The Truth behind Implicit/Explicit form labels

Recently I've been hearing to avoid "implicit" labels and to favor explicit labels for accessibility reason. This was the story I was given, but it felt like there was something missing from the story.

<!-- Implicit -->
<label>
  Text
  <input>
</label>

<!-- Explicit -->
<label for="example">Text</label>
<input id="example">

After doing some research I found, like with all things, there is some more nuance to the story.

1. Both are 100% fine and completely valid according to all language and accessibility specifications.
2. HOWEVER, specs and real world implementations are two different things. Some actual screen readers do in fact have problems reading labels properly.
3. The issue is not related to the code being nested in a label, but instead whether there is an id/for attribute involved.

Sources:

• Screen reader benchmark testing:
  • https://a11ysupport.io/tests/html_label_element_implicit
  • https://a11ysupport.io/tests/html_label_element_explicit

TLDR:

<!--
  Bad: Though all screen readers *should* work with this,
  some just won't play nice with it.
-->
<label>
  Text
  <input>
</label>

<!--
  Works: This has the same support in screen readers as
  the next option, but semantically it's not great, as the
  label header contains non-text content
-->
<label for="example">
  Text
  <input id="example">
</label>

<!--
  Best: This separates the label tag and input so their
  duties are not conflated, it uses the for/id attributes
  that works in all screen readers
-->
<label for="example">Text</label>
<input id="example">

And of course, make sure to use unique ID's.