I sort of don't understand why you thought I didn't Not at all. Actually I'm agreeing with you, but in an obnoxious manner. It's more like looking at different ends of the same elephant.
very professional and specialized system, running on usually very professional and specialized equipment, where errors need to be minimal, or at least recovered from gracefully.
slop it in and everything gets tested in a very buggy world. If anything ever goes beserk out of proportion to its cause, something is bad wrong and needs to be fixed. Use the bugs that you can see to root out the bugs you cannot see.
Which one of us is right? The beauty of Open Source is that I don't even care! If either of us is right, we both gain.
Windows source code isn't like a virus or something.
Humerous but more than a little accurate. Implicit in the source will be subtle assumptions about the way things work that are not precisely accurate. Once contaminated by these inacuracies, errors will be made.
It's not what we don't know that is the problem. It's what we know that ain't so.
It's a nice hack but fails if I get login.c and the compiler from independent sources. That hack depends on my getting both pieces from the same source.
The classic example is to have a cashier that trades tickets for dollars and a ticket taker that takes the tickets. If you have one that does both you have to trust that one. With two of them who are not in cahoots with each other, very little trust is required.
Closed source will never be like that simply because the sheer price of developing millions of lines of code to near perfect standards is astounding
There is no "silver bullet". Ada IIRC manages to make some of what were semantic errors into syntactic errors which are caught by the compiler.
Nothing's perfect. However there is an enormous variation in the degrees of imperfection. What Open Source does do is to make much better use of available resources. It's tempting to think of Open Source as a cheap alternative, but I can assure you that there is no way of rearranging IBM's psyche so that IBM would embrace "cheap alternative". Look at RedHat. For customers of their boxed RedHat Professional Server, they've essentially forced those customers to pay a lot more for yesterday's software or to pay a lot less for tomorrow's software. I can't really blame RedHat, but notice that the RedHat Label went with the expensive line. Also notice that while Fedora may be a sink for RedHat funds, its existence adds much to the value of the RedHat label.
The only argument this guy makes is that it is not good to use software from people you don't trust.
True. Obvious. What's maybe not so obvious is the less you have to trust the vendor, the better.
Contrast: [ ] Always trust Microsoft [ ] Always trust RedHat Why the ^%*^&%&* should I have to trust RedHat? Methinks that an essential part of any con game is that the victim must trust the con artist.
A small and ever-decreasing percentage of users compile their own binaries, let alone check the result. [Emphasis added]
Compare: 50% of 10 is 5.05% of 100,000 is 50 I'd much rather have.05% of 100,000 checking than 50% of 10.
It takes very few to notice something peculiar and investigate. The malefactors get caught out if anybody notices anything. Since anybody can examine everything of interest, it would be extremely difficult for a malefactor to actually accomplish much of anything against Open Source.
I have found hidden files in directories by looking at the location of images and looking in those directories. Those directories and some of the files were not linked to anywere. [Emphasis added]
Errrrrr...... If they show in a generated index of one of those directories, they are linked to! This includes any parent directories and guessable filenames and subdirectories.
"I didn't make an explicit link" in no way implies that the files are not linked.
My philosophy on security: If you actually need security, you'd better be paranoid.
And for reference, yeah, I too have stuck random files up on my site for a friend to grab. But never when it would have mattered if someone else randomly found those files.
You touched on a critical point. It is not necessary to secure everything, and probably counter-productive to attempt to do so. With terms like secure email and secure websites running around, people will click on things. If you think of email and websites like postcards and posting notes on Park bulliten boards, you maybe treat things with a hint of suspicion.
I for one believe enough is enough, and it would be ethically ok to go ahead and create such a worm. All we'd have to do is infect in the same way this new virus does, and run arbitrary code to destroy the virus. Thoughts?
I agree, except.... You have to be good, very good. Do it once. Do it again. It's not really reapeatable. Bad copies will be made. Unforseen consequences. Good copies will have the same unforseen consequences, just slower.
you cannot then hide behind its being free to explain its limitations.
Limitations? If over a span of several months/years the only way to take down a Microsoft Windows machine was something as esoteric as specially built IPv6 stack, you'd have a point. Seems like OpenBSD is in much better shape to play with the big boys than Microsoft.
anything that can be done by port knocking can generally be done better with information sent on a single port.
