Civil engineer that loves technology and developing software tools. Currently building QtHydro - QtHydro.com
Atlanta, GA
- Tweets 7,531
- Following 398
- Followers 407
- Likes 19,126
QtHydro retweeted
If you don’t believe Jesus is God or you don’t believe in God at all then I completely understand that.
BUT there’s a cradle Hindu who became a Christian based on the historical evidence and he’s doing a talk about why this Friday in SF. You should come see.
This Friday, I’ll be giving a talk on the historical case for the resurrection of Christ.
Several years ago, I began looking into the historical evidence as an atheist to prove that Christianity was nothing but a therapeutic superstition. What I found completely changed my life.
Over the course of a year, I spent nearly every waking moment reading whatever I could find on the subject. My entire life was suddenly reorganized around answering this one question: did Jesus Christ actually rise from the dead?
There was a time when I thought this question was too ridiculous to even consider. But looking back, it was by far the most important question I've ever asked.
If you're interested in learning more, check out the link to the event page below. All are welcome to join regardless of religious background or lack thereof.
47
112
2,448
68,861
It’s interesting that Ossoff rarely if ever criticizes Kemp. He is savvy enough to know who the punching bag should be (Trump).
24
When people ask what job is next for programmers, it would be this guy's job.
If done well, this guy is getting paid better than 90% of programmers...
I have reviewed over 75 proposals and sat in a couple dozen meetings over the past two weeks on behalf of a big name fashion brand trying to hire an AI agency to lead their AI implementation.
This will be their 4th agency since they started opening up to AI back in March 2025.
They reached out to me from a referral because they need help finding a good agency that won’t overpromise and underdeliver. I provide the guidance, oversee the training of employees, and ensure the agency is capable of supporting a full organization wide transformation process. Not just marketing, not just ops, not just one little workflow. The whole company.
I have time for agencies that are trying to do right by their clients and know when they need help. The markdown agencies selling expertise they can’t explain and expecting Claude to make every decision for them are the first to be tossed to the side.
Some gaps that keep coming up in proposals and meetings with me and the CTO as we interviewed the agencies:
- You can explain the tech stack and demo for PDF data extraction on Purchase Orders, but what happens when an approved order changes? Who needs to act on it, and what gets updated downstream?
- You quote three months to the first build and pitch another audit. What can you reuse from the last three agencies, and what still needs checking?
- You send every artifact to AI with minimal guidance, so it is easy to tell that we are communicating through Claude. But when we ask them to explain their recommendations in the meeting, they can’t. They run the meeting transcript through AI and send us next steps that are all redundant, but when we ask them in a meeting for specific insights they say they'll get back to us.
- You sell hours saved as ROI by multiplying time by salaries. Those salaries aren’t automatically going away. What work can those employees take on now, and how does that benefit the company?
- You claim to sell "outcomes", but the outcomes you describe provide observability, not actionable steps and proven systems. An AI agent that gathers all the data and lets you query on it is useless compared to a system that uses that data to get more customers and provide more throughput all on its own.
A good agency needs to be able to pick up where they were at, check what actually works, and make better decisions based on that information, and re-assess as they go.
A good agency would come prepared with the ordering, fulfillment, and sales processes they expect to see. Explain where they think the bottleneck is and ask, “Is that accurate?” We can confirm or deny, but at least it shows where they are thinking and the expertise they would provide... then the agency can explain how automating one part leads to the next improvement or depends on a previous step.
For example, you automate the orders coming in and get all of the data extracted, nice and good to go. The people working with it afterward are reconciling it, confirming it, and connecting those orders to fulfillment and cash flow.
Once we’ve proved which numbers we can trust, they can spend more of their time doing predictions, testing, analysis, and bigger scope items that can actually drive revenue.
Now we can start pulling levers. Say an item was successful at a certain time last year. Let’s put a marketing plan in place, test how it does with different traffic, see how the cost changes, and test new prices. Check the margin and whether we have the inventory to support it. Those are all steps that will eventually be automated, and a good agency would need to know how to support that pathway or at least have a strategy to explore. I've done this whole process at over 200 companies and even if you're at your 1st or 10th implementation, it is possible to know these things and come prepared with it.
