#114 Tech Time by Tim is moving

+ Davos 2025, Random Science, Samsung’s S25 Edging, The DeepSeek Depth Charge, But What Are Agents For?

Dear readers, Tech Time by Tim is moving. It'll just be a small shift in emphasis, with my main publication platform becoming Substack instead of LinkedIn. Newsletters will no longer be shared on this account, so be sure to shift with me to Substack and stay connected through my personal LinkedIn account.

The Week That Was

A look back at the tech world of the past week.

Davos 2025

With all the uh…events happening this past week, you’d be quite easily forgiven for forgetting about the World Economic Forum’s annual billionaire ski trip this year. But fear not, for I remembered! and am here to talk you through the technology-related key points of Davos 2025 (as the event is often called due to the location being in Davos, Switzerland).?

Reuters had comprehensive rolling coverage of key events, so rather than a bunch of disparate links, you can just find that as the main link at the end of this item if you want to know more.?

As for what I’d like to focus on, there was actually some positive stuff! A new report coincided with Davos which marked 2024 as the first year in which Europe used more solar power than coal power. This trend looks set to continue amidst the wider ongoing climate tech push that has kept getting stronger over the course of 2024, rolling into 2025.??

Obviously, there was plenty of jostling for AI funding and contracts. More specifically, who will get to implement their agentic infrastructure across enterprises and governments? This was really funny to me personally with the hindsight of the DeepSeek Depth Charge written about farther down in this newsletter. Said depth charge also significantly impacted the next and final topic I’ll touch upon, namely, calls for deregulation and more money.??

There were plenty of calls from European business leaders for the EU to speed up tech deregulation and increase investment efforts. In this context though, ‘leader’ is a bit of a misnomer, seeing as all they seemed to be doing was following the Fear of Missing Out (FOMO) of their American peers. But I digress.

To be clear, an excess of red tape and an upswell of nationalist narcissism are significant impediments to European tech flourishing. These can be overcome though, with discipline and cooperation. LINK??

Random Science Go

Rules of Engagement are usually about ethics and legal matters. But sometimes I just want to share some weird and entertaining stuff with you without necessarily having a deeper point to make about any of it. This is one such time, where I talk about gills and ears, spidermen (not a typo), fluid feline memes according to science, and ok…look there’s actually some promising biomedicine stuff at the end too.??

Now, I’m not much of a comic book fan, truth be told. But I did always think it funny how a lot of the being time, aquatic people like mermaids were depicted as having these sort of fish fin-like ears, or gills on their necks, etc. in comics.

As it turns out, our ears have a lot more in common with fish gills than I thought. I’m not really surprised though, since the first creature to crawl onto land was an amphibian and all. Wait, is this why our ears can go all funny underwater? Is that awful popping sensation the call of the eldritch call of the sea telling us to…nah lol. Calm down, Mr. Lovecraft.?

What is with the liquid-related science news this week though? What if cats are liquid? Have you ever wondered that? Me neither but here’s a physicist weighing in on the ‘fluid feline’ meme for The Scientific American. It’s a pretty interesting read. Sometimes science and technology can just be funny and weird. Sometimes some scientists will just invent spider man’s web-slinging system in a lab just because they can.?

Other times, there will be new insights into how cancer cells hijack and corrupt immune cells, with implications for cancer therapy. Namely, if we know how cancer wrecks these cells, we can think of ways to make them stop doing that. LINK??

Samsung’s S25 Edging?

Would you accept a phone without any ports whatsoever if it meant it could be wafer-thin? I would not. But reactions to Samsung’s Galaxy S25 edge, a teaser of a reveal, seem to place me in the minority.

As I watched the Galaxy Unpacked keynote wrap up, it felt to me as if Samsung had very little confidence in their S25 Edge. This is their entry into the nascent ‘wafer thin’ smartphone niche that will be filled up with devices over the course of 2025. It was the only thing to genuinely excite the crowd at Galaxy Unpacked. Why? I think the answer is the pure novelty factor.??

Because the sentiment around the mainline S25’s reveal was conveyed pretty damn sharply by The Verge’s Allison Johnson when she wrote “Samsung needs to give us a reason to care about new phones every year.”

I respectfully disagree though. I’d change that title to “Smartphone manufacturers need to stop pretending like they aren’t as exhausted as we are with these annual, incremental releases.” The problem with annual releases as I see is that they make a lot of sense, both in terms of yearly revenue and in terms of upgrade cycling. Not everyone upgrades at the same time, so someone, somewhere will always actually benefit from any given year’s new toys.??

However, smartphones have been ‘boring’ for years now. They are a mature product that manufacturers have to keep pretending to be new and revolutionary each year. And that in turn has conditioned consumers to expect their minds to be blown by these promised revolutionary leaps each year.??

It’s this vicious cycle of overpromising and underdelivering that’s gotten us to where we are now. Often leading to seemingly arbitrary design decisions to change things just for the sake of changing them. LINK

Algorithmic Gymnastics

Documenting the wild, wonderful, and occasionally dreadful things done with algorithms.??