Right, but the advantage of port knocking is that the single port doesn't exist until the secret knock is given. Something as simple as three ports, spaced one second apart, opening the real port 3 seconds later would be extremely time-expensive to brute-force, particularly if source-ips are black-listed for wrong sequences.
I not only have a responsibility as SysAdmin of my enterprise of protecting my own users. I must do everything in my power to protect others from the actions of nodes on my network! [Emphasis added]
Correct priorities, even from a purely selfish standpoint. Getting viruses. You can't control that (unless you control the rest of the universe). Sending viruses. You can control that. The universe of what is reasonable to send should be vastly smaller that what what is reasonable to receive. Basically, you don't send strange attachments.
Of course anything that stems the tide, particularly if it can be done cheaply and easily, helps enormously.
Please, please tell me you're kidding, and that you don't work logged in as a domain admin unless you're doing admin tasks?
Dunno about him, but I'd probably give you a heart attack. I'm posting this from work where I'm logged in as "root" which is the domain administrator. When I leave I don't log off or turn the machine off. The problem with admin tasks is that they tend to come up in the middle of something else, and they tend to have a bit of urgency to them. Not good when you're in the middle of several things you don't really want to have to shut down. The user/administrator distinction in Microsoft Windows is not nearly as useful as the user/root distinction in Unix. Microsoft Windows is not geared to being logged into multiple systems as multiple users simultaneously.
FWIW, my users are very good about not even attempting to open things they shouldn't. No problem with virus signatures that aren't quite up-to-date. No virus with the name "anti". Figure that the antivirus software will do a good job of catching the ones that don't matter and completely miss "the big one". The one I'm waiting for uses the antivirus as an integral part of the virus, possibly by "cleaning" a bunch of good data files and documents.
Losing a day's work is recoverable. Losing a few years or decades is not.
But how do you mitigate the effects of having a virus "corrupt" all your documents? Even if you catch it right away and restore from last night's backups (after checking ALL your computers for infection) you still lose an entire day's worth of work for many departments. That's a big setback.... We came up with the obvious: have good backups,
Have good backups.
Good. The classic time to find out you need the backup is as you are destroying it by writing bad stuff on top of old-but-good stuff.
Backups. Plural. Figure three minimum. One's no good. One you write instead of read or otherwise destroy in the recovery attempt. This leaves one for you to recover from.
The only backup that does anything but waste time is the one you will actually need. The only thing you can be reasonably sure of is that when you need backup, something isn't right or working right. What you want is to give yourself the best odds of recovery with the minimum expenditure of time, money and effort. The thing to avoid is an elaborate expensive backup system that works perfectly... until you need it. You're much better off with multiple poor-quality but independent backups.
They talk a good game but let's face it, if you don't run any services on any platform it's about as secure as an OpenBSD install is out of the box. That's not exactly securing the code through audit, it's just locking down a box.
Errrr, not exactly. I'm far from being expert in such things, but OpenBSD seems to be designed for remote administration that must pass through hostile territory (man in the middle, etc). You're behind a firewall, but it's your enemy's firewall and he knows how to use it. You get a fast basic install on site. All the configuration and lockdown is done remotely in a context where the internet is friendly and the LAN is hostile.
There's a lot more to security than just not running vulnerable service.
Good marketing is wonderful... Sounds really simple (sigh) but like all seemingly simple things it's incredibly hard to get it right.
Right. Good marketing is wonderful. Things that claim to be good marketing... usually are not.
I'm always amused by the phrase "reinventing the wheel". The first wheel wasn't invented as a wheel. It was some sorta roundy thingee that things would kinda roll around on. It takes a lot of reinventing to get rid of the bumps. It takes a lot of reinventing to even see the bumps.
Some of it (Linux, *BSD) seems headed that direction, with stuff like real multi-user, chroot, jail, etc. Basically, my browser should only be able to mess with itself, not my stuff, and not with anyone else's stuff. It is secure when I can run something exploitable, unpatched with impunity. Sure the browser can be taken over, but only within that browser. This does require hard and obvious distinctions between what such as browsers claim (which is internal to the browser) and the reality which must be external to the browser and unfakable by the browser. It's a bit like "this email claims to be from tony@foo.bar" versus "this email came from tony@foo.bar". It would help immensely if error messages stuck to what was known and not try to be "user friendly" and guess as the ultimate cause. Microsoft is one of the worst offenders in that everything seems to blame any problems on something else.
