3 ms·

From article: "In its updated statement, Tumblr said that while every image uploaded to the platform is “scanned against an industry database of child sexual ab

From article: "In its updated statement, Tumblr said that while every image uploaded to the platform is “scanned against an industry database of child sexual abuse material” to filter out explicit images, a “routine audit” discovered content that was absent from the database" Same thing as removing Twitter because a non filtered post showed up in someone's feed.

developer2 8y ago
So basically, the same problem rooted in management at most tech-heavy companies: Tumblr likely focuses every single dollar spent on development towards pumping out useless new features in under half the time those features would require to finish properly, sacrificing quality over quantity – every damn time. Some developer at Tumblr is now grumbling, "I fucking told you we needed to spend more time on the filter system, and that it wouldn't work properly when you forced me to ship it to prod after only 3 weeks instead of the 6 weeks I told you it would take". Someone there knows that the cron/API that updates that database is non-functional or untested, but nobody higher up the chain of command gives a shit. Source: have dealt with this kind of management time and time again. In particular, "Agile" development has destroyed this industry. You tell management the task will take 6 weeks, so they assign 3 weeks to the ticket in Jira, and suddenly that means the feature has been promised as being deliverable in 3 weeks. If not, it's somehow your fault for not meeting deliverables.
- devwastaken 8y ago
  There is no filter system in the world that will catch all cp. It doesn't exist because current systems rely upon known content. It's image recognition, and image recognition can be bypassed. Either by an image not in the database, or by modifying one that is.

3 ms·

From article: "In its updated statement, Tumblr said that while every image uploaded to the platform is “scanned against an industry database of child sexual ab

by devwastaken 8y ago

developer2 8y ago
So basically, the same problem rooted in management at most tech-heavy companies: Tumblr likely focuses every single dollar spent on development towards pumping out useless new features in under half the time those features would require to finish properly, sacrificing quality over quantity – every damn time. Some developer at Tumblr is now grumbling, "I fucking told you we needed to spend more time on the filter system, and that it wouldn't work properly when you forced me to ship it to prod after only 3 weeks instead of the 6 weeks I told you it would take". Someone there knows that the cron/API that updates that database is non-functional or untested, but nobody higher up the chain of command gives a shit. Source: have dealt with this kind of management time and time again. In particular, "Agile" development has destroyed this industry. You tell management the task will take 6 weeks, so they assign 3 weeks to the ticket in Jira, and suddenly that means the feature has been promised as being deliverable in 3 weeks. If not, it's somehow your fault for not meeting deliverables.
- devwastaken 8y ago
  There is no filter system in the world that will catch all cp. It doesn't exist because current systems rely upon known content. It's image recognition, and image recognition can be bypassed. Either by an image not in the database, or by modifying one that is.