Rob Hunter
LightNews LightNews
banner
robbobobbo.com
Rob Hunter
@robbobobbo.com
210 followers 110 following 120 posts
I work with authors to build AI practice tools for their users.
https://downpat.substack.com
Posts Replies Media Videos
robbobobbo.com
I get paid to work with computers (admittedly not PCs), so I'm smaht but there's no way I would have done this on my own. I've never touched WinDbg before, and wouldn't have taken the hours to figure it out. For the right problem space, these tools are magic.
December 15, 2025 at 6:51 PM
robbobobbo.com
It pointed me to crash dumps, appeared to write parsers for them, walked me through WinDbg Preview steps to narrow things down. When it was time to take action, it gave instructions on disabling the offending module, and answered questions about the performance risks.
ChatGPT: What this dump confirms It’s UNEXPECTED_KERNEL_MODE_TRAP (0x7F) with trap #8 = Double Fault. The bugcheck parameters in the dump header are: Param1: 0x8 (double fault) Param2: 0xffffe680acdfae70 Param3: 0xffffa00ae29e9000 Param4: 0xfffff807b8152b08 That lines up with the “0x7F / param1=0x8” signature we saw in your logs: a true kernel-level CPU exception, not a normal Chrome/game user-mode crash. There's some thinking output on the right, showing a random sample of the steps it took: Extracting bugcheck parameters from the file I’m thinking we might be able to extract bugcheck parameters by locating patterns from the event log. The user is asking for file details, though, so it makes sense to focus on finding the bugcheck code (0x7f) and process name. I'll search for the key 64-bit values from the logs in the file. Let’s give this a try! Locating bugcheck parameters in file data vals=[0x8, 0xfffff807b810c8e0, 0xffffe680acf3a1c0, 0xffffa00ae29e8d00] hex(vals[1]) '0xfffff807b810c8e0' Searching for bugcheck parameters in file
December 15, 2025 at 6:51 PM
robbobobbo.com
But I didn't want to spend hours reloading the machine only for it to be hardware, and Windows machines must keep crash logs, so I went to GPT 5.2 and left it on Auto. It took n messages, an hour, and a couple hundred megs of debug dumps but it found the culprit - some audio enhancement software.
The top suspects from your loaded driver list 1) msisadrv.sys (very high suspicion) You have: msisadrv msisadrv.sys loaded ChatGPT: This is a well-known low-level MSI driver commonly installed by MSI Afterburner / Mystic Light / Dragon Center / other MSI tuning/monitoring tools. These “hardware poke” drivers are frequent BSOD contributors because they touch chipset / SMBus / sensors in kernel mode.
December 15, 2025 at 6:51 PM
robbobobbo.com
My PC's been reliable for 5 years. Crash here and again, but it crashed 3 times this weekend over about an hour of use. The third time was during light Chrome use. I only know one way to fix such a thing - clean install, and hope it's not hardware going bad.
December 15, 2025 at 6:51 PM
robbobobbo.com
I know there's a lot to be outraged by right now, but "you must download an uninstaller for Adobe Creative Cloud" deserves some too.
October 21, 2025 at 8:36 PM
robbobobbo.com
And OpenAI's Moderation API: platform.openai.com/docs/guides/...
August 26, 2025 at 7:05 PM
robbobobbo.com
It seems obvious to me that if those rails get touched, the correct behavior is to present resources and end the conversation. Apparently that was the initial behavior, and OpenAI changed it:
August 26, 2025 at 7:00 PM
robbobobbo.com
xkcd.com/463/
Xkcd comic featuring two stick figures talking with the following text: Premier ejection summiting has blamed Ohio voting machine errors on problems with the machines McAfee antivirus software. "Wait, antivirus software? On voting machines? You're doing it wrong." "Why? Security is good, right?" "Imagine you're at a parent teacher conference and the teacher sites you that he was always wears a condom while teaching. " "Ah, strictly speaking it's better than the alternative..." "Yet clearly someone is doing their job horribly wrong."
May 23, 2025 at 12:45 AM
robbobobbo.com
This is worth a read, especially if (like me) you're pretty pro-AI. From my friend from Google, who's taken the last year+ to do kickass stuff like start a print zine, write indie printer software, and take art courses.

vrklovespaper.substack.com/p/how-to-dra...
But right now, exploring AI is my last priority. So long as it continues to have all the issues that it has, it will continue to be of no interest to me. But also, it's just not what's exciting me. I look at the art community that has so generously welcomed me to create with them, and I feel infinitely inspired. What might we learn from their wisdom? So much of tech – generative AI included – has been built under the assumption that efficiency is king, and as a technologist, I feel exhilarated by the opportunities that await us if we reexamine all of tech with a radically different hypothesis: What if efficient is not always best? What if speed sometimes deprives us of experiences rather than saves us time? What if we prioritized just about anything else?
March 4, 2025 at 5:03 PM
robbobobbo.com
xkcd.com/937/
Xkcd comic on a tornado warning app. There are several 5 star reviews praising how well the app runs, and one 1 star review saying the app didn't warn them off a tornado. Average 4 stars.
December 13, 2024 at 10:00 PM
robbobobbo.com
I'm not sure that you were wrong
Tweet from Josh Allen Stats: Looks like Amari Cooper didn’t know the audible when Josh Allen checked the play. He looks to Keon Coleman who either tells him what to do or he makes it up 😂 TOUCHDOWN!
