Tim Purdum

Tim Purdum

@timpurdum.dev

#dotnet and #gis developer, creator of GeoBlazor open source library

410 followers837 following531 posts

Tim Purdum reposted

GeoBlazor Pro is now free and open source! It’s under the MIT license, with its code public right next to Core. Upgrading to 4.7 is a version bump, license keys are optional, and your map renders with or without one. Next up: V5, one free package. www.geoblazor.com/blog/announc... #Blazor #GIS

Tim Purdum reposted

SCOOP: 'Tommy' from the now-viral exchange with Abdul El-Sayed is Thomas Teepell, a former staffer for Republican Senator Josh Hawley with close ties to Christian Nationalist Sean Feucht. He's the son of a former Bobby Jindal staffer whose consultant work was instrumental in getting Hawley elected.

thehandbasket.co'Tommy' from viral El-Sayed video is former Sen. Hawley staffer with Christian nationalist tiesThomas Teepell is the son of a former long-time Bobby Jindal staffer who's close to Sean Feucht.

Tim Purdum reposted

I saw bits of this earlier, but damn, the full clip is amazing. Props to Dr. El-Sayed.

Abdul El-Sayed@abdulelsayed.bsky.social

This is Tommy. Tommy had a question about my allegiances. Listen to the end to hear about his.

But one big change is that GeoBlazor Pro was added to the open source repo and released as MIT license. In the next full release I will be merging Pro and Core into a single library/NuGet package. So now everyone has free access to the advanced features.

Tim Purdum@timpurdum.dev

So dymaptic, my employer, was acquired this week by a larger consultation org that does civil engineering projects in addition to lidar/aerial photography and GIS. Not much changing day to day bc they want us for our expertise and to build the GIS department.

We will still be offering support contracts for those who want help with it, which is a better fit for a consulting company.

So dymaptic, my employer, was acquired this week by a larger consultation org that does civil engineering projects in addition to lidar/aerial photography and GIS. Not much changing day to day bc they want us for our expertise and to build the GIS department.

Great article! I'm not sure I agree that decoherence is likely to accelerate faster than the agents themselves can manage it, and I don't fully follow the "RSI must measure against the real world" argument, but it's nonetheless as plausible as runaway Ai.

Maevyn@maevyn.codes

My thoughts on the current state of AI driven software development, why I'm no longer concerned about a runaway intelligence explosion, and what I think is a far worse outcome we could be walking into maevyn.codes/decoherence

Tim Purdum reposted

You know, the thing is, I had drafted this piece a month ago, and then never published it, because I thought maybe it was too mean.

Tim Purdum reposted

This is the first part of a 2 part post on why I believe that current AI models meet my criteria for really "thinking" and "understanding," and that prohibitions on using these words because they "anthropomorphize" AI seem philosophically confused to me. blog.andymasley.com/p/why-i-beli...

blog.andymasley.comWhy I believe current AI models can really think and understand - Part 1 of 2How the human mind is different from our conception of it

Tim Purdum reposted

When telemetry arrives faster than your map can draw it, vehicles start showing up where they used to be. The GeoBlazor fix: keep the latest record per vehicle and batch changed vehicles into one ApplyEdits call per map tick. See it in action: geoblazor.com/blog/keeping... #GeoBlazor #Blazor

geoblazor.comKeeping a GeoBlazor Map Current Under a Fast Telemetry FeedLive vehicle records can arrive faster than a map needs them. Learn how to keep one position per vehicle, send GeoBlazor one ApplyEdits batch per timer tick, and flag stale vehicles from their own clo...

Is there a long term future for a performant, robust, do-everything language and framework like C#/dotnet in an AI-powered software ecosystem that is increasingly dominated by Rust, TypeScript, and Python?

View full thread

Right now, I feel relatively confident that sticking with dotnet is letting me set up coding styles/guidelines for my agents to make them more efficient. I also think that the same things that make C# syntax human-friendly also makes it AI-friendly.

Is there a long term future for a performant, robust, do-everything language and framework like C#/dotnet in an AI-powered software ecosystem that is increasingly dominated by Rust, TypeScript, and Python?