A lot of bad agencies, aka "markdown agencies", are trying to get away with doing the bare minimum by pitching "AI audits", "AI Enablement training", and "maintenance fees" to help sustain the systems... but what they actually deliver are roadmaps generated by Claude, a weak AI guide as they help setup a Claude Team setup for an org whose identity and governance requirements it doesn’t meet (bad for any org that needs to take security seriously), and the maintenance fees are often met with bad service long term.
The last two agencies that have been considered for this opportunity both pitched $75k-$150k per month, but they could easily get $250k per month from an org this size if they just came prepared with the right items and momentum behind their builds.
As of today, I finally hit 250 companies where I have led, implemented, or consulted for full, org wide, "AI transformation". All while maintaining a full time job, what a blessing.
Rewarding myself with a three day weekend goodnight.
1
66
QtHydro retweeted
I spent the better part of 5 hours last night trying to get Gemini Pro to map atoms of a crystal between two thermodynamic states (ended up staying up until 2 AM pulling the AI slot machine lever). I asked it every way I could imagine. Fed it inputs and results, even gave it hints. No matter how convincing the solution sounded, the results were complete slop. I'm now creating the map myself.
I installed every python package it asked for to help it find a solution. While it wrote decent code to implement the strategies it proposed, it just couldn't wrap it's head around how the crystal unit cell repeats, even though it knew what the structures inside should look like. It couldn't look across one boundary and see the other and connect the dots.
What it helped with:
1. Finding existing workflows for solving similar problems to what I am working on (created by humans).
2. Writing boilerplate code for those existing workflows (written and heavily reinforced by humans).
What it failed miserably at:
1. Symbolic reasoning. Even though it knew the exact sub-structure of the chemical groups in the lattice, it could not identify those groups from coordinate data. Trivial for a human to do instantly. It immediately resorted to statistical methods and existing algorithms to try and identify the groups and failed because the algorithms were not written to handle my specific case.
2. Spatial Reasoning. Even though a simple crystal visualization tool can immediately pick out covalent bonds and group them in distinct groups (albeit without wrapping across unit cell symmetry), the AI was helpless to recreate even this. Even when it was telling me that a group was cut in half on a specific plane of the unit cell, it couldn't stitch those two together. The words sounded right but the mental understanding did not exist.
3. Context and logical reasoning. The context was all over the place. I told it multiple times what the ultimate goal was, and it never could make very good decisions about what corners I could cut to make the approach more doable and elegant without throwing out crucial details. Again, it just regurgitated what others have done in its training data for problems of the same class.
This is a great example of what I mean when I say that these things are still stochastic parrots. The things it vomits up are now processes instead of just words, but the underlying statistical methods have not changed over nearly a century.
This is so clearly not intelligence. It is just better, MUCH More expensive, search. Yes, the search blows away existing html metadata indexes, but at what cost, and is the cost justified if it's just really good search?
22
20
223
12,134
QtHydro retweeted
In addition to the books, this is what I grew up with. You have to understand, Greta Gerwig isn't adapting just any book. In addition to having profound religious and spiritual implications, these stories are nostalgically sacred as part of our childhood. They were emotionally and spiritually formative. Modeling Aslan after David Bowie is like modeling Gandalf after Michael Jackson, or Harry Potter after Britney Spears, and then acting surprised when fans are irritated. She isn't stupid. She didn't do this to make fans happy. Quite the contrary. She's giving fans a proverbial middle finger.
Toda vez que eu vejo alguém falando sobre os filmes "originais" de Nárnia, ou "prefiro os antigos", só me vem à cabeça a versão dos anos 80 haha
41
58
632
21,166
QtHydro retweeted
supise suprise!
I have tried having agents port one of my 3D tools (a 2D/3D sketching app) from desktop to web. Not even 1/10th the size of Photoshop. I gave it the full source code. It had Playwright. We spent several hours planning. I set up guardrails. I said "validate everything against the source, do an amazing job, make no mistakes".
