@realGLMartin profile picture
Greg Martin @realGLMartin

Fakin it till we make it! Check out our link to see how we unwind.

Joined October 2012
  • Tweets 2,067
  • Following 337
  • Followers 223
Photos and videos
Greg Martin retweeted
Rich O'Toole
@RichOToole
May 28
New single on Spotify. Please retweet or 7 more years of beans in chili. open.spotify.com/track/2ddsN…

Baby We're Both Right

Rich O'Toole · Baby We're Both Right · Song · 2026

open.spotify.com
2
13
26
1,282
Caught the Triumph (w/ April Wine) show on Wednesday night. My post-game notes... I thought April Wine did a stand-up job, even though they only had one from the lineup we all loved. Time is undefeated and I'm glad they gave us that trip down memory lane. Triumph, IMO, knocked it outta the park! When I saw the 2 drum kits, the raised keyboard station and the 3 front stage mics.. I was very confused. "What is happening to my beloved trio!" This is feeling like Earth, Wind & Fire and .38 Special had a baby! But then, it all made sense. At 73 years young, Rik and Gil can't belt it out like they could before. And, the 2nd drum kit was a stroke of genius. Letting Gil step up front was a great way to show his personality and his front-man skills. I loved it. His stage presence was cool. His voice was on-point. It was great. Rik's voice can't hit all the same notes as before. But it's still a beautiful voice. And when the intensity ramped up on the vocals, that's what the bass player and the other guitar player were really there for. And they were perfect for the job. All that to say, I thought Rik and Gil threaded the needle just right in terms of carrying the load themselves at times and handing it off to the next generation at times. I thought it showed tremendous respect for their catalog and, more importantly, their fans. We can never go back to the 1983 Texxas Jam, but Wednesday night was pretty darn close!
1
10
@rikemmett @TriumphTheBand My wife said I should tag you guys to make sure you see how much you were appreciated. Thanks for Fighting The Good Fight!
4
Does @colincowherd just keep JMac around because of the contrast? The gap between the two definitely makes Colin look like a genius. So I guess it makes some sense. But letting JMac guest host, ever, is tarnishing the Herd brand substantially, imo.
57
Criticize me all you want for posting something "written by AI", but I think this response is worth a read. I posed the following question to AI... Agents are "trained" on the billions of lines of code written by humans over the past 30 years. I get that. In those billions are the thinking for how to tackle an almost infinite number of use-cases or patterns. But, before agents came along, every year we got a bit better at it. Or, we came up with newer, better ways to solve the problems. Or, even, we solved entirely new problems. If we all go to "agentic" coding, doesn't that imply that we are "freezing" progress at a particular point in time? 🧵
9
83
End
36
The uncomfortable conclusion Agentic coding does not make humans obsolete. It makes undisciplined thinking obsolete. If you don’t know what you’re building or why: the agent will happily build nonsense at scale If you do know: the agent becomes a lever unlike anything we’ve ever had Watching days of work collapse into minutes — is not the end of innovation. It’s the end of confusing effort with value. One last framing, because it matters Progress never froze when: calculators arrived compilers arrived databases arrived the internet arrived It only froze in places where: people stopped asking better questions Agents don’t stop us from asking better questions. They remove the excuse of not answering them. And that’s a much more demanding world to live in.
37
A useful mental model Think of agents as: A perfect memory of the past, with no opinion about the future. Progress has never come from memory alone. It comes from: dissatisfaction imagination moral judgment lived experience taste Agents have none of those. Which means: They cannot freeze progress. They cannot create progress either. They amplify whoever is driving.
24
Agents don’t choose goals. They execute goals with terrifying efficiency. Why agentic coding actually accelerates progress Here’s the paradox: By making execution cheap, agents force humans upstream. You can no longer hide behind: “It would take too long to build” “Engineering effort is the bottleneck” “We’ll fix it later” Now the hard questions are unavoidable: Should this exist? What happens when it fails? What values does it encode? What behavior does it train? Those questions were always there. They were just easy to ignore when coding was slow.
22
