Author: jszeder

Everyone is trying to predict the arrival of AGI. From “It will be here next year” to “This will never happen.”

As you might expect, I land somewhere between the two. I think we are about ten years away.

We will not get there on LLMs alone. If you wanted to make some kind of comparison, the easiest thing to do would be to compare AGI to a human brain. After all, AGI means that we will be approximating the output of a human brain sufficiently close to a real brain that it will be virtually indistinguishable.

LLM stands for Large Language Model and represents the ability to infer the next possible symbol that makes sense in a chunk of output, based on what has come before, and what the users ultimately asked about.

I decided to Google what part of the brain an LLM compares itself to, and if you were interested in that kind of thing, you can go do that yourself. For our conversation, let’s say it represents some percentage of our ability to process language. Really excited techbros might say sixty percent. Skeptical naysayers might peg that at twenty percent.

In my opinion, we will hit the maximum possible value coming out of LLM optimization within a few years. I don’t have some kind of law or some math words behind this. I am just trusting my own instinct here after watching four waves of technological evolution since the early nineties.

We are already getting pretty good outputs from these models. Where they fall down is in context management. We are constrained by the size of the processing model relative to people’s power consumption and CPU capacity. How big does your context need to be? And how much juice do you have to pour into the magic box to have it emit the things? That number may be marginally getting better, but as I understand it, we are linearly scaling relative to the supply available of each, at best.

As we converge on the best possible output for LLMs, we will start to see people building simulations of other parts of the human brain. And this is what we need to get to true AGI.

In about a decade, we will have four or five different systems that have the complexity of an LLM. Each of these will represent an abstract version of a part of the human brain. More importantly, we will have a governing system that decides which system or systems are needed at any given moment.

That combined set of systems will be reasonably deep, and I think it will behave functionally identical to a human brain. It will have flight or fight instincts, the ability to be pleased with itself, and a vast sum of both contextual knowledge about just about everything that anyone could know, plus some extra wetware to do different brain functions above and beyond hallucination reduction.

This is awfully handwavy. I think that it is the minimum of what we need to accomplish to generate AGI. We will not get there with just LLMs, even if we have experts writing prompts that make people wonder if their computer has the feels.

That’s it. That is the whole post.

Unrelated to that, Friday’s presentation on Tips and Tricks on Hiring in 2026, also known as “something something ai,” was great. We had a small group of people having a highly intelligent and valuable conversation about things we have all seen in hiring, and I think more than one person took away some new tools and tactics to experiment with when interviewing. I also learned that I should be checking the battery level on my Bluetooth headset. They gave up the ghost halfway through the show, and I had to put my giant air traffic control headset back on. I am going to re-record some of it and put it back into my Leadership Lighthouse site as personal coursework instead of a webinar. I will update you all on how those experiments are going.

See you all next week!

This is going to be short this week, and I do not want it to be, but a brother has gotta git payed. The rent check is due on Monday, you know?

If you have been following along on LinkedIn, you will note I am doing a paid webinar this week. I have spent the past six months furiously coming up with ideas on how to effectively, authentically, and musically interview remote candidates so you don’t have to.

I am not going to wax philosophical here about combating interview fraud. If you want to know more about that, come and listen in on Friday. It will cost you a few bucks, but you know, tariffs. Nothing is really free anymore. You can enter “mofactor” for a 25% off code, though, just to let me know that you read the blog. That is a permanent discount for all future products, not like the limited ones I am handing out here and there.

Real talk, I am the world’s worst Amazon Affiliate Marketer… or you people all have busted mouses. I don’t know which version of the story is worse.

If you are a hiring manager, or you play one on TV, you should easily get your money’s worth on tools and tactics for remote interviewing in the AI age.

Okay, I will shut up about the paid stuff.

I did add some free stuff, too, for job hunters. I added a ten-minute rambly monologue on resumes for your enjoyment. Sure, it is probably a lead magnet. I fully intend to never take profit from people out of work, but my ideas on how to fix your resume are SO GOOD that you will want to attend one of my webinars after you get hired, right? I thought so.

Now that we understand each other, go click the links, please. I asked nicely. If you need some help with your resume, I gave you some free advice. If you are hiring people and want some ideas on how to make that process smoother, I have some “less free” advice.

Thank you for your patience. I have big ideas scheduled for the next few weeks, but this week is all about shamelessly plugging my nascent engineering leadership program. I would love your support.

Thank you all, and especially thanks to Casey Rock, who helped me put together some videos this past week. He did a cute write-up of one of my foaming-at-the-mouth rants about hiring and the kids these days.