It was still a shitshow. It doesn't matter how good 'computer use' gets, coding agents do not understand what is important vs not-important in a complex UI unless you tell them explicitly. They are pretty good at basic buttons and other trivial web UI stuff. Ask them to build a custom widget for a precise 2D/3D interaction without super explicit instructions, and what you get is nearly always dogshit. They can't know what "feels right" because we can't actually describe it in words.
They cannot keep even a small fraction of the actual app code in context. So they are working off 'memory' which means they are constantly deviating. It works OK for math/algorithms/libraries where things are generally componentized and easy to validate. It does not work for framework/UI type stuff where there are little bits of functionality across the entire app, that have to interact *just right*.
(you know, the kind of stuff that gives you Photoshop vs GIMP...)
So even for porting a small app, you have to be on top of it constantly, checking everything it does, because it's going to fuck up *all the time*. And not just in all the small ways. Also ways like "This button doesn't work because the agents interpreted one single phrase in the prompt in an unexpected way and so decided to skip the entire subsystem the button talks to".
Do you think the guy robo-porting the entire Creative Suite in a week is putting in that time and attention? Testing everything (anything)? Clearly not. That's why it takes huge teams to build this kind of software. Because at some point someone - often a whole team of people - had to give a shit about every tiny little bit of it.
The robots are amazing code-typists. I haven't typed a line of code in months. But I don't let it decide how a UI should work, that's stupid.
I wanted to show how janky the AI slop is of the FilmCraft software.
I wanted to set the stage for how bad it is, and how little care there has actually been in implementing this tool, and most definitely the entire Craft suite too.
It's all AI slop that does not even work.
31
24
240
17,624
QtHydro retweeted
Mike was my colleague at Berkeley, and he's been fairly skeptical of deep learning's trajectory for years. Most scientists are skeptics, and it can take a big shock to get someone to consider a possibility that goes against the anchoring effect coded in our DNA.
I too was somewhat of a skeptic until ~6 months ago, and people ask me all the time why I joined Anthropic.
It started in April, when a friend I've known for 20 yrs messaged me saying he'd be in the Bay Area in a week, and he suggested we meet up. I invited him to my house for dinner, and he told me he had just joined Anthropic. He said the models were becoming increasingly more capable, and he believed both that they'd soon be smarter than him and also that all intellectual work would be highly devalued in the near future, and lastly that society was not at all ready for what's coming.
I'd of course heard claims similar to this before, but not from a friend I'd known for this long, and who I trusted, at my dining room table. It was the initial shock needed to destabilize my own anchoring. Also, I've been fortunate to work with and/or mentor many smart people in my life (e.g. Fields medalists, a 4x Putnam Fellow, IOI first place winners, etc.), so I feel pretty calibrated in recognizing very smart people, and this was someone who is definitely very smart and with impressive such credentials. So, it was hard to ignore this friend saying the models would be smarter than even him in particular.
After that dinner, I decided to do an informal survey, calling and meeting up with a lot of my friends who now work on improving model capabilities at these frontier labs. They all agreed with that first friend. These were all people I'd known 10, 15, 20 years, who mentored me, or I mentored them, or we met professionally in academia. These are people I trust, and their beliefs on this topic were consistent.
I then spent a whole lot of time understanding the missions of different companies, and talking to current and former employees and people in leadership. I read Dario's (long) essays and watched videos on YouTube of company leaders being interviewed. Anthropic's stated mission "to ensure that the world safely makes the transition through transformative AI" felt like exactly the right guiding principle to handle such a society-disrupting technology, and everyone I spoke to at the company was very aligned with this mission. It's easily the most mission-focused place I've ever been at, and everyone takes the work extremely seriously and with a profound sense of responsibility.
And now here I am. And being on the inside, the pace of progress has surprised me more than I first imagined. The moment that pure mathematicians are in, to me, is just a microcosm of what I predict a much larger of society will be grappling with in the not too distant future, and I really hope our leaders and the general public internalize sooner than later just how significant a moment we're in.
