## hpr3362 :: Spam Bot Honey Pot: Eating your own dog food

 
In this episode, I revisit my spam bot honey pot method of spam detection for 
web forms. The first part of the episode is a response to rtsn comment asking 
for a follow up on how the method worked out. In the second part of the podcast 
I use the built-in screen reader of my Pop_OS system to review my test form and 
my work form for audible accessibility and to check that the spam catching URL 
field is not reveled by a screen reader.  


For my work site, I pushed the spam catcher to production on 16 February 2021. 
The total submitted form messages from that day until today, 29 May 2021, is 661.
Of the total submitted, the method identified 527 spam messages, and passed 
along 134 messages. Of those 134, 38 messages were uncaught spam--while the 
remaining 96 messages were valid inquiries. Overall, it seems to be doing a 
good job. At one point, I thought about closing down the form, but the data 
does show we are receiving enough valid messages to make it worth keeping around 
and this method makes finding the good messages much easier.


During the second part of the podcast, I find using a screen reader, like any 
tool, takes some practice. It also reveals that making a form audibly accessible 
does take some care, and should be reviewed with a screen reader. My method 
for hiding the honey pot URL field from the screen reader was effective, but my 
overall implementation to make it accessible needs more work.
