Slashdot Mirror
Do You Write Backdoors?
quaxzarron asks: "I had a recent experience where one of our group of programmers wrote backdoors on some web applications we were developing, so that he could gain access to the main hosting server when the application went live. This got me thinking about how we are dependent on the integrity of the coders for the integrity of our applications. Yet in this case a more than casual glance would allow us to identify potentially malicious code. How does this work when the clients are companies who can't perform such checks - either because they don't know how, or because the code is too large or too complex? How often do companies developing code officially sanction backdoors...even if means calling them 'security features'? How often has the Slashdot crowd put a backdoor in the code they were developing either officially or otherwise? How sustainable is the 'trust' between the developer and the client?"
But, thats not to say I lack ethics, am a cracker, or am out to get my client.
How many times have we all heard, duhh.... I forgot my admin password, but I cant reinstall, I need the data.
So yes, I backdoor, and I document it internally (hardcopy stored in a safe). Its just an extra insurance policy for when some moron that I worked for 6 years ago does something stupid.
That said, coding backdoors for the sake of getting access to a web farm so you can host your own services is certainly a bad thing(tm). But hell, what are you gonna do? Everyone backdoors. Don't believe me? Watch someone 'in the know' log in to a random windows box using the System account and come talk to me.
Some of the apps I make have the option to "allow" a backdoor by setting a flag (default on). The client can turn it off if he/she really doesn't trust me, but in most cases they find it useful in case I ever have to bugfix the systems and/or they lose their own passwords.
I know a person who owns his own company and writes code on a for-hire basis. He puts in timed expiration code such that if they don't pay him within 30 days of delivery, his code de-activates.
Where I work, we do similar things, but our motivation is to ensure that users are always running the latest version of our frequently updated codebase. We, as developers, do have the ability to run expired code via the backdoor.
Praying for the end of your wide-awake nightmare.
"I had a recent experience where one of our group of programmers wrote backdoors on some web applications we were developing, so that he could gain access to the main hosting server when the application went live."
/Mcdonnel-Douglas embedded the F15 Eagle fighter plane with a backdoor in its computer systems so if its ever used against the USA it will strangely malfunction.
Its like that theory that BAE
Unlikely, but interesting concept all the same!
Trust and loyalty used to be my main focus... I trusted that those stock options i was offered instead of a chunk of salary would be good, and the company trusted that i would deliver good software, on-time.
I fulfilled my part of the bargain, but when it came to stock option maturity time, I got laid-off.. The company is still in business interestingly enough, and now posting profits even.
Who do you trust, and how is that trust repaid? I can tell you I no longer have the same sense of loyalty and trust in my employer. Companies are paying on average HALF of what they were for the same work 2 years ago.. Trust... works both ways or it doesn't work at all...
But when you think about it, all leaving a backdoor in a system does for you is to provide an opportunity of accessing a system in a way that you shouldn't be. This can lead to trouble down the line.
Clearly, there are legitimate uses for backdoors (to use in case of emergencies, etc.), but unless the backdoor is documented someplace for others in the software development group to be aware of, it's likely the kind of backdoor that is simply not ethical to implement, since it's only usable by one person.
I'm sure people can provide examples that disprove this, but for the majority of situations, as a developer, having a backdoor in a system can only lead to a security breach at some point ...
I am working on an app for the govt, and yes, I have programmed in a backdoor login, as it's very useful for testing and development.
However, the following are true:
1) management knows full well of its existence
2) BY DEFAULT, it is turned off in any build
3) it is NEVER to be deployed turned on
I think it's a good rule of thumb.
http://kered.org
Though it wasn't explicitely mentioned in the question, I feel that such a situation may be more common when the developers are hired as temporary / part-time help. In this case, you are a client and the developers may be looking to get something more. If you have your own in-house developers, they'll have more stake in the company and the project, and surely would care more about security - both the security of the software and the security of their job. A hired hand could code the backdoor then move on before you ever notice. Your own developers would be more hesitant to do this because if and when it gets noticed, they'll still be easily found in the cube on the third floor, east wing.
