smklein.bsky.social
I build some things, break others, and always stop to pet the dogs
Engineer living and running in SE PDX
115 posts
359 followers
103 following
Getting Started
Active Commenter
comment in response to
post
"the loss of that $1.9 million – representing roughly 21,000 hours of overtime pay"
Wild idea, but perhaps we shouldn't be using "police overtime hours" as our unit of measuring the PPB.
Any other industry using overtime to inflate wages this much would be subject to major criticism
comment in response to
post
Ah yeah I'm more concerned with "injection of ulterior motives" into training sets, rather than "did it eat enough data"
comment in response to
post
I'm hopeful about this pathway, but not totally comforted yet - deepseek can be executed on rented hardware, but the training aspect is not really controllable, right?
It's a far cry from "compile from source", and much more akin to "there exist binaries that you can sometimes use, pending laws"
comment in response to
post
Totally. Also, I think I'm understanding your "capabilities precede ethics" point as: "if a tool is sufficiently powerful, we must reckon with ethical implications of its usage".
If LLMs were still really unhelpful we wouldn't really need to talk about all this the same way
comment in response to
post
Also, @steveklabnik.com , I'll clarify something here: I 100% agree that discourse on this topic is sloppy. Me veering off into "ownership issues" doesn't discount your main point, from this post specifically - this stuff is a mess to talk about.
Appreciate you putting this out there.
comment in response to
post
Agree, though we have already surpassed "fancy autocomplete".
The paradigm shift towards working effectively with these tools is one I'm still trying to understand, but it feels shockingly like "becoming good at an internal tool at your company". Effective, but you cede control to the tool's owner.
comment in response to
post
All that said, definitely agree that the voices proclaiming "LLMs cannot do X", where "X" is programming, writing, searching, creating things, etc, are unfortunately uninformed.
They clearly have significant (and also inappropriate!) uses. But I fear consequences of dependence.
comment in response to
post
These aren't unsolvable problems, but there are major hurdles:
1. Locally-run models are inferior to the quality / experience of models owned by mega-corps.
2. **Both** local and non-local models are incredibly opaque.
I find this issue way more concerning than "how capable is the tool"
comment in response to
post
"capabilities inform what is or isn't ethical": I find this an interesting statement. Trying to digest it.
I can see a lot of the utility/risks of LLMs, but one of my biggest beefs is with ownership.
There's a massive push to build dependency on these tools that are overwhelmingly privately owned.
comment in response to
post
Also it's sketchy as hell to send "official city emails" from addresses like "em7526.invoicecloud.com" and expect people to supply their bank details.
This is begging for people to get scammed, it sucks for lots of reasons
comment in response to
post
Fascinating. I was reading up on some people's opinion on the bottle bill, and the idea that "well, you can get high on $1 of fent, so, recycling bottles as a way to get $1 is now dangerous" is such a stupid take.
www.groundscoreassociation.org/glitter is small but shockingly effective
comment in response to
post
No rush, but I'm definitely still interested in this.
Finding an easy path to reliable employment for unhoused folks, regardless of where they're coming from, seems like the sorta thing that should be universally popular (e.g., "Even a harsh critic could not call it a handout").
comment in response to
post
true. a lot of this is personal bias too because I am old and have not *so* much free time, so sometimes a "big open world" is experienced by me as "oh no I won't be able to see a lot of the cool stuff here"
But yeah especially with PL things felt pretty smooth!
comment in response to
post
If they had a really tight experience with much more conservative scope that would actually be kinda sick too
I always prefer a tighter more carefully crafted experience than something bloated and messy
comment in response to
post
Online schedule says 2025 but curious if things are on schedule
comment in response to
post
/me imagines a world where the city quietly reposts all those "Portland is burning" articles, lets the value of land drop... then swoops in and buys skyscrapers for cheap, and ends up with a great carrot for local govt and businesses to come make downtown prosper
... Sounds nice
comment in response to
post
The GLITTER program seems great - for someone interested in tracking the city's involvement and expansion, where should I be looking for updates?
(Aside, have always wondered about a model like this for other forms of labor the city needs, like road maintenance, trail work, etc)
comment in response to
post
Eh, you know what, maybe I'm wrong here. Seeing different definitions between "It's illegal" vs "it's just a fund sitting around with no explicit purpose".
I can see why this would fall into the latter category.
comment in response to
post
FWIW, I totally agree with your proposal here about where to **move** the money, I'm just trying to clarify "why does this organization have this money in the first place".
The term "slush fund" to me implies "they're using it for illegal purposes", which would be a more extreme accusation.
comment in response to
post
Specifically:
- If your budget is the same (actually slightly raised, as pointed out by @councilorgreen.bsky.social ),
- And the SIF drops from $50 million to $40 million,
How does that "amount to a blatant attempt to quietly sunset all our joint work", and "decimate" Prosper Portland's funding?
comment in response to
post
Love to see clear comms about stuff like this. Managing Portland's budget seems inevitably painful, but making the rationale for hard choices clear ("who gets what, and why") is exactly what I'd want from city council
comment in response to
post
FWIW, I think the PP response was wild, and re-allocating a portion of that fund to ensure city staff stay employed seems reasonable, however...
