is the idea that the genetic makeup of an organism contributes to its ability to reproduce (a lot, a little, or not at all). The organism whose ancestors and offspring survive for a long time can be said retrospectively to have a high level of "fitness" relative to the environmnent.
Humans most certainly experience mutation, variation, and copying fidelity today, and we reproduce subject to perfectly real selection pressures.
Some examples of modern selection pressures that might influence the ability of certain individuals to reproduce:
The ability to survive on little sleep. The advantages are obvious. While your less-fit counterpart is snoozing, you are making money, meeting people of the opposite sex, planning for your future, working toward that promotion, playing with the kids, etc.
The inflexibility of institutions toward pregnant women: A woman may put off reproducing until she has a degree or a certain level of income, thereby increasing the risk of birth defects in her children and the risk of miscarriage.
The point is, though we have yet to see the precise evolutionary outcome of our modern selection pressures, they nonetheless exist.
Anyone who doubts this should consider the following truism: Evolution is cleverer than you are.
The premise that Linus started out with, that "development is done by humans" isn't really the gist of his later argument.
Linus' argument boils down to a statement about engineering, not about human nature.
Linus' point is that from a purely engineering standpoint it makes more sense for the network of maintainers to be distributed such that:
those closest to Linus have the most intimate knowledge of the architectural direction.
those further removed from Linus have the most intimate knowledge of the details of the subsystem in question -- it is true that Linus happens to trust these people more, but what he really means is that he knows from past experience that they share his knowledge of the architecture.
Linus made the point that the reason he often rejects patches is because they come from people who lack adequate intimate knowledge of the high-level architectural direction to create a usable patch. He believes that if these patches were submitted to one of the maintainers in the area that the patch addresses (such as networking protocols, for example), then feedback could be given to the submitter about necessary changes and tweaks more efficiently than if Linus simply dropped it on the floor or ignored it -- the efficiency is gained by the transfer of knowledge back to the writer of the patch. Of course, the acknowledgement of inefficiency was precisely what Rob Landley was getting at. Linus' concesion that a Misc. Maintainer would be a good idea shows that Rob's point was well taken and that the discussion was in fact beneficial.
Linus' larger point was that if patches are submitted to the right person, then the 'scalability' of the system won't be limited. He mentions 'trust', but I think he trusts the people who he knows have the necessary high-level architectural knowledge to write a good patch. This issue is about intelligence and efficiency however, not trust.
I agree with some of your points, but do not consider them unique to the ASP model.
1. If an ASP is dumb enough to disregard the cost of end-user training when deciding to undertake a major redesign, then it doesn't deserve to be in business (who knows, maybe this is what happened to some of the.com failures).
2. It would seem to me that upgrading 10,000 hard drives would cost more than upgrading 50 offices from isdn to a T1, for example. 10000 * $50 (for a new hard drive) = $500,000. That doesn't include paying someone to install the drives (suppose it takes 30 minutes to install each one and the person installing them makes $5 per hour, that is an additional $25K. Compare this to the cost of upgrading to a T1 in 50 offices. Call the phone company, buy less than $1000 worth of hardware, and pay $800 per month for the T1 service. Multiplied by 50 offices, that makes $50K upfront costs and $40K per month in bandwidth costs. For the same price as upgrading everyone's hard drive, the organization can get a full year's worth of high bandwidth access for every employee.
If you consider the additional cost of IT support that will be required for the new hard drives, the cost of software installation, and the upfront cost of the software itself, then the non-ASP solution has quite an initial pricetag.
The ASP solution has some addtitional benefits: 1) It is probably not necessary to upgrade the bandwidth at every facility immediately. 2) Minimal IT support cost. 3) ASP runs fine on older computers, thereby extending the time until obsolescense by roughly 30%. 4)Future software upgrades will not require an additional upfront investment.
Also, think about what the company can do with all the money that it saves upfront. This is capital that can be invested and used for other purposes rather than tying it up in 10,000 hard drives that will be obsolete themselves in a few years.