Maybe a good idea would be to bring on a full time development staff and pay them good money so they don't feel the need to try to get something more. Oh, and tell me where to send my resume once you create these new full time positions.
How are you going to keep them down on the farm once they've seen Karl Hungus?
Backdoors can be a good insurance policy, and their theoretical presence might guarentee your continued employment, but if your employers find them, I can guarentee you won't be working there for much longer :P
I work for a small startup that specializes in custom web-applications for indy record labels and small-time bands and clubs. Our main product is a all-in-one web-app that will allow the customer to manage their shows, news, mailing list and numurous other things.
We offer several levels of this product, one being shared (get 1 account on our servers) which we control, standalone, and custom standalone (the standalones go on their own servers.) The latter two are designed to have one back-door login account for myself and the other programmer to go in there and edit their settings or database if the customer breaks something.
So there is my 2 cents. Yes, I put small backdoors in my company's web-apps per boss's request.
What really concerned me though was when we were supposed to store credit card numbers encrypted in the database and I used a simple replacement cypher as a placeholder. Then, when I later asked about putting real encryption routines in was told "we aren't going to do that".
So customers are really in the dark when it comes to the security of their software.
Rich
Doing things like
Anyways, my point is that most backdoors put in by developers seem to be accidental rather than intentional.
C - A language that combines the speed of assembly with the ease of use of assembly.
ok, i don't think there's a backdoor, but i know windows xp comes installed with a special microsoft tech support user. how do they get to use that user to fix problems? that's what i don't understand. it's really odd i think. i wouldn't be surprised if microsoft started putting backdoors into their software that only closed when you entered a unique serial number that bounced back from an online serial number database. similar to the way Q3A uses the cd key. although i know there are some hacks to the Q3A cd key to allow you to use a pirated copy, so this may not work. i just wouldn't put it past microsoft to do something like that.
please me, have no regrets.
When I was a consultant at a Big 6 firm (back in the day), a colleague of mine wrote a Windows app for a client. He added code that would cause the application to stop working after a certain date, so that if the client doesn't pay their invoices, he won't update the code, and the app will simply stop working.
I personally considered this to be very unprofessional, and probably not legal, but he claimed that it was perfectly legit. Of course, the client didn't know this, and he never told them (they did pay their invoices on time).
Definitely not my style, but it is evidence to me that it is done on a regular basis.
I don't but two guys here did just that last year. It was a customer facing website for a large multi-national corporation. The "backdoor" was caught before going live but they were fired with extreme prejudice.
Speak truth to power.
I know that probably was a joke but.. If you think the NSA needs a key to get to your data you need to go read up on the amount of computing power they have in their hands. I recommend "Puzzle Palace" and "Body of Secrets" from James Bamford. Really interesting stuff. They could basically pick through your weak built in encryption like Rosie Odonnell picks through a rack of ribs. Their computers would literally sigh from boredom.
In my experience, back doors tend to be "in the mind of the programmer" as opposed to code that was physically made to be that way. It's not that the programmer sat there and added a way to hack in-- it's just the fact that since he knows how the system works, he knows how to gain access if need be. After all, just about everything electronic can be exploited at some point.
I think you really have to consider the intent. If someone tries to go into the code and deliberately put a back door in w/o proper decision authority, that should be considered unethical. If the client wants it tighter than Fort Knox and loses the key, it's not your fault.
The guy decided to be a dick about it and not pay him the money he deserved.
Fortunately for him he put a backdoor in. He told the guy about it. Once activated the system would not work until a password only he knew was entered.
His payment was promptly received. (He got the idea from a movie.)
-Clio
Karma: Bad (mostly from not giving a fuck)
Blog: http://clintjcl.wordpress.com
Backdoors are very hard to justify and also adds coding (as mentioned here before).
However, I've been involved in projects where we've added an easter egg.