Combined with those three "features" I don't see there can be any lack of future viruses.
Sir, methinks you are an optimist. Everything in Microsoft Windows, including the ads, encourages everyone to click on everything. Seems like it would take much less computer to show the available information than to put up a pretty icon.
From my experince most Mechanical Engineers would call someone to change their tire for them.
Can't say that I blame them. It's dirty, messy, and there's no sense of accomplishment. Further, there's stuff any grunt at a service station knows that is not taught to Mechanical Engineers, particularly if said Mechanical Engineers do not like to "get their hands dirty".
We teach our kids to be vary of strangers offering them candy. [Couldn't resist]
We should teach our kids to be wary of strange software offering them "candy". Adults too. Maybe even more so.
I sincerely believe that all this "userfriendlyness" (is that a word?) in computers is for the worse. Point made, but I have to disagree. Userfriendlyness is good, but rather deceptive in that it takes a lot of skill and design and work and rework to make minor accomplishments. What happens all too often is that a con job is done that claims to be "user friendly" -- "User Friendly" as in a false sense of security.
It is necessary that a few people, preferably self selected, understand how their computers work. For most everybody else, it is only necessary to understand enough to do whatever it is they need to do. This does mean that a lot of skill and expertise will go unnoticed and unappreciated by the masses, except for some vague sense of the feel of it.
Slashdot Mirror
I sort of don't understand why you thought I didn't
Not at all. Actually I'm agreeing with you, but in an obnoxious manner. It's more like looking at different ends of the same elephant.
very professional and specialized system, running on usually very professional and specialized equipment, where errors need to be minimal, or at least recovered from gracefully.
slop it in and everything gets tested in a very buggy world. If anything ever goes beserk out of proportion to its cause, something is bad wrong and needs to be fixed. Use the bugs that you can see to root out the bugs you cannot see.
Which one of us is right? The beauty of Open Source is that I don't even care! If either of us is right, we both gain.
Windows source code isn't like a virus or something.
Humerous but more than a little accurate.
Implicit in the source will be subtle assumptions about the way things work that are not precisely accurate. Once contaminated by these inacuracies, errors will be made.
It's not what we don't know that is the problem. It's what we know that ain't so.
It's a nice hack but fails if I get login.c and the compiler from independent sources. That hack depends on my getting both pieces from the same source.
The classic example is to have a cashier that trades tickets for dollars and a ticket taker that takes the tickets. If you have one that does both you have to trust that one. With two of them who are not in cahoots with each other, very little trust is required.
Closed source will never be like that simply because the sheer price of developing millions of lines of code to near perfect standards is astounding
There is no "silver bullet". Ada IIRC manages to make some of what were semantic errors into syntactic errors which are caught by the compiler.
Nothing's perfect. However there is an enormous variation in the degrees of imperfection. What Open Source does do is to make much better use of available resources. It's tempting to think of Open Source as a cheap alternative, but I can assure you that there is no way of rearranging IBM's psyche so that IBM would embrace "cheap alternative". Look at RedHat. For customers of their boxed RedHat Professional Server, they've essentially forced those customers to pay a lot more for yesterday's software or to pay a lot less for tomorrow's software. I can't really blame RedHat, but notice that the RedHat Label went with the expensive line. Also notice that while Fedora may be a sink for RedHat funds, its existence adds much to the value of the RedHat label.
The only argument this guy makes is that it is not good to use software from people you don't trust.
True. Obvious.
What's maybe not so obvious is the less you have to trust the vendor, the better.
Contrast:
[ ] Always trust Microsoft
[ ] Always trust RedHat
Why the ^%*^&%&* should I have to trust RedHat?
Methinks that an essential part of any con game is that the victim must trust the con artist.
A small and ever-decreasing percentage of users compile their own binaries, let alone check the result. [Emphasis added]
.05% of 100,000 is 50 .05% of 100,000 checking than 50% of 10.
