• 2 Posts
  • 33 Comments
Joined 3 months ago
cake
Cake day: July 25th, 2024

help-circle

  • The approach once worked, but that was back before browsers began including the likes of things like advertiser IDs and other extremely high entropy attributes that no average person would ever think to disable. Contemporary hide-in-the-crowd strategies are mostly curated within efforts like Tor browser where everyone is encouraged to use the exact same configuration. But then it’s still a numbers problem. If only two attendees decide to hide their faces with party masks to a soiree of 100 people, one (large scale observer) only need check the guest list and use process of elimination to determine the identities of the 2% “hidden” attendees.

    Somebody can, and probably will, come along and refute this assessment. I am not entirely convinced myself that it is a losing strategy yet. I’m open to hear opposing takes.

    Privacy Badger: IIRC Privacy Badger operates by logging third party domains connections on a per-site bases, and only begins to actively block connections once a domain seen across multiple visits fits the profile of a likely tracker.

    Nvrmnd, they’ve changed how PB works and it is now closer to a list-based tracker blocker (enumerate badness):

    Privacy Badger no longer learns from your browsing by default, as “local learning” may make you more identifiable to websites.

    So they’ve since corrected one of the core issues with PB. Still it is weak. To see why, please glance through The Six Dumbest Ideas in Computer Security.

    uBlock Origin in advanced mode, with default-deny rules (only allow assets by exception) is going to be much stronger at blocking crap.

    Personally, I use uMatrix with pretty much all asset classes blocked by default. I never see popups. I never see banners begging “please allow our cookies, pleeeeaaase!”.






  • I usually use

    dd status=progress conv=sync,noerror bs=64k if=/path/to/source/drive of=/path/to/destination/drive
    

    Take a reading of the drive health first with something like smartctl and if it has a lot of failing attributes, consider lowering the block size to something rather smaller like bs=512.

    Edit: To elaborate, using the sync and noerror arguments will instruct dd to ignore errors and continue in the event of read failures, while allocating zeroes to the remainder of the block space. Which is also why smaller block sizes are better for potentially failing drives. The copy will take longer, but a single error won’t take large swaths of data with it.









  • I do use ClamAV. Most users just run some sort of daily scan, but this is remedial and not preventative.

    In order to truly harness clamav’s potential, you need to configure clamonacc on-access scanning. It passes items off to clamd with lowered privileges and prevents file access through inotify until its realtime scan has cleared.