... Is there a reason you're calling this a "slush fund"? Aren't these funds a previous city government has allocated explicitly to economic development?
comment in response to
post
@pdxshea.bsky.social ^
comment in response to
post
- There's "Support Bundle" API for collecting diagnostic info. This is available to operators of the rack, can be audited (it's *your rack* - none of this data is "Oxide-only"), and is not "phone-home"!
Our priority is: get data transparently to operators. Make it "opt-in" when to share with Oxide.
comment in response to
post
Some extra flavor here:
- We have an alerting system that uses webhooks, and can notify "whoever you want" on notable events (e.g. hardware failures, capacity, etc). You *could* loop in Oxide, but that's not the default!
comment in response to
post
I appreciate the nuance/candor you're having here. Feels a little bit like this issue is portrayed as either "you do not care about responsible gov't spending" vs "you hate kids". There's clearly nuance here, and we can want the best for our students while trying to avoid wasteful spending.
comment in response to
post
(also that comparison to "we are spending billions to stretch a mile of freeway" hits hard, this is definitely a much better use of funds)
comment in response to
post
Thanks for this - the breakdown in costs is great, and something I'm not seeing too clearly from other reports.
I'm agreeing with your conclusion here about "there are some issues, but it's worth it to pass".
Do you have any idea why the high school costs are so high relative to surrounding areas?
comment in response to
post
Interesting - I figured there would be slightly more dissent due to the price?
That said if no one in city gov't genuinely things this can be done for cheaper I'm down to vote "yes" as well. Just hope the money gets spent well and soon.
comment in response to
post
Also: Do y'all know how "binary" this bill is? If it doesn't get passed, will the school remodels simply not happen, or will there be another attempt to create a budget for a similar effort?
(Trying to balance "This work needs to be done" vs "these would be the most expensive schools in the US")
comment in response to
post
Do you have a "happiest" and "least happy" take on the mayor's proposed budget?
comment in response to
post
OFFICIALLY REGISTERED - come run with me!
gobeyondracing.com/races/lastes...
comment in response to
post
Anyway. Im interested to keep playing with this format. I'm planning on signing up for Lastest Not Fastest in October, which should be another great opportunity to test myself. Looking forward to seeing some of you out there.
comment in response to
post
4) small mental boundaries can help push you over the edge. For example:
"I will not quit at the aid station - I will force myself to start every loop, no matter how bad I feel"
And
"I will get nutrition every lap, no matter what - even if I can only drink calories"
These things help
comment in response to
post
3) the race is SO much more mental than physical. Physically, of course, it hurts. But for this 24 hour race, NO ONE dropped out on loops 21, 22, 23, or 24. Why? Those were the hardest loops, after our legs were crushed.
It's because we wanted to finish! Shy of an injury: quitting is a choice
comment in response to
post
2) fixing problems is the ACTUAL novel challenge on the day. Can't eat after mile 60? Shoes not fitting because your feet swelled? Got a muscle cramp? You need to figure out how to mitigate issues FAST before they take you out.
Practicing the format - and your gear - helps a lot here.
comment in response to
post
1) the race necessitates going slow. Regardless of your normal mile time, you probably want to give yourself 5-10 minutes at the aid station. Less time, and you wont have time to refuel/fix problems. More, and you'll get stiff.
comment in response to
post
yeah. i bought a place in portland with my wife a few years ago, but this is also part of why I never understood NIMBY-ism. I want way more cheap housing around me. Get people (in all the ways) invested in the local community!!
comment in response to
post
That said, for people who "owned a place in PDX for a while and want out" or "just moved here, rent a luxury apartment, and work a high paying tech/finance/legal job", I can understand the financial perspective of hating on these investments.
(Morally, I think that's fucked, but what can you do)
comment in response to
post
I think there's a segment of the population where this is true, but I find it hard to fully agree with this take.
If you own a mortgage in PDX, even in the *most* selfish interpretation, you're literally invested in the success of the county over a 10-30 year horizon.
comment in response to
post
there has to be some leeway. I heard horror stories from employees working on OHA / m110 - budget was unspent, then it was misspent, now we're STILL paying out-of-state contractors for unused services
Responsible spending shouldn't mean "we always spent exactly everything immediately"
comment in response to
post
This seems similar to the www.wweek.com/news/schools... article.... Yeah, it's bad for local gov't to levy taxes and not spend them, but it seems *way* worse to spend that money irresponsibly.
The county holding preschools to high standards to serve kids should be good news??
comment in response to
post
yeah, reading the article, this is a weird choice of headline
- Seems like it's working, if a little slow
- Seems like the county is having some back-and-forth with preschool providers, rejecting ones that aren't up-to-standard
"most rejected providers did not apply again", but should?