There's a lot of discourse on here (and on LI) about AI companies "stealing" copyrighted materials for training.

There are - fundamentally - two different issues at play. The fact that most conversations conflate them creates more problems. The first issue is the use of publicly- available data to train models.

The second (and more serious) is companies using LLMs to unbundle content from creators (people, businesses, organizations, whatever) in a way that could cause real harm.

There's a lot of commotion, confusion and fear about Google's removal of 3P cookies + what that means for advertisers.

Much of this is unwarranted and ridiculous: First, there's a misconception that the removal of 3P cookies from chrome somehow impacts Google's data (1P cookies) - nope.

This will impact many third-party services, from smaller (relative to Google/Meta/Apple) ad platforms, to attribution platforms, to certain UI/UX platforms, to other website service providers (basically - any third party that uses a pixel/tracking tag).

Just spent some time playing around with the Google Demand Gen Beta. Initial reactions & takeaways: 1. RIP Discovery - I've long been a big fan of discovery campaigns, though DG appears to be a level-up from existing Discovery for 4 reasons:

(1) Inclusion of YT placements
(2) ML-driven targeting options (similar to Meta)
(3) Ability to create LALs
(4) Standard Bidding Strats

Google Ads is an area where brands invest heavily, all with little-to-no transparency on what that investment is returning. I've done 100+ audits covering hundreds of millions in spend - and here are the 10 things that result in suboptimal outcomes (+ lots of wasted $$$)👇 "Let's break this down into five core buckets -

1. Strategy & Research
2. Account & Campaign Structure
3. Data Flows
4. Creative & Landers
5. Management

High performance in each of these areas is *essential* if your goal is to build + maintain a highly profitable account."

So #MarketingTwitter you've heard about the big scary antitrust case against Google that was brought by a bunch of states you wouldn't expect to be suing big business (TX, KY, AK, ID, IN, MI, MS, ND, SD, UT) + wondering what's it all about, here are some (preliminary) thoughts: (Disclaimer: I have no idea how long this thread is going to go so :shrug: and stop reading whenever you get bored)

2/x

@NeptuneMoon I don't know if I'm an "expert" -- but as someone who worked in finance + now does lots of digital things, this is a really, really dumb take, for (at least) 5-6 reasons:

Thread time, because I want to procrastinate and way too many people don't understand how MBS or ABS work. @NeptuneMoon Reason #1 - MBS issues resulted in a systemic failure due to a heinous combination of securitatization, proliferation of CDOs, deregulation, fraud + general stupidity (simplified).

The *combination* is what allowed the situation to go from an isolated bad to a global really bad.

A4.1: The only "rule" I can think of for this is, "it depends" -- what is your brand? What is your target audience? What social/political issues? It *can* make sense, but it can just as easily alienate you from your target audience if not done correctly.

#SEMRushchat A4.2: Likewise, the last thing many brands want is to be pulled into a national debate or targeted by certain elected officials for expressing their viewpoints -- which is certainly a possible outcome for engaging in overtly political debate. #semrushchat

While I (usually) *loathe* affiliate relationships, I've been thinking a lot about how they can be used for good - and here's what I've decided: there are a number of platforms I use + love that offer them, so I'm signing up + directing 100% of any $ I make in 2020 to help. 1/X I only have 3 right now, but more in process. The rules are simple:

1. I'll only ever work with companies I know + think add value to their customers' lives
2. I'll only refer companies that share my values
3. 100% of 2020 revenue to causes supporting racial + economic justice.

One of the things I keep thinking about is @justinamash's bill to end Qualified Immunity. It's not a perfect solution + I don't think it goes nearly far enough, but it's a concrete policy that has merit + can have a positive impact for thousands who have been denied justice. Obviously, no one *really* likes tort law, but it's an interesting vehicle that may be able to change incentive structures + drive (much needed) reform to policing (especially at the local level).

So...found a fun fact:

To become a licensed cosmetologist in PA, you must have 1,250 hours of classroom instruction, plus 2,000 hours within 8 months as a qualified apprentice, so about 3,250 hours TOTAL.

To become a PA State Trooper, you only need (at most) 2,415 hours. 2,415 represents 28 weeks of academy training, from 0530 hours to 2245 hours, Monday-Friday (per the website). Of course, not all of this is classroom time or instruction - some of it is physical fitness and some of it is study time or unstructured.