Now, if you couple the aforementioned advantages with the idea that in its ideal form an ASP is also a service, then you can see how powerful the model is.
Regarding your point 3, I don't think office templates are a good example of customizability. Further, there is no reason that a well-designed ASP application couldn't be equally customizable.
4. Recently there were a few worms that were said to cost businesses millions of dollars. Why? Mostly because lazy sysadmins didn't follow the warnings about the worm and apply the existing patch. If you're the sysadmin for an ASP, you'd better protect your servers. With an ASP this is more likely to be the case b/c the web-server is a key part of the business and not an afterthought. Of course, any company that harms its customers due to sloppy system administration of its servers deserves to go out of business, and this applies to ASP and non-ASP companies equally.
One question: Any plan to create a net install so that I can make a floppy and download everything else from the net? Some of the other distros do this, and it makes it very convenient for me, someone who hasn't yet sprung for a cd burner.
I may post this again in a different thread just in case it turns out to be hidden down here...
The points that you make don't really apply. First, number 1 is both a good thing and a bad thing. Ultimately, if a company chooses to disregard its customers' wishes it won't have customers for long, ASP or not.
Point 2: ASP software needs to use a reasonable amount of bandwidth (no extraneous images, etc.). This is ilke saying "most offices have hard drives that are too small to handle the latest release of office". If ASP software offers enough value, it will be worth businesses' while to upgrade their connectivity.
Point 3: This is also not specific to ASP software. I've used a lot of un-customizable software, both ASP and non-asp.
Point 4: Your concern here was addressed by the poster whom you replied to. There were a lot of start-up ASP's that went under and caused people to worry about confidentiality, etc. However, this is not a flaw of the ASP model, merely an aspect of the risk involved with doing business with a small (newly established) company.
The no-brainer is getting caught up in criticizing the "internet bubble" rather than looking at the ASP model as a good way to solve many of the problems associated with traditional software distribution and licensing.
There would surely be some overlap, however consider the way in which readers may discover new topics and perspectives that they would not otherwise have been aware of. Collaborative Filtering is often described as a way to find one's "virtual soul-mate", i.e., the person who has identical preferences. With CF, one's horizons would constantly be broadened. Topics are really quite restrictive. Imagine, too, if users would collaboratively rate new story submissions! The./ editors would become obsolete. In fact, most every story could find its way to the message board, and sooner or later./ would likely cover any imaginable topic, initially due to the fact that most/. readers have other (non-nerdy) interests, and secondly due to the influx of new readers who wanted to share in the discussion of the non-nerd-oriented topics.
How "open" do you think the source needs to be in order to reap the benefits of the bazaar, i.e., which of the following is your favorite license(GPL, LGPL, Sun's Public License, etc.)?
Lastly, what will Sun's biggest challenges be in moving forward with Open Source Software in the future?
I would guess that there are a lot of 'communities' hidden within Slashdot. Factors that differentiate individuals include such things as:
the kind of discussion preferred (long vs short comments, lots of links in comments vs few links, + or - humor, etc.)
level of technical detail appreciated in posts (do I want to read 30 lines of C or not, do I have a strong physics background, etc.)
The fact is, there is no telling what kinds of communities might emerge. A site like/. with CF would be an absolutely unprecedented net community. Right now,/. is managed from the top down with the simple (albeit democratic) majority determining which information is most likely to be viewed by others. CF would allow the true flavor of/. to emerge to the benefit of each and every user.
You are exactly on target. The kind of system that you describe is known as "collaborative filtering", and it is used successfully in sites such as MovieLens [movielens.umn.edu].
You are partially correct about the effect of such as system on the "groupthink problem". Users who always modded down (or +foe'ed) everyone whose ideas they disagreed with would surely get a slashdot consisting of only comments that they agreed with, but users who modded comments according to how well they contributed to the discussion would be richly rewarded: their slashdot woudl be a place of informative discussions and insightful comments.
If you are curious about CF, check out movielens (link above).
