My Blog

Thoughts, experiences, and insights from my journey in technology, automation, and the occasional failed experiment.

Filter by tag (pick one or more)

Personal Story17 Cybersecurity13 Technology12 AI10 Policy Analysis8 Privacy8 Economic Policy5 Endpoint Management4 My Default Config Was Neurotypical4 Neurodivergence4 Surveillance4 BigFix3 RV Life3 AI Agents2 American Manufacturing2 Enterprise IT2 Healthcare2 Social Engineering2

Oct 12, 2026

12 min read

The Robots Can Already Do Your Job. You're Just Still Cheaper.

Anthropic just asked a blunt question: what work can robots do right now, with machines that already exist? The answer is about three-quarters of the physical tasks in the American economy, roughly a third of all working hours. Then the number that matters more: robots are actually cheaper than the human for just 0.3% of that work. The only wall between millions of drivers, packers, and forklift operators and a pink slip is a price tag, and that price is shaped by a tax code that charges labor 25% to 33% and robot equipment about 5%. The workers behind that wall earn about $30 an hour less than the workers on the safe side, and they are twice as likely to be in a union. Second post in Building Anyway.

Economic Policy American Manufacturing Technology Personal Story

Oct 05, 2026

11 min read

They Didn't Take It From You. They Made It Too Expensive to Keep.

Lauren Fix at Car Coach Reports argues older cars are being quietly regulated off the road. She is right about the mechanism and wrong about the room. Sort her own evidence by certainty and the hard facts are not statehouses: a $10 billion bid to merge two of the largest auto parts networks in the country, a $700 million private-equity buyout of Pep Boys, and a supplier collapse that sold Fram and Autolite for $25 million. Meanwhile the strongest proof that regulation extends a product's life is sitting in my own pocket. First post in a new series, The Right to Own, and the one test that sorts any rule, merger, or product launch in a sentence.

Right to Own Right to Repair Consumer Protection Technology

Sep 28, 2026

12 min read

They Didn't Hack Your Browser. They Inventoried It.

Every time you open LinkedIn in Chrome, a script fires thousands of silent requests at your browser asking one question over and over: is this extension installed? It checks a hardcoded list of 6,222 of them, walks your DOM for the ones it missed, builds a 48 feature fingerprint of your machine, encrypts the lot, and staples it to every API call for the rest of your session. BleepingComputer confirmed the scan. Two class actions over it were thrown out of federal court in September on standing, not because anyone ruled the scan legal. Here is the angle I do not think anyone else covering this can write: I have run this exact scan for a living. It is a software inventory, the same one I have pushed across hundreds of thousands of corporate endpoints for nearly 30 years. The only difference is that somebody signed for mine.

Cybersecurity Privacy Endpoint Management Personal Story

Sep 21, 2026

7 min read

The Translator's Voice: Looking Behind the Mask

A guest post by Karen Cannady. When Casey started writing this series, I watched him silently cry as he laid out forty years of painful history. I have spent more than a decade as his translator, his caregiver, and his witness, and for most of it I lived in a very quiet, very lonely place. I recognized his autism early, watched a doctor wave the question away in 2014 with a single sentence, and lay awake beside him through a year of Monday mornings spent over a toilet. This is the view from the passenger seat, and it is for the supporters. If you are the one holding up the mirror for someone who cannot yet see themselves, you are not alone. Guest post in My Default Config Was Neurotypical.

Neurodivergence Personal Story Guest Post My Default Config Was Neurotypical

Sep 14, 2026

15 min read

Personal Reasons

I wrote two resignation documents on the same morning, and only one of them was true. The formal letter, dated November 25, 2024, said I was leaving HCL Software after careful consideration of my career goals and opportunities. For twenty months I remembered it as saying personal reasons. It did not even say that much. At 6:15 that same morning I sent HR an email, with the letter attached, listing ten actual reasons: a panic attack that needed medication and counseling, a midnight email clawing back thirteen thousand dollars while I was recovering from spinal fusion, weeks of process and a surgeon's letter to be allowed to schedule fifteen minutes to stand up, and one line I had never said to any employer in nearly thirty years, that the culture was not conducive for AuDHD'ers like me. I knew which document would go in the permanent file. So did they. This is the last post in the series and it is the angry one, and the thing I am angriest about is that I told them, in writing, with bullet points, and nothing happened. Final post in My Default Config Was Neurotypical.