Why? I don't know, it's fun, easy and it's cool when you can show someone that you actually was part of the development. Ofcourse, these projects were inhouse product development. I would probably not consider doing such a thing in a customer's product that I'm working with, besides, that's not what they are paying me for.
I've written backdoors into web apps I've written, but usually because SOMEONE has to have absolute access to the data. It usually leads to some admin tools I've also written for the site, so that I can change data or whatever when needed. They aren't used for malicious reasons, and don't allow access to everything on the system. Although, hypothetically, if I were ever fired, it might be possible for me to get into those admin tools and issue the Apocolyse command that I've also written into the site, requiring the company to hired me back as a consultant at $125 an hour to fix things. Hypothetically. Not like that's going to happen. Really...
The program came in handy a few times. I finally deleted it about six months later.
Sticking feathers up your butt does not make you a chicken - Tyler Durden
A microserf friend once told me MS had no policy either way on easter eggs. They were there if programmers took the time to put them there. If an easter egg can get through development, peer-review, testing, packaging, why couldn't a backdoor?
I used to work for a company that wrote billing software. Our billing app that we wrote had an easy way to get root on the billing box, via a simple exploit. Although it wasn't planned as that, it was discovered shortly after we shipped it, and was unpatched for years.
As a result of this, anyone who knew our software could get in as root to any of the servers running our billing software. I haven't worked for that company in 4 years, and I don't even think they are still around, but anyone running that billing software can be compromised (hopefully no one is still using it, but you never know).
We did have a standard user that we setup in the database as well, so we could perform maintenance, but we told them about it, and coordinated maintenance with them. That could be contrived as a back door as well though, as it did allow remote access by our company.
Which is to say, most people who went into contracting did so just because of stories like you told. They got tired of being jerked around and decided a little uncertainty and paperwork was worth getting little freedom from the corporate brain washing about team and loyalty.
Granted, many went into it because of money during the dot-com boom. They are no longer contractors now ;-)
I'm loyal--to getting the job done, according to contract, as long as I'm getting paid. I produce results, give advice, and let the customer go his own way--even if they insist on taking themselves to hell in a handbasket.
It beats getting all worked up over stupid stuff at work.
I always loved the "We're a family" line I got when people tried to get me on as a FT employee. I don't know about you, but it is usually true--and they have all the problems that families have too. They can keep them ;-)
"Doubt your doubts and believe your beliefs." -- Switchfoot, Ode to Chin
Just this morning, I write a backdoor into a web project. Very often the testing users give me really strange errors that I just can't verify at all. It's useful to have a "master password" that I'll disable later (probably.) Backdoors are most often used for debugging purposes. Fortunately for the users, I'll be the sysadmin when the system goes live, so there isn't much of a risk (yet.)
I was attached to a software package that didn't have a backdoor per se, so much as an undocumented account with a password of "a" that you could not take out of the database without doing major surgery. The software also (used to, anyways) put the undocumented account BACK into the users table and and restore the specific records to their "default state".
Savvier customers changed the username and password (the rule required the user_id entry to stay in the db. But you could change the username/pass to keep undesirables out of the system. Yet many of the customers didn't ever even officially "discover" it... Before I left I never heard of any malicious things being done with this account, but as I told my boss the day I found out about it, "Its only a matter of time."
I left when everybody around me started getting ".com" fever. Like, wacky. People who made $50k annually were leveraging a fortune in paper stock options to buy brand new Mercedes Benzes and hot tubs...
Who did what now?
the facts are:
As to giving the customers the source code, I remember doing that in the mid-'90s before it was fashionable, and having their "computer people" continuously removing it from corporate systems because it wasn't on the "list of approved programs".
and the one to which "kt" refers is described here. Truly ingenious. Even looking at every part of the source yourself can't protect you in a case like that.
- Codes we share amongst ourselves (dev team and with the testers) get out. This is obvious.