An Intermediate Step toward CF: a manifesto
on
Slashdot Code Update
·
· Score: 4, Interesting
Being able to flag users as "friend" or "foe" is a great intermediate step toward the kind of collaborative filtering-based moderation system that I imagine for Slashdot.
Imagine the following:
Slashdot 'notices' that a bunch of other users who share a lot of 'friends' with you have modded up a posting by someone who is not on your 'friends' list. Slashdot notifies you of the posting, you read it, submit a comment, and add the user to your 'friends' list. You have thus discovered a worthwhile posting that you may have missed had you been filtering out low-scoring comments.
If Slashdot created a true collaborative filtering-based moderation system, then moderation as we know it would cease to exist, and in its place hundreds of closely intertwined 'communities' of like-minded readers would emerge, and the quality of discussion on slashdot (as perceived by its readers) would grow enormously.
To satisfy new readers or those who had not taken the time to express their preferences, comments could be 'scored' according to aggregate moderation across communities. The key of CF would be that everyone would be a moderator all of the time, and everyone's moderations would effect whose comments they themselves saw in the future.
If Microsoft suddenly changes how their browser handles downloaded files, tens of thousands (perhaps hundreds of thousands? any webpage which downloads files) of webpages "designed for IE" will have to be rewritten.
Uhh... I don't know of any sites that fit into this category, do you?
rain on the poor guy's parade
on
This is IT?
·
· Score: 2
Why is it that the typical Slashdotter's comment is so negative?
To me, it's a cool invention. I want one. Will I spend $3,000 on one? Most likely not. But of course, that is just the introductory price.
Slashdot Mirror
A lot of good it has done Netscape in the marketplace.
Taco says it's really more of a tip jar than anything else. If you don't tip, fine.
Ways to help /.
Post good comments
Click through on banners occasionally
Sign up as a subscriber
Read /. to give 'em lots of banner impressions.
Cowboy Neal
sysadmin skill
server power
storage space
code upkeep cost
bandwidth cost
I remember back when
Why is that? I've always wondered...
Evolution is not a force out of the ether that is triggered by a struggle for survival, it is a process that is built into the mechanisms that allow us to reproduce.
Here is a simple description of the modern theory of evolution:
You need to have:
Variation (one organism does not have all of the genes present in the species)
Mutation (when an organism reproduces, its offspring will have some probability of being genetically different from the parent.
Copying fidelity (genetic material must be copied from one generation to the next)
Substrate Neutrality (It doesn't care whether you're a bunch of cells, a bunch of carbon nanotubes, or a bunch of lisp code, so long as the above principles hold.
Some organisms will reproduce a lot, and some will reproduce a little or not at all. Natural selection is the idea that the genetic makeup of an organism contributes to its ability to reproduce (a lot, a little, or not at all). The organism whose ancestors and offspring survive for a long time can be said retrospectively to have a high level of "fitness" relative to the environmnent.
Humans most certainly experience mutation, variation, and copying fidelity today, and we reproduce subject to perfectly real selection pressures.
Some examples of modern selection pressures that might influence the ability of certain individuals to reproduce:
The ability to survive on little sleep. The advantages are obvious. While your less-fit counterpart is snoozing, you are making money, meeting people of the opposite sex, planning for your future, working toward that promotion, playing with the kids, etc.
The inflexibility of institutions toward pregnant women: A woman may put off reproducing until she has a degree or a certain level of income, thereby increasing the risk of birth defects in her children and the risk of miscarriage.
The point is, though we have yet to see the precise evolutionary outcome of our modern selection pressures, they nonetheless exist.
Anyone who doubts this should consider the following truism: Evolution is cleverer than you are.
There is a book dedicated to answering this question... it's called Inc Yourself, and it's been useful to me.
This is true... I was once a newbie Mac user. I will never be a repeat Mac customer.
it's not AGs, it's AsG
The premise that Linus started out with, that "development is done by humans" isn't really the gist of his later argument.
Linus' argument boils down to a statement about engineering, not about human nature.
