The rules of a world you can't trust

Starting from a world you can't trust

When you send a message on a network,
it leaves your hand
and passes through many in-between roads.
You can't know if all of them are safe.
Somewhere, someone might quietly look in,
might slip a change into the contents,
or might intercept it on the way.
Just knowing these risks exist is enough.
You don't need to know how they're done.
What matters is the mindset: never blindly trust the road.

Follow the message's road from sender to receiver. Click to recognize what risks can exist along the way.

All three could happen along the road:
secret reading, content tampering, interception.
But look at these risks closely and
each breaks a different thing.
If it's read, the secret leaks;
if it's changed, you can't trust the content;
if it's intercepted, what should arrive never does.
So there's more than one thing to protect.
Security neatly splits this into three goals.

Three to protect: confidentiality, integrity, availability

Security's goals gather into three.
First, confidentiality.
Only the intended person can read it.
If someone reads it secretly, confidentiality breaks.
Second, integrity.
The content stays unchanged on the way.
If even one letter differs, integrity breaks.
Third, availability.
It can be used without obstruction when needed.
If it's blocked from use, availability breaks.
We take the first letters and call it CIA.

Tap the three goals one by one: confidentiality, integrity, availability. See what each protects and what happens if it breaks.

Seeing the three apart makes it clear.
Confidentiality is not-read, integrity is not-changed, availability is always-usable.
Even for the same message, the three differ.
If it's just not read, confidentiality held,
but if the content changed meanwhile, integrity broke.
So the three goals must be tended separately.
Here, though, one thing is easy to confuse.
All three are things we want to protect;
we haven't yet said how to protect them.
To protect a goal, you need a method.

One means: checking who someone is

There are many ways to protect a goal.
The first you meet is authentication.
Authentication checks that the other side really is who they claim.
In a strange world, a name tag isn't enough,
since anyone can say someone else's name.
So we ask them to show something only they know.
A password is the most familiar example.
Show the right one and you're accepted as yourself;
show the wrong one and you aren't let in.
This sorts out who someone is, right at the door.

Pick the secret only you know and pass the door. Right gets you accepted as yourself, wrong keeps you out.

Showing the right secret got you through.
Authentication sorts out who someone is.
But authentication is a means, not a goal.
Authentication itself doesn't keep a secret;
it just filters out the wrong person
and helps protect goals like confidentiality.
There are many means besides authentication:
encryption that scrambles content so others can't read it,
a check that tells whether something changed, and more.
So it would help to set goals and means side by side.

Goals and means are different boxes

Now split the two clearly.
A goal is what you want to protect.
Confidentiality, integrity, availability are the goals.
A means is the method that protects that goal.
Authentication, encryption, integrity checks are means.
There's a reason they're often confused:
the name of a means can resemble a goal.
For example, encryption is a means,
but its goal is to protect confidentiality.
Sorting cards into a goals box and a means box makes it clear at a glance.

Tap each card to sort it into the goals box or the means box. CIA is what to protect, authentication and encryption are how.

With the cards sorted, it's tidy.
The left box gathered the goals to protect,
the right box gathered the means that protect.
Not mixing the two is the eye for security.
Whatever security story you hear,
just ask first:
is this a goal to protect, or a way to protect?
If the goal wobbles, what to protect goes blurry;
mistake a means for a goal and you lose the real goal.
Now let's wrap the whole thing up.

Let's wrap up

Gathered on one line, it is this.
The network world is a road you can't trust.
So security sets up three goals:
confidentiality, integrity, availability. CIA for short.
These are goals to protect, not methods.
The means that protect them come separately.
One of them is authentication, checking who someone is.
Not confusing goals with means is the first step.
Deciding what to protect in an untrusted world, and choosing how to protect it. That is security.

Tap the key points in order to review. (untrusted world -> goals CIA -> means authentication -> goals vs means)

Now you know what security asks.
What to protect, and how to protect it.
These two questions are the skeleton of all security.
The three goals of CIA will keep coming back.
Authentication is just the first means to protect them;
more means, like encryption, are waiting.
Next, how does encryption,
scrambling a message so others can't read it,
protect confidentiality?
Let's keep exploring together.