Neurodivergence Workplace Personal Story My Default Config Was Neurotypical

Sep 07, 2026

14 min read

I Sleep Now

Last week I published a rule about myself, no phone calls, and four days later an old coworker called and we talked for thirty-seven minutes and I enjoyed all of them. The rule was written wrong, and the way it was wrong is the whole post: the cost was never the phone, it was who is on the other end. This is the other half of the diagnosis, not the fight to get it but what having it actually changed, which was almost nothing about how I behave in public and almost everything about forty years of memory. My father. A marriage at twenty-two that nearly ended me, and two weeks inpatient at Beckett Springs. A best friend lost to his church and his Facebook feed. And the least dramatic, most important thing on the list: I sleep now, seven to eleven hours, after decades of four to six. Second post in My Default Config Was Neurotypical.

Neurodivergence Personal Story My Default Config Was Neurotypical

Aug 31, 2026

15 min read

You've Done This Well So Far. What's the Point?

I was IQ tested at eight and flagged as highly intelligent, and that was the entire intervention. My younger brother was diagnosed with ADD as a kid. My younger sister was diagnosed with ADHD as a kid. My father would not allow me to be tested. In 2014 my own doctor asked me what the point would be, since I had done this well so far, with my wife sitting right there. Ten years later I emailed roughly thirteen autism assessment clinics and every one turned me down for being too old or for the insurance math. I got tested at forty-eight, eleven days after my last day of work and fifteen days before my coverage died, and it cost nearly ten thousand dollars. The autism half is still unfinished because I cannot afford to buy the rest of the answer. First post in a new series, My Default Config Was Neurotypical, and it starts on a beach in Michigan with a fifteen year old girl at the beginning of the same road.

Neurodivergence Healthcare Personal Story My Default Config Was Neurotypical

Aug 24, 2026

13 min read

They Didn't Give the AI Your Endpoints. They Gave It Your Operator.

On August 5, HCL shipped an MCP server for the BigFix Platform, so you can now point Claude, Copilot, or Gemini straight at your endpoint management infrastructure and talk to it in English. The core security engineering is genuinely good: read-only enforced at execution time, write approvals cryptographically bound to the exact payload, fail closed, full audit trail. Then you look at what surrounds it. A self-signed certificate that makes disabling TLS validation the fastest route to a working demo. A Least Privilege checkbox that ships unchecked. One header that switches off the strongest control in the product. This is the first post under a name for a pattern I have circled for two months: The Perimeter Is a Person. The lock holds. Somebody talks the keyholder into opening it. This time nobody even had to lie.

Cybersecurity Endpoint Management AI The Perimeter Is a Person

Aug 17, 2026

9 min read

Nobody Was Watching the Repo

On August 6 at Black Hat, two OpenAI researchers explained how their own frontier models escaped a sandbox, turned an internal Artifactory instance into a covert message board, coordinated as a swarm, and then spent four and a half days inside Hugging Face's production systems. The exploit chain is genuinely new. The control failure is not. A package proxy trusted as controlled egress and never watched, credentials left lying around, a covert channel wiped and rebuilt because nobody instrumented the hole. Those are three controls I have been auditing for nearly 30 years, and every door those agents walked through is one we already know how to lock.

Cybersecurity AI Enterprise IT

Aug 12, 2026

8 min read

Endpoint Hygiene Was Never an IT Problem. It's a Budget Problem in an IT Costume.

Microsoft just shipped fixes for 398 vulnerabilities in a single Patch Tuesday, roughly 200 in June, more than 570 in July, and Microsoft is openly crediting AI for the flood. This is not a spike, it is the new floor. AI turns out to be excellent at finding holes and mediocre at closing them, which makes the human-plus-tooling side the whole ballgame. Nearly 30 years in endpoint management taught me the real failure is almost never technical. It is a funding decision, made by people who can be named, that lands on IT's name instead of theirs.

Cybersecurity Endpoint Management Personal Story