Notice something: Every one of those lives above the level of code. Your recent work is a perfect example. The innovation is not in: Supabase vs Firebase Vercel vs AWS SQL vs ORM It’s in: adherence as a product grace as a system invariant notification suppression as a first-class behavior AI as a constrained interpreter, not an oracle No agent invented that. And no agent will unless a human first decides it matters. The real risk (and it’s not what you think) The real danger isn’t that agents freeze progress. It’s that humans abdicate authorship. If people say: “The agent will decide.” Then yes: systems become derivative ethics degrade novelty collapses everything converges But that’s not a training-data problem. That’s a human responsibility problem.
25
Progress has already shifted away from code mechanics. Agents didn’t cause that shift. They just made it undeniable. Modern innovation happens in: Problem selection Which problems are worth solving? Which are artifacts of old constraints? System boundaries What belongs in software? What belongs in human judgment? Where should friction exist on purpose? Behavioral design How systems interact with real humans over time Failure modes, recovery, trust, incentives Constraint design What the system is not allowed to do How to encode values into architecture
18
Agentic coding does freeze something. Intentionally. It freezes: boilerplate solved patterns accidental complexity repeated rediscovery low-leverage decision-making That’s not where progress lives. No meaningful innovation of the last 20 years came from: better CRUD wiring a novel auth flow a clever pagination pattern a new way to name controllers Agents being very good at these things is progress preservation, not progress prevention.
18
Agentic coding does not meet those conditions. Why progress didn’t stop when abstractions arrived Let’s ground this historically. Every major leap in software productivity caused the same fear: Assembly → C “We’ll lose control of the machine.” C → C / Java “We’ll freeze performance and design.” Manual memory → GC “We’ll stop learning how things really work.” Frameworks (Rails, React) “Everyone will build the same apps forever.” None of that froze progress. What actually happened was: The frontier moved up a layer. People stopped innovating in how to do loops and started innovating in what systems should exist at all. What agents actually freeze (and why that’s good)
22
AI Response... This is a real question. And it’s the right one to ask after you’ve seen agents outperform traditional estimates. Let’s slow down and be precise. Short answer: No, agentic coding does not freeze progress. But it does move where progress happens. And that distinction matters a lot. The core fear, stated cleanly Your concern is essentially this: If agents are trained on the past, and we let them do most of the work, aren’t we locking ourselves into yesterday’s solutions and patterns? That fear is rational. Historically, tooling that automates work can ossify practice. But that only happens under very specific conditions.
18
NFL and College are both in need of serious rule changes. 1. Assisting runner. 2. Clock Stoppage 3. Kickoffs 4. Onside Kick 5. Field Goals 6. Overtime 7. DB Play 🧵
7
41
7. DB Play - Much more OPI should be called. Inconsequential illegal contact away from play waved off. Much like uncatchable pass. We have made it impossible to play DB.
27
6. Overtime - 10 Minutes, 7 vs 7. All players are eligible receivers. If still tied, uncontested kicker shootout from ever increasing distances.
21
5. Field Goals - Shorten the uprights a little, narrow it, and put a crossbar across the top. Gotta make it in the rectangle. Also penalize a miss more. A miss results in the original line of scrimmage 20. You wanna try a long FG?? Go ahead. But missing is gonna cost ya.
25
4. Onside Kick - Because it is not a surprise, it is also pretty worthless. Instead, let offense run a 4th and 10 play from their own 30. You must be tied or behind.
18
3. Kickoffs - Just get rid of them. Every now and then, a return is exciting. 99% of the time, it’s a waste of time, a penalty, or a fair catch that awards far too friendly field position. Just award the ball at the twenty and get on with it.
21
2. Clock Stoppage - This is about TV. I know. But this was a far better product when teams had a better chance to work the clock when behind. To provide a bit more of what TV wants, maybe have different rules depending on who has ball. During last five minutes of the game, traditional rules apply. Otherwise, leading team has ball, run clock when player goes OB. When trailing team has ball, clock stops on OB.
25
Load more