smash-the-firewall/slides/metadata.md

<!-- .slide: data-background="img/metadata.jpg" -->


<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<small>
> If you give me ~~six lines written~~ **metadata** by the hand of the most honest of men, 
> I will find something in them which will hang him.

>                            Probably Richelieu

</small>
--

## Metadata

Pieces of information that describe everything BUT the content. Examples:

- phone call: my location, the duration, who I called
- a PDF file: the author, the location, the authoring software
- an encrypted file: date of creation, modification and last access

<br><br>Content is easier to protect, metadata often **leaks**.

--


## I don't need to know the content if

- you called a sex line at 2:24 and you stayed on the phone for 12 minutes
  <!-- .element: class="fragment" -->
- you called a suicide prevention line
  <!-- .element: class="fragment" -->
- You watched a video on anxiety management for 27 minutes at 8:05 AM
  <!-- .element: class="fragment" -->
- Your phone was near the one of famous activists during a protest
  <!-- .element: class="fragment" -->
- You entered a cannabis dispensary and stayed 44 minutes
  <!-- .element: class="fragment" -->

--

## Why are metadata important #1

In many legal systems, the content is usually protected much better
than the metadata. For example, in Italy, phone calls are only
recorded in case of ongoing investigations for certain level of
crimes, while phone operators are legally required to maintain the
metadata of phone calls for 24 months.


--

## Why are metadata important #2
Metadata are easy to mine:

- large scale research: metadata are textual and rigidly structured. There are many algorithms to navigate structured data by keywords, time, descriptions, etc...
  <!-- .element: class="fragment" -->
- indexing: there are methods to cluster metadata so that it's easy and fast to retrieve parts of the information on demand
  <!-- .element: class="fragment" -->
- statistical analysis: extract current and historical trends, unforeseen categories and forecasts from metadata
  <!-- .element: class="fragment" -->


--

## Why statistical models matter for L.E.
Law Enforcement make wide use of statistical models.
- <a> Resource allocation  <!-- .element: class="fragment" --> </a>
- <a>Performance evaluation of different units  <!-- .element: class="fragment" --></a>
- <a>Assessing crime trends  <!-- .element: class="fragment" --></a>
<br>
- <a style="color:red;">Criminal profiling</a>
  <!-- .element: class="fragment" -->
- <a style="color:red;">Build correlations and associations</a>
  <!-- .element: class="fragment" -->
- <a style="color:red;">Predictive policing and crime forecasting</a>
  <!-- .element: class="fragment" -->

--


## How do I protect myself?

The only real protection here is the **mindset**.
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<small>
<br> start with commonsense
<br>
<br> end up with paranoia


</small>