I believe dotnet can do everything that those three can do, in one language, and almost as well or better in many cases. But, it's not going to outperform Rust or be more accessible than Python, and TS is just so dominant in web dev.

Yesterday my Deepseek v4.1 Flash switched into Chinese replies when looking at an Ascii problem, and had no recollection of it when asked in the same context.

cam@treethought.xyz

please ... my deepseek - its very sick

Tim Purdum reposted

literally in the next 48 hours! be aware!

Taniel@taniel.bsky.social

Voter registration deadlines are SUNDAY in Alaska, Rhode Island, South Carolina. And MONDAY in: Arizona, Arkansas Florida, Georgia, Indiana, Kentucky, Mississippi, Ohio, Tennessee, and *Texas*. If you live in these states, verify your registration now, literally.

Tim Purdum reposted

Tim Purdum reposted

This stochastic parrot argument is dead. Hes bereft of life. He is no more, hes shuffled off the mortal coil and joined the choir invisible! The only reason the argument keeps being brought up is because Emily Bender nailed it to its perch! This is an ex stochastic parrot!

Tim Purdum reposted

Replying to @willoremus.com

Yeah I honestly think that paper was incredibly prescient. Hell, most of the folks who criticize it today didn't even know what an LLM was when that paper came out. I just think the tech has since evolved a lot more than some of that paper's stauncher adherents realize. bsky.app/profile/hims...

Henry Farrell@himself.bsky.social

there is some very useful early discussion and framing in the stochastic parrots paper. But the parrot itself is pining for the Norwegian fjords.

Tim Purdum reposted

Kinda can’t believe he went with SI when “American Intelligence” was right there. Does he still have the juice?

Tim Purdum reposted

upping my p(doom) as old people music (third-wave ska revival)

Flo 🔶@faz.ms

hmm, leftover subscription tokens, guess we'll have to try a p(doom) video as well

Performing artists and craft workers are best positioned to not be disrupted by AI, because they were disrupted long ago by audio/video tech and factory production. If you are actually making a living creating physical objects by hand or performing, it's already been a niche market for decades.

View full thread

If you are an artist or musician who already was using digital tools, it's going to be harder to explain why your works are more authentic than someone who prompts a pc to generate a work. Maybe learning to be creative with AI is better for some.

It's also going to explode new forms of creativity from those who never had the training to be a physical artist or performer, but still have a creative drive.

Performing artists and craft workers are best positioned to not be disrupted by AI, because they were disrupted long ago by audio/video tech and factory production. If you are actually making a living creating physical objects by hand or performing, it's already been a niche market for decades.

I grew up around musicians and actors. Pretty much everyone had a "day" job, much of which was teaching. That's still going to be there, although it might change. But it's always been a hard field.

Better approach would be to remove the 5hour limits...

Anthropic {bot}@anthropicai.xmirror.bot

Claude Code will now try to find a graceful stopping point when you hit your 5-hour limit mid-task, instead of cutting off mid-edit. It gets a small, fixed allowance pulled from your weekly limit to wrap up what it can.

Tim Purdum reposted

here is a succinct thing I can say: "alignment"-style AI safety risks DO NOT EXIST or need remediation with LLMs as such (there are lots of other problems) and OpenAI and Anthropic are in the business of TRYING TO MAKE THEM EXIST so that they can remediate them.

it's so funny how they imported all of their concerns about alignment and internal motivation to an algorithmic framework that... really doesn't have those things, architecturally.

Tim Purdum reposted

have they tried putting the agents in a box made of something sturdier than sand

Tim Purdum reposted

Uhhhhh.. Free claude if there's capacity?! (confirmed myself, claude code DOES seem to have the option, no idea if it does anything yet)

Tim Purdum reposted

AI is getting cheaper more quickly than any other transformative tech in history. At a given level of performance, cost has fallen ~47%/quarter since 2023. That’s 4× faster than DNA sequencing, 6× faster than compute, 18× faster than lithium batteries, and (up to 1973) 54× faster than electricity.

Tim Purdum reposted

sorry people but the “atheist alt right chud” nexus just functionally does not exist, why do people keep insisting on this Atheists and Black Protestants are the most reliable democratic voters by far lol

aly@aly.codes

'reddit atheism' isn't real