- Some codes that I do not share amongst the team got out. And these are complicated as HELL. Like doing a series of things in the game and it having another effect minutes later that you must also know to activate. I show just a few people that these things are even possible, without sharing how. (Yes it's fun to feel cool.)
- Some codes that others added that I do not know about show up in public.
Conclusions: No backdoor is safe. Once released (or delivered), the code is in the hands of the enemy. Period.Corollary: People have way more time on their hands than they should.
If a back door is exploited and the company looses lots of money, etc... They simply go through their source integrity system and hold the coder partially responsible... Even if the programmer no longer works for the company...
It's responsibility...
But then again this country's general populace doesn't accept the concept of responsibility... Just look at the number of stupid lawsuits that are out there and the number of criminals that have gotten off with a good lawyer...
my 2 cents
Writing backdoors is an extremely stupid and dangerous thing to do.
Obviously you take the risk to get fired on the spot. But of far greater risk is your liability. What if your backdoor causes (unexpected) damage to the company? Do you have the pockets to make up for that? Because you can rest assured they will be knocking at your (front) door.
If you have to write a backdoor for 'good' reasons, make sure the company is aware of it, and there should be no problems at all.
The 'putting in a backdoor to make sure a customer pays in time' is stupid as well. If someone writes software for me and comes back with an 'update' after we pay that removes a backdoor, that was exactly the last time that person would work for me.
In fact, that programmer would have signed a contract that specifically states that we do not allow backdoors, but I guess not all companies think of that... Regardless, that programmer has wasted the company's time, with all sorts of (legal) consequences.
I have been pondering the implications of this type of thing for a while now. I am currently building in a backdoor AND telling the user about it. You can read more of the discussion here:
Remote Code Hosting
www.jmagar.com
-
I was working for a company that went belly-up around 9/11. I was one of the few people kept on as a contractor, but I wasn't guaranteed that I would be paid. The product was going to be installed at a single (large) client. I put in a simple check to exit with an error message if the app was run past a date a few months in the future. I planned on eventually taking it out if all went well.
I was at least a little clever in how I put it in; didn't make the check too obvious. But our source control system gave me away (I knew it would, but I wasn't trying to be that secure). I've wondered though, how difficult it would be to plant the change in a much older version of the source file. We were using SourceSafe, but I've never looked into the format of the actual files or how that would be done.
I have never been asked to write a back door, but I have been asked write code to covertly send us reports about how the customer is using our software. The motovation behind this is that our software (Health Care) is price based on the 'number of lives' covered by our system.
My response was that I could do this, but that I thought that doing this without notfying the customer that it was being done was wrong. Stangely enough they did not argue with me, and as far as I know it has not been done.
Real legal actions
5
http://www.harktheherald.com/article.php?sid=7394
In my opinion, this is a terrible way to build the system, and "debugging" methods like this are the reason so much code sucks.
You're not developing the project for a "master user," you're developing it for normal users. Debugging code while in the "master" mode will do nothing more than give you a false sense of whether your code is buggy or not.
It's like installing an app, and then testing it as root. It doesn't tell you anything, and it makes user's lives miserable when they can't get something to work.
Here is something to think about, at my last job, we ran a very popular piece of software that is used to track and manage student loans for our financial aid department. One of our financial aid reps was having problems with her peecee and I sent out one of our IT support guys who was a full time student and part-time employee. He ended up calling the support desk for the piece of software in question and the rep gave him a backdoor account that was full access right over the phone. Luckily for me, the kid was honest and came back and told us about it. We all had a laugh that such an important piece of software could be compromised so easily and the support rep didn't even think twice about giving info on the backdoor to a student. Just to be safe, I periodically checked that student's account and made sure no phony changes were made. The more I thought about it, the more paranoid I became. I talked to my boss (the IT Director at the college) and found out that we really didn't have any choice in the matter, that piece of software was the only one like it. Thankfully, I don't work there any more, and I don't think the problem was ever exploited, but I wouldn't be surprised if I ever read about it. I won't divulge the information about the backdoor, but I will say that the software in question is called WhizKid and there may be college employees here on Slashdot that are familiar with it.
