Martin Karlsch

@eric_seufert All while avoiding billions in profit taxes, repatriating them at conveniently reduced rates, and using that cash for stock buybacks. The hypocrisy is real.

@eric_seufert Calling regulatory fines “de facto tariffs” misinterprets tariffs and mischaracterizes the purpose of these laws. The fines reflect scale and non-compliance, not nationality. American tech isn’t being tariffed; framing it this way is oversimplified victimization 🤷🏻‍♂️.

But American tech is tariffed. DMA fines, GDPR fines are predominantly levied on American firms. These are de facto tariffs.

@evanlapointe Primed to perform and Under the hood are both great

What is the single best book or resource about culture that you’ve come across? I’m wondering if there might not be a truly great resource or if the best once still don’t quite nail clarity about such an abstract topic.

@eric_seufert Thx for the clarification. For anything in the minute range the accuracy is still good enough given what is left (i.e. on iOS in app). Enough to slow down the transition you describe.

@dynamix My point is not that it’s technically impossible to construct any type of fingerprint. My point is that fingerprinting doesn’t provide enough fidelity to be useful for user-level attribution, especially now that *only* the IP is available.

Advertisers accept ad tech vendors’ use of fingerprinting — and the misguided characterization of it as “probabilistic attribution” — to their peril. These methods are not future-proofed, but more than that, they mask the broader trend of user-level attribution ultimately, eventually being impossible. Advertisers should be adapting to an increasingly restrictive privacy environment with system-level changes. Not only does simply swapping deterministic identifiers with fingerprints introduce measurement imprecision / risk, it delays the inevitable, which is a wholesale transition to econometric ad measurement. Even genuine probabilistic methods shouldn’t be seen to facilitate user-level attribution. That’s a dead end. Advertisers should accept that, absent persistent deterministic identifiers, their measurement infrastructure should seek to answer three questions: 1. What’s the optimal level of ad spend? 2. What’s the optimal distribution of ad spend across historical channels? 3. Does my spend on any given channel deliver incrementally additive revenue? This — and the introduction of new channels — requires moderated, systematic experimentation. There is no “bottoms-up” methodology that can provide guidance. Advertisers should seek to achieve platform-level ad spend efficiency and maximize revenue using first-party data through personalization. It’s not just that fingerprinting isn’t reliable — what fingerprinting attempts to deliver, which is user-level attribution, isn’t possible as privacy interventions reduce the availability of deterministic identity.

@eric_seufert ATT kicked in three years ago, Apple isn’t policing fingerprinting or putting the money on the table to funnel everything through private relay = market builds „probabilistic“ solutions instead of moving away from the user level

@eric_seufert Yeah imho that’s unfortunately wishful thinking. As long as fingerprinting still works on a technical level it will be used. It’s not even „advertisers accept“ it. Many understand this well and ask for it ala „hey XYZ offers us bla % more reach and the performance is fine“. 🤷🏻‍♂️

@eric_seufert Is the point not user level attribution is EOL, pivot to econometric methods? Pls correct me if I misunderstood this. If it is - iOS is a good example what happens if you leave the market with stuff like the IP, many take the path of least change even if they agree with ur points

@dynamix You’re missing the point.

North Atlantic sea surface temperatures hit a record daily high on March 6, 2023. 325 consecutive days of record temperatures later, the North Atlantic is still setting daily records. It's f&%king insanity that the collapse of the planet is not the lead story every day.

@antoniogm „here is why somebody else’s misery is not so bad“ or „four reasons why you don’t have to show empathy or compassion for THESE people“

@ral1 Awesome book! In part IV you mention that centralised L&D investments are mostly wasted and mention research that supports this statement. Could you point me to the research? That would be highly appreciated as I couldn't find much myself.

I wrote this book (whentheywinyouwin.com/the-book) because people deserve to be led well, and the overwhelming evidence is that managers are failing, despite the mountains of content aimed at helping them be better. Don't worry, gang, I brought receipts (they're in the book!)

@Plinz Great poll to reason about your followers 😁

When you see that someone has pronouns in their bio:

@sokirill @ejames_c I can definitely recommend it - if only for the first third. It helped us to formulate and execute a comprehensive and better strategy compared to what we had before. Some multi year real life positive impact for a 100+ people company vs “a book review” 😉

@ejames_c I was just thinking of reading it and thought it was actually recommended by you :) "tackle the hardest bit, which is the context-dependent diagnosis" => I think this is the bit that makes most books kind of useless, they are just too abstract to draw practical conclusions.

I am about 90% convinced that Good Strategy Bad Strategy is a bad book. Two questions: 1. What is something actionable you took away from the book? 2. In a year’s time, circle back and ask: what changes did I make as a result of that takeaway?

@ejames_c 1. A shared framework how to formulate and execute a strategy with a clear structured understanding of what strategy can be 2. A lot of positive changes that created better alignment and more focus (200 people company) Out of 20 books on strategy it was the most impactful.

@sokirill sounds like survivorship bias @ PayPal ;-)

If I ever have employees, I want to create a culture where when they go and start hacking on a side project, I can create an environment where they can explore other options and also invest in them. Why do companies don't do that? If only PayPal invested in all their alumni.

@sokirill I booked a hype train ride

When people say I am transitioning from web2 to web3, what does it actually mean? I am learning solidity? I am moving all my $ to $ETH? I am taking a job where I get paid in crypto? I am storing the data on publicly available ledger instead of a privately-held S3 instance?

@NataliePis Used in production 24/365 since eight years :-)

Gophers, have you ever used gob? #golang

@eric_seufert At Apple it’s just @johnwilander 😉

Idle thought: the very small cadre of product and engineering managers working on advertising privacy across Apple and Google -- perhaps totaling fewer than 1000 people with direct impact? -- are defining the future of the internet economy. Almost as a modern-day Bretton Woods.

Care to join a company that puts culture first? We are looking for a VP Engineering: remerge.io/careers/496450…

@sokirill @diannamallen Anything you can do with a normal computer you can pretty much do with a piece of paper ;-)

@diannamallen You can exchange bitcoin for $ :) Jokes aside though, crypto or blockchain is just a matter of storing information on more than 1 place and then retrieving it for whatever purposes. So anything you can do with a normal database, you can pretty much do with blockchain.

is there anything you can do with crypto that is NOT crypto? serious question, i’m just trying to learn

@antoniogm They provide infrastructure+sdk to collect trillions of mostly worthless events and play blackbox attribution arbiter - a centralized inconsistent modifiable ledger - i don’t know, to me this doesn’t have much to to with a blockchain …

As a side note, what do attribution providers like Branch, AppsFlyer, etc. really do? They provide a selectively-readable, publicly writable ledger of past actions, allowing multiple parties to agree on shared state with no central control. Sounds a lot like a blockchain....

Since the intersection of ads and Web3 seems to be rather ... desolate ... at the moment and nobody has written much here (kudos to @aripap for doing so), I'm going to do a point-by-point analysis of this thread, some of which I agree with (and some not).