Compare:
50% of 10 is 5
I'd much rather have
It takes very few to notice something peculiar and investigate. The malefactors get caught out if anybody notices anything. Since anybody can examine everything of interest, it would be extremely difficult for a malefactor to actually accomplish much of anything against Open Source.
I have found hidden files in directories by looking at the location of images and looking in those directories. Those directories and some of the files were not linked to anywere. [Emphasis added]
Errrrrr......
If they show in a generated index of one of those directories, they are linked to! This includes any parent directories and guessable filenames and subdirectories.
"I didn't make an explicit link" in no way implies that the files are not linked.
My philosophy on security: If you actually need security, you'd better be paranoid.
And for reference, yeah, I too have stuck random files up on my site for a friend to grab. But never when it would have mattered if someone else randomly found those files.
You touched on a critical point. It is not necessary to secure everything, and probably counter-productive to attempt to do so. With terms like secure email and secure websites running around, people will click on things. If you think of email and websites like postcards and posting notes on Park bulliten boards, you maybe treat things with a hint of suspicion.
I for one believe enough is enough, and it would be ethically ok to go ahead and create such a worm. All we'd have to do is infect in the same way this new virus does, and run arbitrary code to destroy the virus. Thoughts?
....
I agree, except
You have to be good, very good.
Do it once. Do it again. It's not really reapeatable.
Bad copies will be made. Unforseen consequences.
Good copies will have the same unforseen consequences, just slower.
what everyone needs is an easy-to-install, easy-to-update, and easy-to-configure linux
... /usr/local/bin/knx-hdinstall
You mean like
Uhhhhhm... You do realize that the ABA is a private entity, not a government organization???
Actually, I don't. It's one of those meta-organizations which effectively govern government. Very much the wrong place to have any hint of corruption.
I don't think I would get anywhere with something like the Unamerican Bar Association, regardless of membership.
I would assume that you put the door knockers on the outside of the NAT box, not on the inside.
One practical use would be to forward port 80 to the proper internal server based on the prior knock-sequence on the firewall.
you cannot then hide behind its being free to explain its limitations.
Limitations?
If over a span of several months/years the only way to take down a Microsoft Windows machine was something as esoteric as specially built IPv6 stack, you'd have a point. Seems like OpenBSD is in much better shape to play with the big boys than Microsoft.
anything that can be done by port knocking can generally be done better with information sent on a single port.
Right, but the advantage of port knocking is that the single port doesn't exist until the secret knock is given. Something as simple as three ports, spaced one second apart, opening the real port 3 seconds later would be extremely time-expensive to brute-force, particularly if source-ips are black-listed for wrong sequences.
I not only have a responsibility as SysAdmin of my enterprise of protecting my own users. I must do everything in my power to protect others from the actions of nodes on my network! [Emphasis added]
Correct priorities, even from a purely selfish standpoint.
Getting viruses. You can't control that (unless you control the rest of the universe).
Sending viruses. You can control that. The universe of what is reasonable to send should be vastly smaller that what what is reasonable to receive. Basically, you don't send strange attachments.
Of course anything that stems the tide, particularly if it can be done cheaply and easily, helps enormously.
Please, please tell me you're kidding, and that you don't work logged in as a domain admin unless you're doing admin tasks?
Dunno about him, but I'd probably give you a heart attack. I'm posting this from work where I'm logged in as "root" which is the domain administrator. When I leave I don't log off or turn the machine off. The problem with admin tasks is that they tend to come up in the middle of something else, and they tend to have a bit of urgency to them. Not good when you're in the middle of several things you don't really want to have to shut down. The user/administrator distinction in Microsoft Windows is not nearly as useful as the user/root distinction in Unix. Microsoft Windows is not geared to being logged into multiple systems as multiple users simultaneously.
FWIW, my users are very good about not even attempting to open things they shouldn't. No problem with virus signatures that aren't quite up-to-date. No virus with the name "anti". Figure that the antivirus software will do a good job of catching the ones that don't matter and completely miss "the big one". The one I'm waiting for uses the antivirus as an integral part of the virus, possibly by "cleaning" a bunch of good data files and documents.
Losing a day's work is recoverable. Losing a few years or decades is not.