I was subcontracted to build a Delphi app for some other folks. The payment was going to be some pocket change up front and then a percent of sales.
Paranoid that the minute I gave them the program it was going to turn into "Mooman? We never heard of no Mooman" and screw me from the sales, I made a backdoor/easter egg: While the splash screen was showing, if you type m-o-o, the splash would change to information about my little company.
Since the people I was providing the code to weren't Delphi folks, I figured it was a safe CYA to make sure that I got credit where it was due...
I also wrote a perl-based self-registration CGI for them too, and in it I set up a backdoor just so I could get the count of the number of registrations.. Again, just to keep 'em honest.
Not malicious by any stretch but I feel completely justified in what I did...
In the Portland, Ore area and like card games? Check out: http://groups.yahoo.com/group/portlandgames/
Sure it tells you something. When it works for root and not a regular user, you know you buggered the permissions.
I carry around all sorts of hacks and backdoors for the software project I work on. Toggle the hidden switch and bingo! You get access to parts of the code that require a different license. Helps me to debug things (extra state analysis routines) and I don't have to get the customer a temporary license.
Don't know if it's a back door, but....
I mostly have worked on embedded systems. One of them typically didn't have a keyboard installed, but did have a standard PS/2 keyboard port hidden on the back. Pressing any KEY on that would get you past any password prompt on the front interface (LCD/Membrane 10-key pad.)
On another system, a POS system, there's a master password that changes daily, based on a date calculation.
Both of these are used for support purposes, though. On the POS system, there are some changes that can only be made using the "master" login and password. On the controller, field techs use it to reset a password after the owner of the system has lost or forgotten it.
Interesting question... I have never and would never build a back-door into anything I code, unless it was approved and well-documented (some clients might want it).
But, on the other hand, I have build Easter Eggs into systems I've written. One system that is used by thousands of users at my current employer has a silly little Snake game in DHTML if you find it, another high-volume system has Blackjack built-in.
Neither has ever been found, but I have told a number of people, including the managers that sponsored the projects, about them (after the systems were deployed). They didn't seem to mind too much (looked at me kind of funny, but didn't really bitch).
The question... is THAT ok? I know probably most big-time sofware has eggs, but as a matter of course, should it be acceptable, or is it generally unacceptable like back doors seem to be, judging by the general tone of the responses here?
Many of the same arguments apply, such as extra code that could break and put blame on you... They might even be exploitable as security risks if really pooly written... And of course, it most probably was NOT in the client's requirements, so should you do it, even if your intention is not nefarious (mine certainly weren't).
I don't know, I'm kind of torn now that I think about it. I've done it before and didn't think twice, not so sure I would in the future though. Thoughts?
If a pion (n-) collides with a proton in the woods & noone is there to hear it, does lamdba decay into the source pa
A good example is a web-based app I developed for our intranet. We were having trouble debugging customers problems with it (normally PEBCAK - and we'd get useless reports back from customers, and be unable to replicate the problem since we couldn't log in as them).
So, we created a method whereby an administrator could log in as any non-administrator user by supplying the non-admin username and using their own admin password (on a page separate to the normal page). No global master password. One-way encryption of passwords.
When someone logged in this way it was logged to the database that the customer was being spoofed (and by whom) - audit trail. Once the login check had passed the admin could act as if they were that user.
This was considerably more secure for the customer than asking for the customer's password and telling them to "change it later" (which was what we'd had to do previously).
This became an official feature of the app - albeit a not-very-well-known one. To use it you had to have superuser access - which usually meant that you had direct access to the back-end database anyway.
This reduced the time required to deal with problems by a massive amount.
The NSA, interestingly enough, measures their computing power in acres. Yes, not Megahertz or flops or anything lik e that. They measure their computing power in terms of how many acres of computers they have. I think the current value is several thousand acres of computing power.