October 20, 2024 at 7:41 PM
robbobobbo.com
Pretty incredible use of LLMs
“In late 2022, Santa Clara piloted a manual review process with two staff members manually reviewing 89,000 pages of deeds, finding roughly 400 racial covenants,” the paper reads. “At an average of a minute per page, performing a manual review of the entire collection of 84 million pages of records would require approximately 1.4 million staff-hours, amounting to nearly 160 years of continuous work for a single individual.” California law still requires the county clerk’s office to manually validate each racial covenant the model found before recording the redactions, but the researchers estimated the model saved the local government more than 86,500 hours of work and crushed the project’s cost down to just a few hundred dollars to cover computing costs, a miniscule fraction of what a vendor would likely have charged.
October 20, 2024 at 7:30 PM
robbobobbo.com
"How's this different from a person reading a book in a library" is fun question, and I don't even know enough about copyright to set aside legal merits. It feels like how they got the book or article matters - even buying a book and scanning it feels better than downloading a shady PDF.
For now, case law in the United States appears to favor AI companies. Although the Supreme Court's recent decision in Warhol v. Goldsmith narrows the scope of fair use, some legal scholars argue that the use of copyrighted texts to build computer systems remains protected, even if the texts were illicitly obtained (Henderson et al.; Samuelson et al.). Verbatim reproduction of copyrighted works, as alleged in the New York Times suit against OpenAI, does seem illegal, especially if it reduces the market for such works. (Given that LLMs are not designed or used to store individual texts, verbatim reproduction will likely remain marginal.) By contrast, stylistic imitation of human artists by AI systems, as by humans, appears permissible; the dismissal of most claims in the Andersen v. Stability class action suggests as much. Most troubling of all are the claims, still pending in Andersen and elsewhere, that AI firms break the law by using copyrighted works to train their models. Setting aside the legal merits, how do such claims differ in principle from the idea that a person must not enter a library, read a few books, and leave with the knowledge they contain? The copyright police will rightly say that language models are not people, but both operate within networks of language technologies that affect what we can read and write. Both generate complex but predictable language based on texts we have previously read.
October 13, 2024 at 12:09 AM
robbobobbo.com
We hired a writer rather than a DS or engineer to help write prompts and fix weird outcomes, so I totally agree that opportunities are being created. It's way less obvious to me that people who got into the humanities will enjoy this work.
For literary criticism, the rise of LLMs should further motivate the already decades-old return to aesthetics, including judgments of value. It has become easy to generate coherent texts of any length in any familiar genre. With the issue of quantity rendered moot, that of quality remains. Kirschenbaum and Raley ask, “Are we sure we know what ‘good’ output actually is?” This question has preoccupied aesthetic criticism for centuries, and it should not remain so marginal to literary studies today. Having earlier dismissed my poetry friends for saying that machine-written poems lack a certain something, I concede that we should join them in trying to say what makes good art good, but we should seek answers that avoid politically backward ideas about the human, the creative, the authentic. However we approach questions of aesthetic value, we should abandon the belief that aesthetics inherently provides cover for elitism, quietism, or conservatism and recognize it as a politically enabling field.
October 13, 2024 at 12:02 AM
robbobobbo.com
www.smbc-comics.com/comic/2010-1...
September 27, 2024 at 2:28 AM
robbobobbo.com
When I left Uber, I managed 3 people in SF, 3 in Seattle, 3 in Toronto, and 3 fully remote. I can imagine having some teams in person and some remote, but forcing everyone to the office to be on Zooms all day is the nonsensical norm.
Others still noted that it seemed pointless to go in to an in-person office when the teams they worked on were already distributed across multiple offices around the world, so they'd mostly still be on Zoom calls anyway.
June 21, 2024 at 2:32 PM
robbobobbo.com
Not great
Screenshot of the linked WIRED article: That secret IP address-44.221.181.252-has hit properties at Condé Nast, the media company that owns WIRED, at least 822 times in the last three months. One senior engineer at Condé Nast, who asked not to be named because he wants to "stay out of it," calls this a "massive undercount" because the company only retains a fraction of its network logs. WIRED verified that the IP address in question is almost certainly linked to Perplexity by creating a nev website and monitoring its server logs. Immediately after a WIRED reporter prompted the Perplexity chatbot to summarize the website's content, the server logged that the IP address visited the site, This same IP address was first observed by Knight during a similar test.
June 19, 2024 at 5:59 PM
robbobobbo.com
I'd recognize that alligator car anywhere
A bunny driving an alligator car in Richard Scarry's Cars and Trucks and Things That Go
June 7, 2024 at 7:21 PM
robbobobbo.com
Occasionally we get as much joy from the BarkBox as Steven. Drool Intentions from Barkbuster.
Steven racking up late fees with his Barkbuster chew toy. Back of the Barkbuster toy, describing Drool Intentions: Three pure bred pups get more than they trained for in this unneutered thriller. Quentin and Eloise are two rival litter mates who treat their pupper crust obedience school like a dog park of debauchery. When new puppy Dozer shows up, leashes get tangled in a dangerous game is cat and mouse and dog and dog and dog. Barkbuster dog chew toy, a parody on Blockbuster video rental cassettes
May 17, 2024 at 12:51 AM
robbobobbo.com
Slack autocompleted ":+1 <tab>" to 👎 instead of 👍...

If you've had a coworker thumbs down your suggestion today, you may be entitled to compensation
Slack's autocompletion menu suggesting :-1: before :+1: with the input ":+1"
April 19, 2024 at 10:27 PM