Taggart 

@mttaggart@infosec.exchange
6.8K Followers
847 Following
6.6K Posts

Displaced Philly boy. Threat hunter. Streamer. Educator. #infosec, #programming #rust , #python  #haskell , and #javascript . #opensource advocate. Runs @thetaggartinstitute. Made https://wtfbins.wtf. Co-creator of https://github.com/mttaggart/OffensiveNotion. Not your bro. All opinions my own. Dad. #fedi22 #searchable

Pronouns: He/him.

M.Ed | Sec+ | CISSP | eCTHPv2 | eWPT | CRTO

The Taggart Institutehttps://taggartinstitute.org
Bloghttps://twitch.tv/mttaggart
Codeberghttps://codeberg.org/mttaggart
YouTubehttps://youtube.com/taggarttech
GitHubhttps://github.com/mttaggart
Keyoxideaspe:keyoxide.org:G4ADJFWICZZZXGR4STZQVMBJNM

@merospit I feel like this response eats itself. In the first 'graph, the contention is that decreases in sales will lead to reverted changes. In the second, the contention is that anyone around would be unable to attribute the losses to models in use. And then finally, the current phase of enshittification is used to demonstrate that companies can decrease product quality with impunity.

So, kinda seems like these companies would be attracted to a "good enough" crap generator if it costs way less than people without losing business.

I'd also add that you don't have to be an idiot to be unable to measure the value of a product that actually sucks. In fact, most of the big cyber vendors have spent fortunes creating "metrics" that mean nothing for true efficacy. One can be forgiven for losing the plot amidst the forest of Gartner quadrants.

@ThreeSigma

  • Not all standards are created equal.
  • Why couldn't they be hoodwinked as well? I think it will be a while before actuarial data could inform decisions.
  • @cR0w @wrw Maybe "they" can't, but I'm pretty worried about all the hospitals who will decide to do patch management by bot.
    @cR0w @mttaggart Luckily we have absolutely no historical precedent for what happens when the rulers of a decadent and decaying empire decide to stop paying the mercenaries that guard them.
    @Muddobbers When I realized that basically everyone was selling one form of lie or another, I became a changed man.

    More space junk and future ionosphere-zapping metal bits launching into the atmosphere tonight, expected at 8:36pm PT Mon June 16th, Vandenberg SFB.

    #vandenberg #launch #spacex #space #sonicboom

    @jimi I'll also add that since this was focused on security, we're already talking about a cost center. The idea that you can reduce its cost without sacrificing quality is a lie every CFO wants to believe.
    @jimi I have yet to see a business really make anything but short term decisions, so if it's a limit, it's a pre-established one.
    @cR0w But that's the same thing in almost every case, because appearance is 90%. And for the other 10, the folks selling these tools can rainmake you for ages before someone catches on, and by then it'll be too late.

    For the kazillionth time, generative models don't have to be good to threaten your job; they just have to be good enough.

    And what is "good enough?" It's good enough to pass muster, to answer leadership's needs for a Thing To Present, to hang in front of the Mission Accomplished banner.

    And here's the kicker for cybersecurity. The answer almost never has to work; it just has to look like it does. The tests are too few and far between for every wrong answer to be discovered—especially if those who would discover them have been given the boot.