Some years ago I worked for a really big telecom firm shipping POTS telco equipment to almost all countries in the world. We had this really smart guy who was working with a command module (a module that receives the CLI command the operator would type to configure something). During system test, one of our testers saw something in the code which looked really weired. It was not really weired but the code in this particular piece of code did something which wasn't obvious. The tester tried to enter a commad which excercised that particular piece of statements. Nothing happend. After a while he discovered this code was dependant on a specific type of hardware configuration. After this was set up he tried again to fire off the command. The reply was:
Perre was here!
(his name was Per, Perre is slang). He was sent to the boss and he had to take it away. But I actually think this code got out to customers. If Per reads this you might wanna fill in what really happend at Ludde's office.
My house has a back door. Even if I'm the only one to use it, it's still a back door. It's still there and it can be used by others if they find it unlocked.
RMN
~~~
I saw first hand how back dooring software could provide job security for one developer.
I worked at a company that produced some very complex financial and utilities management software. They needed a way to have these two applications talk to each other and their solution was a daemon to act as a conduit between the two. Since it had to assume user privs the daemon was set to run root suid.
The code had been in production for quite some time when it was assigned to developer to maintain. The code was a mess (it had been written originally by people unfamiliar with programming in the Un*x environment). The developer was tasked with cleaning up the code if he could. Since they were very busy there was little or no supervision over him. As long as the daemon worked everyone was happy.
Eventually the development department decided to restructure and investigated letting this guy go. He had a reputation of being a bit of a hacker so they came to me (I being the Un*x/Network admin at the time) to see how we might protect ourselves from reprisals should he be let go.
I was fairly confident that my systems were tight. The biggest weakness as I saw it was this daemon. So I checked out the source code and started going through it. As I did, I discovered that this simple daemon had developed some new and interesting features. Along with it's normal duties, it also doubled as a telnet daemon (you could telnet to the listening port and login just as in telnet - except this one would ignore /etc/securetty thus allowing remote root logins over an
unencrypted protocol). Another feature was it's ability to tunnel other
ports through it's own listening port.
The code was too convoluted for me to get a complete grasp on it in the time alloted. I went back to the VP of development, pointed out what I had found, and suggested that he would need to have every piece of code this guy had worked on audited to make sure it was clear of back doors. He visibly paled. The developer in question had been there for over 5 years by this point and had touched nearly everything at one time or another.
In the end they simply moved him to another department (he is still there as far as I know). They felt it was too cost prohibitive (and dangerous) to let him go.
They never did tell their customers about these gaping security holes either.
Lessons learned:
What my friend meant was that it usually takes a good amount of bitterness to make anyone consider contracting. It's a scary step and most are intimidated by their manager's comments about "contractors have no benefits".
That's all it meant--i.e. "perhaps you are now bitter enough to take a risk". But what you said still applies. I had many friends who were bitter enough to give it a try. Only one friend of mine did and is still doing it.
My comment was the same thing: if you feel that betrayed, realize there are other options.
As for me, I did it because I was fed up with flushing my life down the drain in salaried positions. When I get paid by the hour, I find I get more equitable treatment. Employers *think* about what they want me to work on. And usually they are more serious--since my time equals money in a very easy to use formula. And I don't feel like I am being cheated when I get a heavy work load--more hours equals more compensation.
Less pay or not, salary is for suckers. Even if contracting is making half the money, right now pay is down across the board and salaried employees are being asked to work twice the hours. So do the math.
(OTOH, during the dotcom days, I made some serious money. Ah, things will never be like that again *g*)
For me, contracting is about quality of life--as in, I have one now.
"Doubt your doubts and believe your beliefs." -- Switchfoot, Ode to Chin
--main reason I am against computerised voting. I'd bet a LOT there's all sortsa nifty "features" in the code that got "reviewed" and judged "ok".
But, it's just so gosh darn conveninet to have the computer vote for me! And it's so modern and techy trendy! And "they" wouldn't do anything to affect a vote would they? I mean it's not like anything as important as control of a state or the world's most powerful country is actually enough motive and incentive for the nice people at E-VOTIN'TECH.con Inc. to do naughty things, is it?