You should go read it too.

See you all next week.

I finally watched “TRON: Ares” over the past five days, and kept falling asleep and having to restart it. I suppose I should be grateful that I accidentally found it available on Disney+ for free. I have clicked on the movie at least once a week since it became available on demand, with a ridiculous “Buy Now for 19.99” right beside “Buy A TRON Bundle for 17.99.”

I also have about three more weeks of this for “Wicked: For Good.” Yes, I am going to watch it. It is as culturally relevant as the new TRON movie, but it has the correct amount of Jared Leto in a movie for me.

This goes beyond “something something Disney.” HBO Max pooped out the worst finale for a tv show in the history of the universe with the end of Peacemaker Season Two, and then politely emailed me that my rates were going up to that mythical 20 dollars a month. If you want some free life advice, do not ask for money right after you shit in someone’s living room. This is the first time I have unsubscribed from a streaming service, and it will not be the last.

I signed up for Peacock for 1.99 a month (with ads) last year, and when I went to cancel it before it went to regular price, they gave me an offer to resubscribe for 2.99 a month for a year. I could have waited until Black Friday and seen if they would have given me the same deal, but I am not going to fight them for a dollar a month. This is where my SNL is at, if nothing else.

Why are these movies so expensive for so long at home? Is it tariffs? Did the streaming people test the length of the pricing window to optimize revenues? Or maybe it’s tariffs?

I don’t know, but I am getting pretty sick of it. You might find this hard to believe, but I have never pirated a movie in my life. I work in intellectual property, and I do my best to respect its value.

Last week was the first time I thought about it, though… I was not so much excited for TRON: Ares as I was curious about it. My low expectations were almost met.

I managed to get most of a whole page written without saying anything about LLM technology, and I have a few more things before I pop the clutch and jump back into full tech-bro mode; Which is odd, because at my advanced age of fiddy-sumpin, I am more like a tech-poppi than a tech-bro. For real, I do not even lift.

Hollywood, dear sweet Hollywood. You need to blink twice. You are in danger.

The past few years have yielded tepid slates of movies, followed by crazy digital pricing, and flat-out franchise destruction in the eyes of many fans. Star Wars and Star Trek, I am looking at you. There are entire generations of people who are nervously showing people where on the teddy bear their favorite franchises have touched them. I am one of them.

So we covered the insane pricing for digital goods, we covered the movie releases being unexciting, and finally, the fact that there are parties in the street that Kathleen Kennedy has stopped doing whatever it was she was doing to Star Wars. Yes, I did a little dance when I read that.

Let’s add to that some of the franchises that are putting out new content too fast for the franchise. I am seeing both “you should go see 28 Years Later: The Bone Temple” and “28 Years Later: The Bone Temple has mixed reviews.” With that much mixed signal, it was certainly not going to be a “first week must-see movie” of the kind that Hollywood needs to feed its financial beasts.

And wasn’t it like… March… that we just had an Avatar movie?

Why does this matter?

Because of Sprint, Verizon, T-Mobile, and ATnT in 2007.

Bear with me for a few minutes. Hold off on calling security. I will attempt to explain everything.

Once upon a time, you did not buy your mobile games from Apple and Google. They were put for sale inside your feature phones and charged to your carrier bill. This is one of the places I cut my teeth as a developer, and I had a great time shipping tons of cute little games and helping people with porting from one phone to another.

The carriers were not very good at managing games. They had around 500 games under management each year, and it made them very tired to deal with all of that. So one day, someone woke up and attempted to do some deep reasoning about their business.

They said, “We push out 500 games a year, but only 100 games a year are successful. So… Let’s just release 100 games a year!”

This coincidentally destroyed the whole middle tier of mobile game publishers, who were financed and staffed around getting 10 to 20 slots on a carrier deck for their digital goods. Suddenly, that would be 2 to 4 games.

For yours truly, I was at a mobile company doing a mix of contract work and self-publishing/co-publishing fun little games. We were succeeding with a 1 in 6 hit ratio.

If you divide that number by five, like the carriers did, then we would not break even as a business anymore, so we shut down our company.

Many people got out of mobile games at that point.

The carriers deciding to strangle their games business to death came at a perfect time for Google and Apple. The carriers taught people to play games on their phones, they gave developers an opportunity to make many fun little experiences and figure out how to work within constraints for the devices, and essentially created this massive unfulfilled need between these two groups of people who could not connect through experiences like they used to.