Michael Jordan (the machine learning one): "Superintelligence is Science Fiction" 🔥🔥🔥
di.ens.fr/~fbach/MJordan_ITW…
72
208
1,968
328,781
QtHydro retweeted
My story with God.
I mostly talk markets on here, but today I want to share the story of the day I knew there was only one true King.
I grew up in church. Had a great foundation of what a relationship with God was supposed to look like.
Church on sundays. Bible study. Summer camps. Singing. Giving. Serving. Volunteering. You know the typa fam.
I was "doing good" in the world I guess, but didn't really feel like I truly knew Jesus or had a relationship with Him.
Felt like i was kinda larping as a christian to be honest. But then in my late teens & early 20s, life was getting ugly and I needed someone. Something. There was a gaping hole in my life/heart.
Because of my upbringing, i knew God was the answer but i didn't understand how to find or build a relationship with Him. Just felt like prayers were going unanswered.
^ spoiler: they weren't. I just needed to be patient. Here's where my story changes.
In the midst of struggles/trials in my life, i went to church one sunday and for whatever reason i felt different this day. Just off. Cant really explain it but whatever.
But the message from the pastor that day was about "miracles in the marketplace"...meaning the power of the Lord isn't just contained to the church walls or some building. I thought that was a cool idea.
Miracles can happen anywhere. So, as I got in the car after that church service. I was praying out loud (which I rarely do) for God to perform a miracle in the real world, my world, before I arrived home from that drive.
I had one stop during the drive. At an Aldi grocery store. I went in and did my shopping. Nothing crazy. Normal stuff.
But as I was walking out, i was thinking "wouldn't it be cool if a homeless person walked up so I could give them some money or something". Kinda a dumb thought in hindsight haha but whatver.
But if I gave a homeless person money then I could chalk that up as the "miracle" (was my thinking lol), but as I was walking out to this huge open parking lot, there was no other cars within 50 feet of my car. No people anywhere in sight...and I mean I was looking hard.
Not a soul was around.
Literally nobody. so I get to my car, open the trunk, toss my two bags of groceries in in less than 10 seconds...and as I'm swinging my big toyota rav4 trunk closed...
I kid you not, this man appears out of thin air. He did not exist. I have chills writing this. He appeared out of no where. He didn't ask for anything. He just said "hello".
I stood there for a sec and said "hey man, you scared me" (because it made me jump). He replied, "do not be afraid, i was just wanted to see if you wanted to grab a pizza. this is a little embarrassing but im hungry"
There was a domino's in the same parking lot, so I said of course. It was pouring rain but we walked over. I got him a pizza. His name was John. He was a homeless veteran. Just fallen on some bad luck.
As we were waiting for the pizza, he was telling me his story, and he said earlier that day he was praying for a sign from God. For someone to be kind to him. To eat with him. Treat him like a human again.
We sat for hours. He's weeping in the Domino's saying God has answered his prayers, and he's going back to the homeless shelter to tell everyone everything. Meanwhile, he just changed my life forever.
A miracle in the marketplace. An answer prayer. A supernatural encounter. A relatinship with the one true King.
So, on this Sunday, remember that while this life can be blast. Making money is fun. Traveling is cool.
But it's all fleeting. One day, none of the money, the status, the followers, or the "things" we spent our lives chasing will matter.
I spent years thinking I needed some massive sign from God to know He was real...turns out, all I needed was a pizza with a stranger. My DM's are always open.
Happy Sunday everyone, let's have a week this week!
163
125
2,064
125,521
QtHydro retweeted
If DOE and DOD team up, they will build one type. That changes everything. See my op-ed in the WSJ today. wsj.com/opinion/hegseth-empo…
"Almost every reactor has been a special snowflake." DOE official on our nuclear call today. She reminded us that 50 "first of a kind" projects have robbed us of lower nuclear costs. We have to change our approach.
1
4
7
552
Installing a closet door is the bane of my existence today.
I have to try and take the minimum off the edges so I don’t go through the solid sides. Out of square opening. I hate doors.
1
67
QtHydro retweeted
Guys on this app who have never built anything in their life will casually shit like “AI powered humanoid robots will end factory work”.
Bro.
