In recent news, Google has put forth a proposal known as the "Web Environment Integrity Explainer", authored by four of its engineers. On the surface, it
It would stop beneficial bots like the ones I create¹ as a small-time hobbyist because the little guy does not have the resources for this arms race. You may be right when it comes to large-scale scraping ops that are done by a business (e.g. scraping RyanAir or Southwest airlines so an airfare consolidation site can show more fares).
① e.g. I wrote a bot that scraped the real estate market sites, scraped the public transport sites, and found me a house with the shortest public transport commute.
It was coded 8 years ago in Tcl¹ for a one-off project in Belgium. Would you really be interested?
The APIs would have changed dramatically by now & some of the real estate sites no longer exist. Some of the sites brought in CAPTCHAs. It was coded to use Tor & the public transport site has become Tor-hostile and also changed their API. It’s also very user unfriendly… a collection of scripts & variety of hacks because I was my only user.
I didn’t publish the code at the time because I worried that it would trigger the target sites to become bot-hostile.
① Also note that I use #Tcl for personal use but I resist publishing any Tcl code because I would rather not promote the Tcl language. Why? Because the Tcl folks have jailed a large portion of their docs in Cloudflare’s walled garden. I believe programming language docs should be openly public.
I wouldn’t be trying to use or adapt it; it was more of an intellectual curiosity. I haven’t written scrapers or used Tcl, so, yeah, I would be interested to take a peek :-)
I’d like to write something similar, but I’d probably use Lisp or Python.
But the limitation of using #Selenium is a big one. Being forced to work in java, forced to use the resource hog of a modern gui browser, forced to reveal more browserprint info, being browser-dependent, etc. Selenium is my last choice when desperation is sufficiently high.
Am I the only one thinking these trust tokens are not going to prevent bots from scraping websites?
Eventually, somewhere, someone will just develop the infrastructure to work their way around this, right?
It would stop beneficial bots like the ones I create¹ as a small-time hobbyist because the little guy does not have the resources for this arms race. You may be right when it comes to large-scale scraping ops that are done by a business (e.g. scraping RyanAir or Southwest airlines so an airfare consolidation site can show more fares).
① e.g. I wrote a bot that scraped the real estate market sites, scraped the public transport sites, and found me a house with the shortest public transport commute.
Are you willing to share the source for said bot?
It was coded 8 years ago in Tcl¹ for a one-off project in Belgium. Would you really be interested?
The APIs would have changed dramatically by now & some of the real estate sites no longer exist. Some of the sites brought in CAPTCHAs. It was coded to use Tor & the public transport site has become Tor-hostile and also changed their API. It’s also very user unfriendly… a collection of scripts & variety of hacks because I was my only user.
I didn’t publish the code at the time because I worried that it would trigger the target sites to become bot-hostile.
① Also note that I use #Tcl for personal use but I resist publishing any Tcl code because I would rather not promote the Tcl language. Why? Because the Tcl folks have jailed a large portion of their docs in Cloudflare’s walled garden. I believe programming language docs should be openly public.
I wouldn’t be trying to use or adapt it; it was more of an intellectual curiosity. I haven’t written scrapers or used Tcl, so, yeah, I would be interested to take a peek :-)
I’d like to write something similar, but I’d probably use Lisp or Python.
How would this impact selenium based bots?
Perhaps not at all.
But the limitation of using #Selenium is a big one. Being forced to work in java, forced to use the resource hog of a modern gui browser, forced to reveal more browserprint info, being browser-dependent, etc. Selenium is my last choice when desperation is sufficiently high.