The DeepSeek Depth Charge

Hey, you guys remember those times I wrote about how the insane levels of investment in the AI sector couldn’t possible be sustained for much longer. Kind of like some kind of bubble that keeps getting inflated until it pops? Many observers and analysts think this was the week in which it did.??

The narrative goes that a small team of Chinese researchers had about $10 million dollars to use on a side project for a venture capital firm. The idea was to show off to potential partners and clients. Well, mission accomplished there lads.??

At the time of writing, details about DeepSeek, including the reproducibility of their proclaimed benchmarks, are yet to be verified. I mention this for the sake of clarity and fairness, but it doesn’t actually matter whether or not DeepSeek itself is telling the truth or lying. There are already allegations though, that the team may not have been entirely forthcoming about its AI training methods. As stated earlier, though even if these developers are ‘frauds’ the underlying truths do not change.??

Those truths being that no one has a moat, and everyone that knows what they are doing can actually do a whole lot with relatively modest means. You do not need infinite resources to ‘win’ at AI. You just need to be smart about working with what you’ve got. I will say though, if it does turn out that a VC firm of all things ended up blowing that jealously guarded open secret to the previously ignorant wider world…. Wow. This story would have more layers (of irony) than an onion in the movie Inception.??

And sure, much like their American equivalents, the DeepSeek app and API have grabby hands. But unlike certain American peers, their local AI model is actually Open. Not that most users will engage with that one of course. This is actually why there was such a big stock wipeout due to DeepSeek.??

The people least knowledgeable about AI are the ones most easily excited by it. Excited enough to keep pouring money in no matter what. So just imagine, on top of all those previous sunk costs, you just committed uh….let’s just pick a random silly amount….$500 billion more dollars. Boy, wouldn’t that be really awkward right now? LINK??

But What Are Agents For??

I’ve been working on a bunch of AI stuff at work, perhaps that’s why I’m in such a contemplative mood about agents right now. Oh yeah, and OpenAI released a new thing called ‘Operator’ as well. Combined, they have me pondering what AI agents are even supposed to be for. Well, I’m going to poach the 80/20 AI split that a colleague explained to me a while ago to explain.

The 80/20 split is also known as the Pareto principle.? This principle holds that 80% of an employee’s time is wasted on menial,? repetitive work, with only 20% left for high-impact, strategically valuable work.

Thus far, much of the undelivered hype of AI has focused on trying to replace the 20% whilst leaving the 80% mostly on the table. They want to replace the artist instead of giving him a more versatile brush or a bigger palette.??

We know now that hallucinations can’t be fixed, and that AI mistakes are not like human error either practically or philosophically. Simply put, AI is and will remain too unpredictable to behave according to human goals.??

Agents then, can be one of two things.??

1. A doubling down on trying to get that 20% of ‘meaningful’ human labor no matter the financial or human costs.?
2. A recognition of the limitations and pitfalls of AI. If recognized and worked with, agents could be a step towards genuine, 80% oriented utility.??

Practically speaking, even if we can’t stop the agents from making stuff up or lying, we can control the contexts in which they operate. Techniques geared towards very controlled and focused usage have shown promising results, and that’s why so much of AI coming to your phone (when implemented properly of course) are thing you already did, but more convenient. In other words, agents as quality of life rather than substitutes for life. LINK??

A Nice Cup of Serendipity

Cool bits and bobs from around the web.?

Deep Cavemen LINK??

Brake Bucks LINK??

Car Hopping LINK??

Congestion Pricing LINK??

Brain Dones LINK??

Electric Tongue LINK??

Robotics Predictions LINK??

Slacking Off LINK??

DM Derps LINK??

TikTok Censorship? LINK??

Atomic Labs LINK?

Attention Payments LINK??

PC Game Metrics LINK??

Game Library LINK?

Jet Raid Doubts LINK??

More? Lynch In Gaming LINK??

Crypto Crazies LINK??

Social Secret Sauce LINK?

Hyperloop History LINK??

Vidya Opera LINK

One More Thing…?

So my mother and I were experimenting this past week with how to make a tasty snack that was cheaper and healthier than the stuff you can buy in the supermarket. Specifically, she used to like the sort of candy bar thing that was basically just dates with other stuff mixed into it. The fact that I did not know the true specifics of this ‘other stuff’ set off the control freak part of my little tech goblin brain.??

So anyway, we made something really great, ate it a bit too fast because it was really good, and then my mother managed to make a second batch that ended up being even better. Part of that was simply because we’d proven the formula and knew which steps of the process did and didn’t work.??

So I took some samples to work to test on some co…I mean, to share with them. And whilst some were apprehensive at first, the reception was universally positive! I talked with our company chef about potentially making some for everyone at our (Alkmaar) office at some point, and have already gotten ideas for variant flavors from some of the colleagues who enjoy the culinary arts as much as I do! Snack Time by Tim incoming!?