Careful there, gluadys and crawfish, mark might lose all respect for you any time now!
Here's an analogy for the processes in the Central Dogma.
=========
Imagine that our world is populated by nothing but architects and builders. Every architect has, in his house, a big Mac which stores all his architectural designs. Now the big Mac burger is pretty portable (and the architects carry them around everywhere they go) but the big Mac computer isn't. So every day, before going to work, the architect prints out the blueprints of whatever building he's working on, and brings that to the construction site where the builders are working to construct those buildings.
Here's another twist: the architects have no idea how to change their designs manually. They're all PDF files, but for some reason the code for Adobe Acrobat has been lost, and all they have is Acrobat Reader. (All the computer programmers are down in their basements playing World of Warcraft; nobody has seen them for close to ten years now.) So at the end of the day, no matter how many annotations and markings they make on their printouts in the field, none of that information manages to go back into the hard disk.
What happens when a new architect comes along? He doesn't get to start from scratch; instead, he brings his hard disk over to whichever architect is mentoring him, and simply copies everything over. (Remember, the computer programmers are downstairs playing WoW. How are the architects to know what they need to copy and what they don't need? Best to get everything.) That's what the architects do. What they
don't do is grab a pile of all the printouts and try to scan them back into the computer. Hoo boy, scanners are rare and expensive pieces of junk that don't work half the time. Much better to go back to the source, unless for some perverse reason the architect involved actually wants to
introduce errors into the designs he's inheriting.
=========
So that's how the Central Dogma works. (It's called a dogma only because Watson and Crick weren't creative enough to call it anything else.) The nucleus of a cell is somewhat like a big, clunky computer (clunky by the standards of the cell) with the DNA being its "hard disk". Ribosomes are the places where proteins are made (much like the builders at their building sites); information gets from the DNA to the ribosomes by means of RNA.
Why is it important for the cell to have this kind of a dual system? It's done because storage and transmission are two fundamentally different (and often incompatible) goals for an information system. When I want to store data, I want it to be untouchable in some sense. I don't want it to be too easily altered or damaged. When I want to transmit data, on the other hand, I want it to be easily seen and changed and commented on. So think of the difference between a paper printout and a hard drive. A paper printout is light and easy to carry around, and can be read at first glance; it can be written on, colored on, even folded into interesting shapes. But paper is notoriously easy to lose
precisely because it's so portable; and it's easy to mess up
precisely because it's so easy to alter. A hard disk, on the other hand, can only be accessed through a computer, and only "at a distance" at that. Nobody can look directly at the spinning platters of a hard disk and deduce what information is on it. But that helps keep the hard disk's information secure and controllable: an encrypted hard disk is much harder to read than any kind of encryption you could apply to a piece of paper.
So a cell needs both a "long term memory" and a "short term memory" to get protein synthesis done. DNA is a long term memory: it's massive, bulky, double-stranded (so in-built error checking). The double-stranded-ness in particular makes it hard to modify, but also hard to access: the molecule needs to be
unwound physically for the information to be copied. RNA, on the other hand, is a short term memory: it's light and easily read. It can move around the cell fairly quickly (because of its size); interestingly, the unmatched bases on the single strand of RNA can pair with each other, so that RNA can already fold into base-specified 3D structures - much like how a piece of paper can be scribbled and annotated on to increase its information content.
And that's why mark's error is so important. Not just because I am a pedantic evolutionist but because you get in serious, serious trouble if you confuse DNA and RNA. They do such different things! And if the processes of replication and transcription seem similar, well, that's only because in both cases the information in DNA needs to be retrieved. It's similar to how, with computers, whether you are printing out information or copying it to another hard drive, the first step is reading it off the initial hard drive. So also the DNA molecule needs to be unwound at the location where it is being read.
But that's where the similarities end. The paper printout and the copied hard disks are headed to very different locations, for very different purposes, with very different properties. Similarly, DNA and RNA (despite their apparent similarity, both being nucleic acids) perform radically different purposes in the cell.
What's a transcript error? It's when the printer stuffs up and prints something differently from what was stored on the hard drive. What's a mutation? It's when the hard drive itself stuffs up and stores something differently from the original. Can a transcript error change the DNA? It's as simple as asking whether a printer mistake can change the data on your hard drive: the answer is obviously no. Well, there
are a few indirect channels. As rcorlew points out, the proteins used to replicate DNA are themselves produced through the normal cellular process, so conceivably an error in transcribing those proteins would lead to a mutation in the DNA. Even then though, the link is tenuous and unlikely.
The more important way that DNA can be affected through transcript errors is during the process of
reverse transcription. Reverse transcription is akin to the scanners introduced in the last part of the story: they have the ability to copy RNA information back into DNA format. RNA is far more prone to modification than DNA though, so reverse transcription is only applicable either when quantity is more important than quality, or when the goal is precisely to introduce some kind of variability into the genome. The former applies in the case of
retrotransposons. Those are fascinating little snippets of RNA who hang around DNA for the express purpose of copying themselves into DNA: the retrotransposons that copy themselves well get to introduce more copies of themselves into the DNA, which then get expressed, leading to more of them, which leads to ... an interesting, intra-organismic evolution arms race!
The latter case (deliberately introducing variability) applies to retroviruses and immune systems. Interestingly both are two sides of the same coin: retroviruses need variability to get past the immune system, and the immune system needs variability to check the viruses. In both cases reverse transcription of error-prone transcripts does play a role in variability. This is particularly true into the production of antibodies. Antibodies are essentially proteins which contain receptor components that can "latch on" to specific proteins on the outer coats of invaders. And for every antibody to respond to a different invader, its receptor must actually be different. Reverse transcription is actually used to induce "hypermutation" in the DNA that codes for these antibodies, so that over time each B lymphocyte in the body is capable of producing wildly different antibodies, even though they started from one single copy of ancestral DNA.
But these are exceptions that prove the rule. Reverse transcription can be such a madcap process that, as far as I'm aware, it doesn't really enter into the normal replication of DNA. Normally RNA intermediates don't directly enter into the replication of DNA (imagine copying a hard drive by printing out all its bits and bytes and then scanning them into another hard drive with a scanner!); instead, each strand of DNA unwinds and becomes the template for direct production of another strand of DNA. (This has been proven in studies using radio-tagged DNA strands; when a radioactive double-stranded DNA molecule is replicated, both resulting molecules are radioactive, showing that there is some mixing; but when both molecules are replicated again to give four DNA molecules, only two molecules are still radioactive and the other two are not, showing that the individual strands still retain their identity instead of being mixed up.)
And so that's why it's critical to not confuse the two processes of transcription and replication. A lot happens to RNA (hence its usefulness as a messenger molecule), but not much happens to DNA (hence its usefulness as a storage molecule). Confusing the two is a recipe for sure disaster, and mark's amusing rants on this thread are more than ample evidence of that.