Dave

Dave

@dps952.bsky.social

Seattle patent nerd hockey dad.

155 followers289 following1.3K posts

A radical idea for the patent folks: A new Beauregard-like claim preamble: “A method, comprising: submitting a prompt to a genAI that causes the genAI to output computer-executable instructions that [regular Beauregard clam language]” An additional way to combat AI-based reverse engineering?

Dave reposted

Thrilled to share that The Terms & Conditions of Generative AI is forthcoming in the Texas IP Law Journal!! 📣🎉 W the inimitable @johnmarknewman, we study the boilerplate for 86 genAI tools: what's good, not so good, & actually quite bad. 🧵👇 buff.ly/3tMJbG0

buff.lyTHE TERMS AND CONDITIONS OF GENERATIVE AI<div> Generative AI raises pressing questions for myriad doctrinal areas. Policymakers and scholars have responded rapidly, devoting substantial attention to t

A question for the client-responsible lawyers out there: Say a new matter comes in to your firm, and you give out matter numbers sequentially. The next matter number happens to be “1488”.

Do you consider changing it? Or do you just let it be, with the assumption that no one either in your firm or the client spends as much time on bsky as you do, and would have no idea why you would want to change it?

I am reasonably amused that I was able to guess who this was in 1 try.

Mike Masnick@masnick.com

I was at a conference last week where a well known 1st Amendment professor claimed it was silly to think that free speech was under attack in the US because people at that conference were criticizing the administration, so clearly everyone felt free to speak.

Dave reposted

I am going to a meeting with the Pope and plan to be surprised when the Pope argues for the unique dignity of the human soul. I am a smart person.

Parkscope Joe@parkscopejoe.bsky.social

This is insane.

Dave reposted

Hey, Seattle folks! I will be at the Northgate Barnes & Noble this Friday from 6pm until I'm not! This is a ticketed event, and I will have pins and possibly stickers as a thank you gift for folks who pick up a book and come out to see us! www.eventbrite.com/e/seanan-mcg...

eventbrite.comSeanan McGuire celebrates A DIVIDED DUTY at B&N NorthgateB&N Northgate in welcomes Seanan McGuire to sign and discuss her new book A DIVIDED DUTY on Friday, October 2nd, at 6PM PST!

Hey, fellow #patent nerds. Anybody else notice that 425 patent applications entitled “SYSTEM” filed by SoftBank were published last week? That’s… weird?

Walking distance from Pinehurst station to Shoreline station: 18 mins. Walking distance from Pinehurst station to Roosevelt station: 29 mins. Walking distance from my house west of Aurora to closest light rail station: still 62 minutes. A great victory for transit access!

Mayor Katie B. Wilson@mayor.seattle.gov

Seattle is a growing city, and we can’t just keep adding more cars to the road. That’s why I was so excited to welcome the newest addition to the system this week, the Pinehurst station in North Seattle.

Dave reposted

The Star Spangled Banner being played by the Mariachi Herencia de Mexico band before the Bears-Eagles game in Chicago on Monday Night Football was a deeply cool thing to see tonight

Can we point out for one second that the Bears’ third-string QB was out there zipping balls into tight windows and making completions while backpedaling in a way their QB1 has only been capable of a handful of times in the last 50 years???

You know a guy actually spends time with his kids when he says, “I’ll try to watch the game with them tomorrow, but we’ll probably end up watching Bluey.”

Da Bears@chicagobearsx.bsky.social

Case Keenum on his primetime performance 🐻

Is it just me, or is having a historic year-over-year performance drop-off and then saying, “well, we all did bad” not a great look?

Seattle Sports@seattlesports710.bsky.social

After the #Mariners were eliminated from playoff contention, Cal Raleigh addressed what led to the club's disappointing season. “I think it was a collective failure." sports.mynorthwest.com/mlb/seattle-...

A fascinating story! Please don’t let US patent examiners find out about this one, or that’s the end of our “impractical to perform in the human mind or with pen and paper” responses to mental process rejections.

Bart⚓️@bartgonnissen.bsky.social

1/x In the 1840s, a lame American naval officer, grounded at a desk in Washington, cut weeks off the world's ocean voyages without inventing a single machine. His raw material: a warehouse of logbooks nobody had ever read. Matthew Fontaine Maury, Pathfinder of the Seas. A 🧵 #maritimehistory

I don’t care for this latest season of Parks and Rec.

Aaron Rupar@atrupar.com

PATEL: The previous regulation automatically disqualified any individual who participated in either side of it KENNEDY: Either side of beastiality? So you disqualified the animal? P: Well, we have great canines, but we're not going to disqualify them K: I'm trying to understand why you did this.

My 12yo was telling me about how they were talking about 9/11 in class yesterday, and a classmate said, “I hear Trump claims he was carried away from Ground Zero by two firefighters. That’s pretty good for him, normally Trump would say he was the one carrying the firefighters.”

Just needed to share that one, since it’s better than any reaction I read on bsky. Middle schoolers remain undefeated in talking smack.

Me, sitting in the basement of my Seattle house watching the Apple Cup at Husky Stadium across town on TV: “wow, it really looks nice out there.”

Looking forward to the briefs on the eventual copyright infringement suits here that all cite the Betamax case *against* Sony.