Do you understand how overpowered humans are?
- We can see in 360 degrees in two planes
- We know if something is touching us
- We can pick things up without destroying them
- We can smell a gas leak
- We can hear with location
- We require zero maintenance and recharge on just 8 hours of sleep a day
- We can move from place to place
- The entire world for the past 10,000 years has been designed for us
I could go on and on.
AI is no doubt amazing and a game changing technology that is more powerful than we are in the digital world, but in the physical world we are and will be for a long time, king.
171
60
824
44,705
QtHydro retweeted
Joe Biden surrounded himself with people who embraced modern monetary theory, a heterodox academic fad that taught you could write as many blank checks as you liked without the consequence of inflation. Larry Summers, Bill Clinton’s Treasury secretary, warned Democrats not to go down that road. Conservative economists warned Democrats not to go down that road. Biden listened instead to the modern monetary theorists and their loud defenders online, ultimately signing a $1.9 trillion spending bill on top of everything else, and gave the country inflation that peaked at 9.1 percent in June 2022. Democrats paid for it at the ballot box.
Unfortunately, President Trump has made a similar mistake. He has surrounded himself with men who staked out heterodox views on the economy decades ago and now treat his administration as a laboratory. Peter Navarro has spent his career pushing trade theories that nearly the entire economics profession rejects, and he once propped up his own books by quoting an expert named Ron Vara, who turned out to be Peter Navarro.
Outside the White House, Oren Cass, a lawyer and former management consultant with no formal training in economics, founded a think tank, gave himself the title of chief economist, and has become the most prominent defender of the tariffs on the right. Around them is a chorus of loud social media defenders who insist tariffs are a tax on foreigners.
The MMT crowd promised that massive spending would not cause inflation. It did. The tariff crowd promised that tariffs would not raise prices or hurt wages. They did both. We no longer have to argue about it as a matter of theory. We have the data. The tariffs raised prices and held down wages.
The administration now refuses to change course. That is arrogance. It is also bad politics. The tariff regime has turned into a favor generator, with exemptions going to whoever hires the right lobbyist. Voters now rank corruption as a top concern for the first time in a generation. Republicans are heading into a midterm with voters angry about prices, angry about corruption, and trusting Democrats more than Republicans on the economy. Much of that is self-inflicted, and the tariffs are the biggest self-inflicted wound of all.
ewerickson.substack.com/i/21…
143
201
1,192
134,122
Feels like this report is going to generate a lot of articles, posts, and takes: wsj.com/economy/affluent-and…
43
137
894
106,066
The doomers on this site need to log off and get some perspective on AI.
Go walk most businesses and there are years worth of optimization opportunities sitting there. There is a complete lack of true skilled workers across industries.
Stupid policies and governance will kill more jobs than AI will for the foreseeable future.
2
77
QtHydro retweeted
Humans provide a fun comparison point here.
In the 80s, the "cognitive consequences of programming" was a major psychological research focus. The prevailing hypothesis was that learning formal logic and algorithmic structures through computer programming would act as a form of mental gymnastics that could upgrade general reasoning, planning, and novel problem-solving abilities. But foundational research, followed by subsequent meta-analyses, demonstrated that this does not happen. Students who learn programming become highly proficient at algorithmic thinking and coding, but these gains do not transfer to general cognitive tasks or unrelated problem-solving scenarios. Similarly, intensive mathematical training improves mathematical deduction and structural mapping, but does not increase an individual's baseline rate of skill acquisition in unrelated fields.
General intelligence seems to be a fundamental property of the brain rather than a skill you can train. Practicing in a domain makes you better at the domain but does not make you generally smarter.
What if the jagged frontier is mainly math + code (which you can push arbitrarily far with RLVR), and everything else starts to plateau because it is still bottlenecked by human generated data?
Model performance in non-verifiable areas has kept improving steadily, albeit much slower than for math and code. But is that steady improvement a side effect of a higher G (itself driven by RLVR), or only a function of the amount of new human data getting injected into training (which is still continually happening on a massive scale)?
A lot of things depend on the answer to this question
93
136
1,345
74,481