...
... until you need it. You're much better off with multiple poor-quality but independent backups.
But how do you mitigate the effects of having a virus "corrupt" all your documents? Even if you catch it right away and restore from last night's backups (after checking ALL your computers for infection) you still lose an entire day's worth of work for many departments. That's a big setback.
We came up with the obvious: have good backups,
Have good backups.
Good. The classic time to find out you need the backup is as you are destroying it by writing bad stuff on top of old-but-good stuff.
Backups. Plural. Figure three minimum. One's no good. One you write instead of read or otherwise destroy in the recovery attempt. This leaves one for you to recover from.
The only backup that does anything but waste time is the one you will actually need. The only thing you can be reasonably sure of is that when you need backup, something isn't right or working right. What you want is to give yourself the best odds of recovery with the minimum expenditure of time, money and effort. The thing to avoid is an elaborate expensive backup system that works perfectly
They talk a good game but let's face it, if you don't run any services on any platform it's about as secure as an OpenBSD install is out of the box. That's not exactly securing the code through audit, it's just locking down a box.
Errrr, not exactly.
I'm far from being expert in such things, but OpenBSD seems to be designed for remote administration that must pass through hostile territory (man in the middle, etc). You're behind a firewall, but it's your enemy's firewall and he knows how to use it. You get a fast basic install on site. All the configuration and lockdown is done remotely in a context where the internet is friendly and the LAN is hostile.
There's a lot more to security than just not running vulnerable service.
You control the results of the search, you are in a position to profit from it.
And as you distort those results in order to profit and are perceived to do so, those results become just so much spam.
Amazingly effective. To the point of being annoyingly effective.
It's a shoot first, ask questions second tactic.
Good marketing is wonderful ... Sounds really simple (sigh) but like all seemingly simple things it's incredibly hard to get it right.
... usually are not.
Right.
Good marketing is wonderful.
Things that claim to be good marketing
I'm always amused by the phrase "reinventing the wheel". The first wheel wasn't invented as a wheel. It was some sorta roundy thingee that things would kinda roll around on. It takes a lot of reinventing to get rid of the bumps. It takes a lot of reinventing to even see the bumps.
Some of it (Linux, *BSD) seems headed that direction, with stuff like real multi-user, chroot, jail, etc.
Basically, my browser should only be able to mess with itself, not my stuff, and not with anyone else's stuff. It is secure when I can run something exploitable, unpatched with impunity. Sure the browser can be taken over, but only within that browser. This does require hard and obvious distinctions between what such as browsers claim (which is internal to the browser) and the reality which must be external to the browser and unfakable by the browser.
It's a bit like "this email claims to be from tony@foo.bar" versus "this email came from tony@foo.bar". It would help immensely if error messages stuck to what was known and not try to be "user friendly" and guess as the ultimate cause. Microsoft is one of the worst offenders in that everything seems to blame any problems on something else.
Combined with those three "features" I don't see there can be any lack of future viruses.
Sir, methinks you are an optimist. Everything in Microsoft Windows, including the ads, encourages everyone to click on everything. Seems like it would take much less computer to show the available information than to put up a pretty icon.
From my experince most Mechanical Engineers would call someone to change their tire for them.
Can't say that I blame them. It's dirty, messy, and there's no sense of accomplishment. Further, there's stuff any grunt at a service station knows that is not taught to Mechanical Engineers, particularly if said Mechanical Engineers do not like to "get their hands dirty".
We teach our kids to be vary of strangers offering them candy. [Couldn't resist]
We should teach our kids to be wary of strange software offering them "candy".
Adults too. Maybe even more so.
I sincerely believe that all this "userfriendlyness" (is that a word?) in computers is for the worse.
Point made, but I have to disagree. Userfriendlyness is good, but rather deceptive in that it takes a lot of skill and design and work and rework to make minor accomplishments. What happens all too often is that a con job is done that claims to be "user friendly" -- "User Friendly" as in a false sense of security.
It is necessary that a few people, preferably self selected, understand how their computers work. For most everybody else, it is only necessary to understand enough to do whatever it is they need to do. This does mean that a lot of skill and expertise will go unnoticed and unappreciated by the masses, except for some vague sense of the feel of it.