Linus' point is that from a purely engineering standpoint it makes more sense for the network of maintainers to be distributed such that:
those closest to Linus have the most intimate knowledge of the architectural direction.
those further removed from Linus have the most intimate knowledge of the details of the subsystem in question -- it is true that Linus happens to trust these people more, but what he really means is that he knows from past experience that they share his knowledge of the architecture.
Linus made the point that the reason he often rejects patches is because they come from people who lack adequate intimate knowledge of the high-level architectural direction to create a usable patch. He believes that if these patches were submitted to one of the maintainers in the area that the patch addresses (such as networking protocols, for example), then feedback could be given to the submitter about necessary changes and tweaks more efficiently than if Linus simply dropped it on the floor or ignored it -- the efficiency is gained by the transfer of knowledge back to the writer of the patch. Of course, the acknowledgement of inefficiency was precisely what Rob Landley was getting at. Linus' concesion that a Misc. Maintainer would be a good idea shows that Rob's point was well taken and that the discussion was in fact beneficial.
Linus' larger point was that if patches are submitted to the right person, then the 'scalability' of the system won't be limited. He mentions 'trust', but I think he trusts the people who he knows have the necessary high-level architectural knowledge to write a good patch. This issue is about intelligence and efficiency however, not trust.
but now I use LFS
I agree with some of your points, but do not consider them unique to the ASP model.
.com failures).
1. If an ASP is dumb enough to disregard the cost of end-user training when deciding to undertake a major redesign, then it doesn't deserve to be in business (who knows, maybe this is what happened to some of the
2. It would seem to me that upgrading 10,000 hard drives would cost more than upgrading 50 offices from isdn to a T1, for example. 10000 * $50 (for a new hard drive) = $500,000. That doesn't include paying someone to install the drives (suppose it takes 30 minutes to install each one and the person installing them makes $5 per hour, that is an additional $25K. Compare this to the cost of upgrading to a T1 in 50 offices. Call the phone company, buy less than $1000 worth of hardware, and pay $800 per month for the T1 service. Multiplied by 50 offices, that makes $50K upfront costs and $40K per month in bandwidth costs. For the same price as upgrading everyone's hard drive, the organization can get a full year's worth of high bandwidth access for every employee.
If you consider the additional cost of IT support that will be required for the new hard drives, the cost of software installation, and the upfront cost of the software itself, then the non-ASP solution has quite an initial pricetag.
The ASP solution has some addtitional benefits: 1) It is probably not necessary to upgrade the bandwidth at every facility immediately. 2) Minimal IT support cost. 3) ASP runs fine on older computers, thereby extending the time until obsolescense by roughly 30%. 4)Future software upgrades will not require an additional upfront investment.
Also, think about what the company can do with all the money that it saves upfront. This is capital that can be invested and used for other purposes rather than tying it up in 10,000 hard drives that will be obsolete themselves in a few years.
Now, if you couple the aforementioned advantages with the idea that in its ideal form an ASP is also a service, then you can see how powerful the model is.
Regarding your point 3, I don't think office templates are a good example of customizability. Further, there is no reason that a well-designed ASP application couldn't be equally customizable.
4. Recently there were a few worms that were said to cost businesses millions of dollars. Why? Mostly because lazy sysadmins didn't follow the warnings about the worm and apply the existing patch. If you're the sysadmin for an ASP, you'd better protect your servers. With an ASP this is more likely to be the case b/c the web-server is a key part of the business and not an afterthought. Of course, any company that harms its customers due to sloppy system administration of its servers deserves to go out of business, and this applies to ASP and non-ASP companies equally.
this is undoubtedly one of the best posts I've read on slashdot in quite a while... let me see... how do I add Arandir to my "friends" list?
One question: Any plan to create a net install so that I can make a floppy and download everything else from the net? Some of the other distros do this, and it makes it very convenient for me, someone who hasn't yet sprung for a cd burner.
I may post this again in a different thread just in case it turns out to be hidden down here...
Point 2: ASP software needs to use a reasonable amount of bandwidth (no extraneous images, etc.). This is ilke saying "most offices have hard drives that are too small to handle the latest release of office". If ASP software offers enough value, it will be worth businesses' while to upgrade their connectivity.
