It's interesting how MTurk reached out to script creators and workers for input on the new worker site but as far as anyone has said, MTurk didn't reach out to anyone for input on these changes, which sound like they're potentially very major.
Here's a scenario that illustrates just the type of worker that these new policy changes apply to:
1.
Malicious worker decides they want to easily make $300-$500/day the dishonest way.
2. Worker makes a script to scrape batches of HITs (e.g. Barcodes) shortly after they drop. This is done while logged-out. The script opens each HIT in the batch, extracts the HIT ID and the UPC and saves them, and then skips to the next HIT. This process generates a ton of page calls to Mturk.com (10's-100's per minute) The whole batch can be scraped within minutes of it dropping.
3. All UPCs and corresponding MTurk HIT IDs are stored offline in a custom offline database, and the UPCs are automatically searched via another script against existing online UPC databases. UPCs that result in no products being found are flagged. The UPC search results from the online databases are automatically saved to the workers offline database (this takes <10 minutes)
4. Dishonest worker now logs into MTurk, and runs another script that previews each Barcode HIT in the batch and checks each against their custom database. Once a HITs ID containing a UPC that corresponded to no product is encountered, the script automatically accepts the HIT, checks the box for "No Product Found", and then auto-submits the HIT. Script moves on to find the next HIT, and repeats the process until the batch is exhausted.
Such a malicious worker would violate all three of these 3 key policy points:
1. High frequency page calls
2. Extracting & storing data
3. Substituting their human judgement for letting a script automatically search for and input the information into the HIT.
To reiterate what was recently discussed above, this is VERY different from how the average Turker uses scripts and does not apply to something like HitDB