HNHacker News
TopNewBestAskShowJobs

nijave

2,500 karma · joined August 22, 2018

[ my public key: https://keybase.io/nickv; my proof: https://keybase.io/nickv/sigs/qMdHcggaKRYKr002ExXFme5MjZeOQbHt-PyAl_07HDk ]
submissionscomments
nijave··on Show HN: Open-source engine running Gemma 4 26B in 2 GB RAM on any M-series Mac
Any idea if madvise helps? Admittedly I have very limited experience and only on Linux
nijave··on Keychron announces first open-source firmware for gaming mice
Hopefully better than their QMK fork. They force push over their branch in their fork each new release effectively erasing history of the last version...

Of course there's no change log anywhere either...

nijave··on MAI-Cyber-1-Flash inside MDASH
>does this mean Microsoft's model is best at fixing Microsoft products

In theory, they have tons of network-type data and at least some Linux data from Azure. In practice... Good question...

Identity is somewhat standard-ish LDAP/Kerberos/OIDC (ok, sure, they've added a bit of creativity)

nijave··on Elevated errors on Claude Opus 5
I should have mentioned Claude Code, specifically. It sounds like other harnesses are better. I'm a bit entrenched with CC at this point since that's what we have a subscription to at work and they don't support other harnesses on subs
nijave··on Opus 5 is currently #1 on Artificial Analysis Intelligence Leaderboard
Forgot about the ol "but how many tokens did you spend to get _there_"--wish benchmarks would include the number of input/output tokens to achieve the score. I think the closest is Android Bench https://developer.android.com/bench although best you can do is extrapolate off time/cost (iirc they claim to prefer using provider's native API)

In general, smart models work fine with any tools, dumber models need better tools to achieve same results but better tools can eat more context/take more turns

I've gotten decent results with Llama 3.1 8b on Hugging Face tester with Exa MCP since it seems to dump sufficient context into WebSearch/WebFetch type calls even a crappy model almost always gets back what it needs as long as it calls the tool at least once. I had Claude Code look at previous sessions with SearXNG vibe MCP compared to Exa MCP and results got better when it modified SearXNG to work very similarly to Exa. Ended up with this https://gist.github.com/nijave/604c43e3e0fdcd60f5280d3a6b109... although it's really only optimized for "search" not "fetch" at this point. Fetch is basic niquests without Javascript or anything clever

nijave··on Elevated errors on Claude Opus 5
Anthropic has an outage nearly everyday. GitHub seems like they're down to about once a month again
nijave··on Elevated errors on Claude Opus 5
I wish the harnesses would auto resume but I suppose that would also add more load without more money for subscription customers...
nijave··on Some more things about Django I've been enjoying
1000ms / 12 = 83ms per request which sounds normal
nijave··on Opus 5 is currently #1 on Artificial Analysis Intelligence Leaderboard
I think on swebench verified luna was only like 3% points lower for 1/5 the cost

Like 96% vs 93% or something

nijave··on Self-host your mail server
There used to be some web form for msn/Hotmail/outlook you could submit your IP to and 1-6 months later it would magically start working _most of the time_

Not sure if that's still the case

nijave··on Self-host your mail server
Email hosting isn't too bad. IP reputation management is a pain. Everything works perfectly until one day it doesn't.

SMTP relays are cheap--worth a few pennies a month to not deal with IP rep.

nijave··on Self-host your mail server
Easier to self host inbound and relay outbound. SMTP relays are effectively a commodity and pennies for personal usage. Worth <$1/yr to not screw with IP rep unless you own a /24. Less than /24 you risk some knucklehead getting the whole block listed
nijave··on Self-host your mail server
Well, generally you need 2 different programs regardless. Usually an SMTP server for accepting and storing emails and a IMAP server for clients to connect to and read stored emails.