Two to three years after this massive self-inflicted content desert occurred, modern-day smartphones put out their own app stores and started filling the empty hole in people’s hearts and heads where mobile games used to live.

You might see a slight comparison—Hollywood is releasing fewer movies, making them more expensive, and generally taking fewer and fewer creative risks.

They have stopped feeding their audiences the right kind of content, the right amount of content, or the right price for that content.

Here comes my big fat LLM / AI comment.

Much like the carriers salted the earth for mobile games in many different ways, Hollywood is chasing away its fans and customers. They are poisoning the well with consumers, overcharging for windows that are too long, and failing to innovate on value creation.

So when we are all watching AI-generated micro-movies at home to pass the time because we do not want to see a Star Wars movie with Jared Leto in it, I am going to sigh and shrug.

Hollywood is not going to be taken out of business by AI-generated movies. It mostly has itself to blame for its own demise; Hollywood has been slowly strangling itself to death through all of the ways I have described above.

So what is to be done about it? Pretty much nothing, honestly. We will politely smile and nod as we get fewer and fewer big blockbusters and tentpole franchises out of incumbent media businesses. The death spiral has started. Anything they try to do now will be too little too late. By the time people get good at making movies with LLM technologies, we will be so starved for content that we will dive into new shows and formats with both feet.

On that note, I am going to go and check on Wicked: For Good pricing.

See you all again soon!

Ah, yes, Settlers of Catan. What a great game.

My wife and I are convinced that “Settlers of Catan” is what we will write on the line for “Reason for divorce.”

If you are a math nerd and know how to count dots better than other people do, or know why you should count dots, and also love to play the game according to your own twisted, psychotic rules, then this is the game for you.

But what is so special about it?

Quite a lot. Which is why Catan is probably the single most toxic licenseable digital product ever.

“We got a license to Settlers of Catan” is a statement that irrationally exuberant people make when they have spent some kind of money to get a license, sure to drive their adoption.

People like new toys, right?

People like to play familiar games on new toys, right?

So how could you go wrong here?

It turns out that the real fun in Settlers of Catan is not to be found in the joyless rejection of trade offers from randos on the internet.

The real fun of Catan is in knowing the rules for when you can turn over your final victory point to make someone at your local table so mad that they flip the table over as you crush them.

And yes, I have done this before.

I know more than one person who has smugly admitted they have licensed this particular product for a new digital platform.

And for the first few times I heard it, I was impressed. And then after watching a few of these businesses catch fire, fall over, and then sink into the swamp, I began to wonder about that.

It turns out that some things do not transition well to digital.

I was a director of engineering at Zynga Poker for a while, and unless you are playing with the high spending VIPs, playing Texas Hold ’em on your phone is far from playing real poker. The stakes have to matter, and some random person on the internet named FlibertiGibbet6969-420-67 going all in and LOLing with nothing in their hands over and over again just ruins it for everyone.

Intimate social experiences matter for some people, for some experiences. If you want to know what I mean, go put $100 into an online hold ’em poker game and go play on Christmas Eve. I got cleaned out so fast by sharks, it made my head swim. The filthy casuals all have gone home.

And that is what is true about Catan. The casuals do not want to show you, on the teddy bear, where Catan touched them. It is a savage game. It is designed to make someone angry. It is a comfortable destination for sociopaths to flex their math skills and flaunt their superior understanding of psychological warfare.

And all of that is lost on the internet. No one will trade wool for stone. There is no clever barter with people where you reason with them, to cause everyone to start an embargo against you. And there is no ability to make a 3-for-1 trade in someone else’s favor to break that embargo, to make that person suddenly the victim of everyone else’s ire.

So even though we all love playing Settlers of Catan as a board game, the vast majority of the people who have put it online are making a pale imitation of a shadow of a whisper of a dream.

And when that is your strategy to de-risk your business, you are likely to wind up a pale imitation of a shadow of a whisper of a nightmare.

I am become Catan, Destroyer of Platforms. License me and despair!

Welcome to 2026, everyone.

It is time for me to start lining the bottom of my monitor with topics. I have a variety of topics to talk about for the first month. I am so excited, and I just cannot hide it. I am about to lose control, and I think I like it.

That last line isn’t mine. I borrowed it from a song. Do I owe someone a royalty? It seems like that is what fifty percent of the Carl Complainy-pants people who hate AI complain about. I play whack-a-mole with them on LinkedIn, cheerfully, because whack-a-mole is fun, but also because I am trying to help people stay employed.