Nawww-never happen in a billyun years, people are all honest, rilly and trully!
In my opinion, this is a terrible way to build the system, and "debugging" methods like this are the reason so much code sucks
I have had situations where code runs fine on the development box but when it is copied to the (supposedly) identically configured production box things fall over. I don't have login access to the production box so I have a backdoor that allows me terminal access for debuggin purposes. Without this bugs that currently take me 5 minutes to find would take someone else sever hours of sorthing through httpd.config files and trying to see what diference is breaking my code.
someone just has to look through the executable for strings.
Oh my... looks like you never wrote something remotely similar to a backdoor. And I mean no just-for-debugging-and-then-i-remove-it backdoor, I mean a real one. A backdoor that's meant to be misused. E.g. you are writing some software for your bank ;) That kind of backdoor. The first rule is that the backdoor should be as invisible as it is possible. And some strange password-like string is the simplest way of shouting "hey, I'm here!"
A real backdoor should look e.g. like the one legendary that once was in a C compiler... for more info see the jargon file entry on backdoor.
Why? Because I REALLY do not like to lose. If I ever got screwed by a client, they would stand to lose more than me.
Have I ever USED one of my backdoors? Only once in over 24 years of working with computers. I wrote a program for a college professor who then turned around, changed the opening banner/credits/logon page to HIS name, and then sold it to the college as his own work. I went in, changed the page back, blew away the user and password file, and disabled the logon sequence. Everyone on the college's staff who had a computer got to see what I said about him the next morning when they tried to log on.
A few weeks later, after all the shit was through flying, I gave the college the program for free. Along with the source code (Open source circa 1979!).
The number 1 problem of working in a cubicle - 23 power cords, 1 outlet...
For this reason, if I write in backdoors like that in a PHP script (a single admin password, for example), I don't actually store the password in a database or the script plaintext, I use an MD5 hash. Even if someone somehow manages to see it (they shouldn't anyway), it's still hard for them to guess what the password is (which they need to POST to the script).
After I quit, I was reading my former boss's email for the next year (having set up the email and web for the company), though he tried to get an injunction to "prevent me from hacking their system" (to muddy the case I had to claim unpaid wages), he never bothered to change his password, till they changed ISPs. At least, there were no surprises in the case...
The less sophisticated customers would never authorize anything like that until and unless they were in a panic, so we learned to pre-install it. In general, saving the customer from himself is necessary to maintain good customer relations, and is probably the origin of the term "Customer Engineering".
I18N == Intergalacticization
This is something I wish more systems had, but not in a hardcoded (backdoor) way. Cyrus SASL/IMAPd does it correctly: they've completely separated the concepts of authentication and authorization. So you can say "until further notice, user slamb can log in and do anything user bob can do." It serves the same purpose, but in a maintainable, open way. You can have a group of administrators that can log in as anyone, but without needing to know either everyone's password or some master password that's difficult to change if someone leaves...they can just use their own, and it can be disabled/enabled on a per-person easily.
But I don't remember the guy's name. Or maybe it was a chick. But whoever it was, they were definitely a staid and steadfast compiler writer.
A customer I was working for deployed our web application on a system we (the developers) had no access to at all. The application did not go through any form of testing before being deployed, so as expected, bugs were numerous.
We asked for access to the machine so we could run simple SQL queries just to figure out WHY things were bugging when the reports started dropping in - but it was against the customer's security policies to allow us on the machine.
It all ended up with us coding a web page which simply was one form and a submit button. Into the form you could type any SQL query and it would be executed directly on the tables, and the result would be displayed in an HTML table on the page. The page was of course password protected and not linked to anywhere, but still...
This is possibly the worst case of security policies gone wrong I've ever seen, and fortunately also the only 'backdoor' I've ever coded.
Posting anonymously to protect the customer.