For the parent post, you configure an outgoing SMTP relay

Iirc called smarthost in postfix

nijave··on Self-host your mail server
I had a Digital Ocean VPS and it was fine about 7 years but on year 8 the entire /24 got blacklisted and it was game over

Took about 6 months initially of fiddling to get other email services to accept mail DKIM, DMARC, SPF, PTR, web form requests

nijave··on Short sellers notch $8.7B profit as SpaceX shares dip to IPO price
Starlink is so cheap because economies of scale ie volume of launches. They only have a large volume of launches because Starlink (123 ish of 165 ish launches in 2025)

Without Starlink, they would need to charge government launches much higher.

Starlink has really fast asset depreciation at ~5 years as well as competition from ever expanding terrestrial wireless (5G NR)

nijave··on SpaceX bond worth 10% less than issue price – heading for junk bond status
https://en.wikipedia.org/wiki/Bypass_Paywalls_Clean

It is now on sketchy Russian forge due to DMCA takedowns

nijave··on Short sellers notch $8.7B profit as SpaceX shares dip to IPO price
There's a radiation section in the article, I'll let you draw your own conclusions

https://arstechnica.com/space/2026/07/how-hard-is-it-to-buil...

nijave··on The state of open source AI
My (unverified) AI research claimed generally Chinese models are cheaper to train because Chinese data scientists are cheaper to hire and they're also under more pressure to optimize training cost due to limited hardware availability

Seemed believable but not sure where that's true

Chinese AI companies are generally smaller tho and the models they're releasing are also smaller (I think estimates put OpenAI and Anthropic SOTA into trillions of parameters)

nijave··on The state of open source AI
I think (and have heard) the Chinese govt is also interested in trying to embarrass the US as well by showing off their capabilities (so there's a political angle, too)
nijave··on Short sellers notch $8.7B profit as SpaceX shares dip to IPO price
When I checked, Starlink seemed somewhat valuable but highly questionable addressable market (limited growth)
nijave··on Short sellers notch $8.7B profit as SpaceX shares dip to IPO price
>but in an affordable and sustainable manner

Affordable is a bit questionable. It's heavily subsidized by Starlink and government.

Not completely clear what happens to price if SpaceX doesn't keep launch demand propped up with their own business.

nijave··on Short sellers notch $8.7B profit as SpaceX shares dip to IPO price
Training, no. Latency is too high.

Inference the latency becomes trivial.

Other things, I suspect latency is too high again.

nijave··on Short sellers notch $8.7B profit as SpaceX shares dip to IPO price
Ars had an article that cited some HPE testing on the international space station that said regular hardware is _probably_ fine up to about 5 years

Definitely not definitive but it's plausible current hardware could survive with minimal modification

nijave··on Short sellers notch $8.7B profit as SpaceX shares dip to IPO price
GPUs depreciate super fast. It might last 5-7 years but it's already outdated at 2-3

Also space has more radiation

nijave··on SpaceX bond worth 10% less than issue price – heading for junk bond status
magnolia1234 bypass-paywalls-clean

https://archive.is/tnSeY

nijave··on The infinite scroll may become endangered if controversial Calif. law passes
position: fixed has been around a long time

A lot of implementations also break the back button and "page" or scroll position links (you can link to a singular item but not a spot in the list)

They usually also break psuedo binary search and force linear/sequential search (you can't skip a head to page 10 when you're trying to zero in on a date unless they've added an explicit date filter and you remember the exact date versus the relative position)

Also breaks parallel loading--can't queue up the next few pages of gigantic media in new tabs while finishing the current page

They also tend to break ctrl-f and if they don't they tend to get progressively slower as more crap stays loaded in memory

nijave··on What are Forward Deployed Engineers, and why are they so in demand? (2025)
Everyone got tired of renaming sysadmins so they moved on to a new role to rename.
nijave··on TLS certificates for internal services done right
Yeah, specific RRset
nijave··on You're Not a Better Engineer Because You Type Git Commands by Hand
There's too much nuance--it would take a dedicated skill
nijave··on You're Not a Better Engineer Because You Type Git Commands by Hand
I suppose that's fair. The assumption here is things get linked back to the issue tracker and it's not very interesting boring business apps.

Anything technical tends to be a link back to discussion on someone else's issue tracker

← PreviousPage 7 of 34Next →