Dexerto@dexerto.bsky.social

Sony has responded to a digital games lawsuit by arguing that players don't actually own the games they buy Its lawyers say they are licensed rather than sold

This is a good thread about human perception of language, but it also suggests the right way to approach LLM usage:

Emily M. Bender@emilymbender.bsky.social

This sign is a great example for helping to explain why chatbots present such a potent illusion. A probably not short thread 🧵>>

View full thread

When it says "there are two 'r's in strawberry," what it's really saying is "when someone has asked a question in the past about how many letters there are in a given word, it's statistically likely that the answer that was provided was 'two.'"

If your mental model of the LLM includes that aspect of its operation, then you can reasonably evaluate the outputs for quality, and you can ask it questions appropriate for its capabilities. If your mental model of the LLM is, "it counted the letters," you can't.

This is a good thread about human perception of language, but it also suggests the right way to approach LLM usage:

Emily M. Bender@emilymbender.bsky.social

This sign is a great example for helping to explain why chatbots present such a potent illusion. A probably not short thread 🧵>>

View full thread

So, I guess my point is: responsible LLM use requires building an accurate "mind model" of the LLM, so that you can recognize its capabilities and its limitations, and so that you can understand what it's really saying.

It's like how people make fun of LLMs for not knowing how many "r"s there are in "strawberry": if you understand that the LLM doesn't have a letter-by-letter representation of the words in the prompt, you understand that any answer it provides to that question can't be relied upon.

This is a good thread about human perception of language, but it also suggests the right way to approach LLM usage:

Emily M. Bender@emilymbender.bsky.social

This sign is a great example for helping to explain why chatbots present such a potent illusion. A probably not short thread 🧵>>

View full thread

...it will provide a reasonable answer if I ask it for a strategy for replying to a subject matter eligibility rejection in a new application, because the language data it's drawing its statistical inferences from is a good representation of the problem being presented.

Meanwhile, if I know that it's trained only on patent application prosecution history, I would be much less confident that it would give me a good answer about whether an issued claim is patent eligible, because that data is less representative of how a court would decide.

This is a good thread about human perception of language, but it also suggests the right way to approach LLM usage:

Emily M. Bender@emilymbender.bsky.social

This sign is a great example for helping to explain why chatbots present such a potent illusion. A probably not short thread 🧵>>

View full thread

Bringing this back to LLMs: I used to push hard on the idea that you shouldn't use an LLM for substantive work (in my case, legal work) unless you know how the LLM works (at least at a basic level), you know what data it's been trained on, and you know what data it accesses.

That's the parallel to this "mind model" idea: If I know that an LLM is trained on, say, the entire corpus of patent applications and their prosecution history from the past 5 years that received a subject matter eligibility rejections, I could be more confident that...

This is a good thread about human perception of language, but it also suggests the right way to approach LLM usage:

Emily M. Bender@emilymbender.bsky.social

This sign is a great example for helping to explain why chatbots present such a potent illusion. A probably not short thread 🧵>>

A premise of the thread is that when a human reads these notes (particularly the handwritten note), you form a mental image of the state of mind and the situation of the person writing it, and that is an important part of receiving all of the meaning conveyed.

Like, this person probably hears the bell ring all the time but reacts slow, delivery drivers worry they didn't hear it so they ring it multiple times, and everyone on the other end gets annoyed. That's a lot of meaning!

If we’re not careful with generating all of this discourse about traffic cameras, we’re going to end up summoning Volokh talking about the slippery slope like he’s Beetlejuice.

Dave reposted

HALF NELSON was a 1985 Joe Pesci TV series that was cancelled after 9 episodes, but its intro seems to last forever via an escalating cavalcade of absurd cast reveals, underscored by a theme song that feels like a gag gift when you can't license "I Love L.A."...

Fascinating one, here!

Justin White@jwhite.bsky.social

I am going to try to explain what just happened in the Blue Jays/Red Sox game. Reskeet if you want to. So the Red Sox had runners on first and second, no outs. The batter hit a low liner toward 1st base which bounced once on the dirt just as Vladdy dove to stop the ball. It would have looked like...

Can’t decide whether to send this to my friend in private equity to congratulate him, or to his wife to warn her.

Ann Monster Lipton@annmlipton.bsky.social

For your consideration: “If you live in the celebrity bubble, party after party, with the same people—the same 25 people—life gets really boring,” said Kevin O’Leary, the businessman and “Shark Tank” panelist. “When you end up with a private-equity guy, your life gets far more interesting.”

Not that anyone cares, but this home run hit me in the hands and I biffed it. I’m never going to get called up to Banana Ball fandom now.

Circling Seattle Sports@circlingseasports.bsky.social

Julio Rodríguez hits a solo shot to the Home Run Porch in left field, and Seattle is within a run. The 39,468 fans here at T-Mobile Park get something to cheer about. Tampa Bay Rays 2 - 1 @mariners.com B9th, two outs #TridentsUp #SeattleMariners

Reading “this city would be worse if we didn’t get to have giant piles of trash on the curb” is like reading “the outputs of AI are stupendous and everyone should use it for everything.” The mind rebels.

This is it boys, the platonic "New York can never get better" piece www.archpaper.com/2026/07/litt...