“I hate these AI tools; they will destroy my job!” is mostly a problem for people who go to work wearing shirts with AI on them featuring a red circle around it and a line through it. When the bean counters come in and say, “We are doing great with the people who have leaned in on productivity-enhancing tools, but we have some redundancies; who is not on the bus to AI-Town?” You can guess who gets in the crosshairs of the pink slip cannon.

But that is not today’s conversation.

Today, we are going to do what I vaguebooked last week.

I want to talk about hiring in the age of AI.

This came up as one of the things I wanted to talk about when last year, I accidentally shot my impressions through the stratosphere on this subject. I have done some successful engagement farming on LinkedIn, but this? This was some next-level stuff.

So it stands to reason that people want to talk about how to hire in a world loaded with AI tools, some of which are designed to foil the hiring process.

So what are the problems?

The first one is the easiest one to solve for, and it is self-obvious enough; I am just going to put my junk on the table.

“A different person showed up to work than I interviewed.”

People are having other people interview for their jobs. How crazy is that?

The best thing to do here is to take a picture at each stage of the interview and post it to a hiring medium (folder, file, Slack channel) so that everyone can see that it is the same person, or not.

Make sure that person matches what they look like when they get to work, too.

It goes without saying that you should never hire a remote worker who will not turn on their camera. I will accept a blurry background.

If you are running your interview in expert mode and are afraid of someone being a bad actor trying to get a job, the easiest thing you can do to enable them is to not look at them.

Camera isn’t working? The interview is over. Next candidate, or reschedule for when the camera is working.

This is an example of the problem, and it is not even the freakiest one. Someone built a tool to sit in on your interview and answer your questions for you.

In 2025, I did not see these problems for about half of the companies I was working with, and I will confess, “I want to interview a bad actor as a candidate” became a bucket list item for me.

Why wasn’t I seeing these candidates? I kind of panicked. It turned into an obsession.

The answer was ultimately not surprising. It was because I was interviewing for hybrid roles. It is harder to be a remote super secret spy angling to get at your digital goodies when you need to show up in Hoboken, NJ, twice a week.

I have recently started helping companies recruit for fully remote roles internationally, and boy oh boy, that is a very different world.

You have a lot of questions that you have to answer, and sometimes people will sneak through the interview. It has gotten crazy enough that when someone is interviewing, I am tempted to take a screenshot of their wristwatch, or a clock in the room, or the reflection of their screenshot to look for what time it is… Some people I have interviewed have lied about their country of origin and making sure they are in the correct time zone by detective work has crossed my mind. Are you in London? Open the windows at noon, I want to hear some church bells.

So to recap:

• People might have other people do their interviews
• People might be in a country other than the country they declared
• People might be cheating in real time on their code tests
• People might be cheating on any take-home assignments

So, how do you hire successfully in 2026?

I have refined a hiring playbook, and I am going to use this as a launchpad for joining the modern-day expert coaching community. I will have more on this later, once I have gone beyond the Ghostbusters “Guys, get her!” level of planning. I have Meetings With Experts on this. Everyone says they want a solution. Let’s see who has some dead presidents for an hour-long conversation on this subject. There is a lot more to come on this subject soon.

Hello everyone. I am back in the blogging business. I had some reasons for stopping last year, and in hindsight, it was a bad decision for me. When you are asked to change something you believe is an important behavior, you should ask yourself, “Who benefits?” In this case, it was not me.

I am going to kick this year off by making some predictions. This is probably the most important thing for anyone who has ambitious goals for the future because the best way to predict the future is to create it.

That being said, I have read about a dozen “predictions” for 2026. And almost all of them are boring and obvious. What is stopping people from making big, bold bets? Probably the same thing that kills a large swath of startups every year: Risk aversion.

We can debate that in the inevitable LinkedIn comments, since as a fractional CTO, I just offended half of my potential employers and customers. Although if you are talking to me about helping your business, you are clearly in the other half of the leaders, right?

Enough pillow talk.

Let’s make some crazy predictions!

In 2026, we will see:

• A billion-dollar company’s valuation decreased by 30% due to issues using LLMs (compliance violations or data corruption).

• The release of a mass-market AR device, which will reach 4m users before EOY. I should caveat this one, because I want this to happen very badly. I have a steak dinner in SF riding on its outcome. If I get this one wrong, it is because the powers-that-be want to see me take someone out to dinner. Possibly two, because I have a double-or-nothing on the bet due to another prediction for 2026 that is audacious, and perhaps will not happen until 2027.

• Two. Billion. Dollars… Of mobile publisher consolidation as the red ocean accelerates in a winner-take-all market.