Point 3: This is also not specific to ASP software. I've used a lot of un-customizable software, both ASP and non-asp.
Point 4: Your concern here was addressed by the poster whom you replied to. There were a lot of start-up ASP's that went under and caused people to worry about confidentiality, etc. However, this is not a flaw of the ASP model, merely an aspect of the risk involved with doing business with a small (newly established) company.
The no-brainer is getting caught up in criticizing the "internet bubble" rather than looking at the ASP model as a good way to solve many of the problems associated with traditional software distribution and licensing.
i pasted them in from a web page... oops.
I was just lazy... go ahead and mod it down.
Lastly, what will Sun's biggest challenges be in moving forward with Open Source Software in the future?
A list of licenses follows for reference...
The GNU General Public License (GPL)
The GNU Library or "Lesser" Public License (LGPL)
The BSD license
The MIT license
The Artistic license
The Mozilla Public License v. 1.0 (MPL)
The Qt Public License (QPL)
The IBM Public License
The MITRE Collaborative Virtual Workspace License (CVW License)
The Ricoh Source Code Public License
The Python license (CNRI Python License)
The Python Software Foundation License
The zlib/libpng license
The Apache Software License
The Vovida Software License v. 1.0
The Sun Industry Standards Source License (SISSL)
The Intel Open Source License
The Mozilla Public License 1.1 (MPL 1.1)
The Jabber Open Source License
The Nokia Open Source License
The Sleepycat License
The Nethack General Public License
The Common Public License
The Apple Public Source License
The X.Net License
The Sun Public License
The Eiffel Forum License
The W3C License
The Motosoto License
The Open Group Test Suite License
the kind of discussion preferred (long vs short comments, lots of links in comments vs few links, + or - humor, etc.)
level of technical detail appreciated in posts (do I want to read 30 lines of C or not, do I have a strong physics background, etc.)
The fact is, there is no telling what kinds of communities might emerge. A site like /. with CF would be an absolutely unprecedented net community. Right now, /. is managed from the top down with the simple (albeit democratic) majority determining which information is most likely to be viewed by others. CF would allow the true flavor of /. to emerge to the benefit of each and every user.
You are partially correct about the effect of such as system on the "groupthink problem". Users who always modded down (or +foe'ed) everyone whose ideas they disagreed with would surely get a slashdot consisting of only comments that they agreed with, but users who modded comments according to how well they contributed to the discussion would be richly rewarded: their slashdot woudl be a place of informative discussions and insightful comments.
If you are curious about CF, check out movielens (link above).
Being able to flag users as "friend" or "foe" is a great intermediate step toward the kind of collaborative filtering-based moderation system that I imagine for Slashdot.
Imagine the following:
Slashdot 'notices' that a bunch of other users who share a lot of 'friends' with you have modded up a posting by someone who is not on your 'friends' list. Slashdot notifies you of the posting, you read it, submit a comment, and add the user to your 'friends' list. You have thus discovered a worthwhile posting that you may have missed had you been filtering out low-scoring comments.
If Slashdot created a true collaborative filtering-based moderation system, then moderation as we know it would cease to exist, and in its place hundreds of closely intertwined 'communities' of like-minded readers would emerge, and the quality of discussion on slashdot (as perceived by its readers) would grow enormously.
To satisfy new readers or those who had not taken the time to express their preferences, comments could be 'scored' according to aggregate moderation across communities. The key of CF would be that everyone would be a moderator all of the time, and everyone's moderations would effect whose comments they themselves saw in the future.
For those who speculate about Miguel's
intentions as they pertain to his work on the
Mono project, simply consider the meaning
of "Mono" in Spanish...
Shouldn't the "top retailers" section actually be called "retailers with the hardest-to-remember URL's"?
Uhh... I don't know of any sites that fit into this category, do you?
Why is it that the typical Slashdotter's comment is so negative?
To me, it's a cool invention. I want one. Will I spend $3,000 on one? Most likely not. But of course, that is just the introductory price.