• The job market retreat will slow and even stop. Jobs will come roaring back, due to AI, and the need to shepherd and grow the LLM-features-to-customer market.

• Sovereign LLMs and on-prem deployments will cause valuation shifts in the big AI LLM providers, due to the first point, and general cost management. The LLM provider counterattack will be in these companies selling LLMUaaS; LLM Updates as a service, and they will have a haven in Docker-style models with a support contract for updates.

Phew! 

I am going to stop there. I only made half as many as a top ten list. But these are meaty, and they pack twice as much of a punch.

Stay tuned for next week.

I am going to make some announcements about products and services for engineering leaders who have 2026 problems.

First up: How to hire in the age of AI.

Now that I have decided to stop adding topics to the blog, it feels like the air is slowly being let out of the balloon. What is left to teach? Maybe I should just blow out the list of topics in one long rant and then be done with it. Yes. Let’s do exactly that.

So, let’s tear apart my backlog and give you one final blog post filled with rapid-fire advice.

Have pride in your work – When you ship your products, ensure the little details are handled. Are pixels out of alignment? Are your controls properly formatted? If you have a Date Picker UI control, did you ensure it is clean, or have you left the insignificant time elements with minutes, seconds, and microseconds at the end, with the actual digits word-wrapped out of the display like some sort of insane hobo? If you do not take small amounts of time to ensure your products are shiny, then you are just putting garbage into the universe. Please don’t.

Learn from everyone – Everyone around you has something valuable to teach, not just your boss, not just best-selling authors. Learn from your teams. Learn from your peers. Learn from your kids. This last one is quite serious. Youth sports has taught me so much over several years as a coach, a referee, and a spectator.

How will I teach this to future me? This blog encapsulates a lot of this question. What are the things I can do to help someone who is on my career arc but is several years behind? How do you provide coaching, storytelling, projects, and feedback to people to get them to where you are now, but relatively earlier for them? This is one of the more active questions I ask myself daily.

How do you avoid FOBU? The Fear Of Blowing Up is real. You have to try things, and you have to be prepared for them to fail. You cannot wait until you are 100% confident and have 100% of the needed information. You have to make imperfect decisions. Sometimes, things will blow up.

How do you demonstrate leadership as an IC? I have identified the biggest career gap for most people wanting to be a senior IC and, eventually, a leader. You have to do more than just identify the problems in your team and organization. You must propose credible solutions and spearhead the work to fix those problems. “This is what is broken” is not as effective as “This is what is broken, here are some possible ways to fix it, and this is how I want to fix it.”

Drive-by management – Some of the best managers and leaders I know seek opportunities to grow their people. Delegating your work creates massive growth opportunities for your team. Look for curious and motivated people, and give them interesting work. Discuss how you would do it differently. If there are gaps between what you would do and what they would do, explain them. Make sure you regularly know who is interested in this kind of work, and confidently assign tasks that give room for career development.

Be loudly wrong – According to Amazon, Leaders are often correct. This creates a fear of speaking up and saying the wrong thing. Be sure to give your teams a voice to be heard by letting them know it is okay to share ideas and propose things, even if they get shot down. Whenever I attend large engineering events, the same ten people are always willing to speak up or ask questions. Those people are the people to watch professionally for how fast they will ascend professionally.

Core hours are amazing – When you work from one side of the continent to the other, core hours are a great thing. Knowing that you have a window of time when you can meet with people in a remote-first world helps businesses share information smoothly. It also gives people the freedom to do their IC work and time to think independently.

Know the blast radius for failure – Finally, one of the important things to do as a leader is to understand where you can fail and how. You need to give people opportunities to fail to grow. One of the biggest challenges in today’s engineering team culture is creating opportunities for senior engineers to become staff engineers and architects because of their current engineering leadership. Either the current leadership is unaware that they have put all of their team’s output on a super-safe railway, prohibiting the ability for people to stretch and do more architecture work, or they know and are too focused on day-to-day business results to give their people the space to stretch and grow. Both will cause your best people to churn. Know what projects you can give over completely to your teams to design and build without causing irreversible damage to your business or your customers.

With that last point, the bottom of my monitor is now clean for the first time in four years. It is an emotional moment for sure. It has been a great voyage together. Thank you all for reading along as I do some sort of strange career penance by writing down all these things. I also want to thank our good friends at 3M for their high-quality Post-it Notes, without which I would have had to do some kind of Trello bullshit to track this work.

The last person on this web page, please turn out the lights.