>I think OpenAI may have taken the medal for best available open weight model back from the Chinese AI labs
I have a bunch of scripts that use tool calling. Qwen-3-32B handles everything flawlessly at 60 tok/sec. Gpt-oss-120B breaks in some cases and runs at mere 35 tok/sec (doesn't fit on the GPU).
But I hope there's still some ironing out to do in llama.cpp and in the quants. So far it feels lackluster compared to Qwen3-32B and GLM-4.5-Air
Drew Breunig has been doing some fantastic writing on this subject - coincidentally at the same time as the "context engineering" buzzword appeared but actually unrelated to that meme.
How to Fix Your Context - https://www.dbreunig.com/2025/06/26/how-to-fix-your-context.... - gives names to a bunch of techniques for working around these problems including Tool Loadout, Context Quarantine, Context Pruning, Context Summarization, and Context Offloading.
Rule #1. BQ is not a standard database. If you use it like one, it will cost a fortune.
Rule #2. BQ is amazing for being able to churn through and analyze massive amounts of data, and can very well be the best option in some use cases.
Rule #3. Letting "just anyone" run queries is as dangerous as casually handing a credit card to your drug-addicted cousin. Just wait until you get the bill!
Rule #4: Partition and cluster your data wisely. You don't have indexes.
Rule #5: Duplicate data. Throw all of the normal forms out the window. Storage is cheap, computation is expensive.
Rule #6: BQ is not meant to be used like MySQL. It's "spin up" time is too slow, but you would be hard-pressed to beat its performance on truly large data sets.
My perspective: One of our customers has a database growing by 17 gigs a day. One of them. There's several on the same scale. Yes, it's necessary. Another instance: One of our customers spent $8k in one month because limits were not properly placed on the account and we didn't catch it until the bill came. We monitor better now. A different instance: We had a dev trying to optimize a query, and they spent $250 in queries to get the cost down from $50/query to $15/query. Most of the time, though, our queries are only pennies.
Now that I've written all of this out, I feel like I need to record a video about it. There's not a lot of BQ info aside from the marketing fluff put out by "teh Google".
There's a very recent talk that they did at 37c3 that you can watch, if you're interested in the process of them gaining root on the various platforms [1][2]
There’s a really good article from Spotify Engineering that looks at exactly how Spotify bridged this gap between “random” and the “random” people actually expect.
It a good read on understanding what people generally expect when they ask for a random stream of songs (or comics), and how you can meet that expectation by carefully engineering how you generate “random” lists.
I recommend using an ad-blocker while visiting that site :-/
Lately, I find myself using more and more plugins to make the "modern web" tolerable. To list a few:
Channel Blocker (lets me block channels from search results on Youtube);
uBlock Origin;
Disconnect;
F.B Purity;
Consent-O-Matic (auto fill cookie consent forms);
Kagi Search;
PopUpOFF;
Facebook Container;
Privacy Badger;
ClearURLs;
Return YouTube Dislike
Basically, if I visit a website and don't like the experience, I either never go back (Kagi lets me exclude it from search results) or find a plugin to make it tolerable.
What I really want now is the ability to exclude entire websites from any permissions I grant to plugins. I feel like in the last year, I've read a couple stories about companies buying successful plugins and then using them to track you or show ads or whatever. I'm worried this will be the next stage in the battle for our attention -- best case: companies will buy popular plugins to track us and show us intrusive ads; worst case: nefarious actors will buy them to scrape information we think is private and collect it.
IE: I just want to be able to say "Hey, Firefox... those permissions that I granted to plugins x, y, and z? They don't apply to www.myfavoritebank.example.com"
Is there a browser that has that feature yet? I spent a few hours trying to figure out if Firefox did. It did not appear to.
edit: Added semicolons to separate plugins in list b/c HN stripped the newlines from my comment.
Since July of this year I've been CTO for a small games company, I've worked in this industry before in a more specialised role but this is much broader scope.
I've gone from being responsible for 1 person to 10 which will likely become 50 by the end of next year.
I'm terrified.
I'm terrified because I feel like I will let my reports down.
I'm terrified because a lot of the work I'm doing lacks transparency, how do I tell my compatriots about things that may not come to pass, especially deep political things -- and the amount of work necessary to onboard/offboard people, about "strategic partnerships" and why they're taken, or not taken or any of the hundreds of other things which are changing direction 3 times a day each.
I am technical, deeply, passionately technical.
I will not do service to people by being technical, my days are too fragmented, my worries too broad, it's impossible to dig into a small focused problem and give it the treatment it deserves - and even if I could then I'd be dooming one of my colleagues to maintain it forever because I couldn't possibly..
My being technical is good because I know the problems and I know what can make people feel valued; I can give people autonomy and direction and I can understand when help is needed and how to give it.
However, almost nothing in my work is technical, it's all presentations, discussions, breakdowns, timelines, contracts and negotiation.
The most code I write now is to do cost reporting.
My most frequently accessed cloud console pages are the billing sections.
This is not fun at all (and frankly, I didn't think it would be); but I fear what will replace me if I leave.
Great article. I have hosted hundreds of happy hours, dinner parties, cocktail parties etc. Here are a few things I'd add:
- Name tags. Please, please use name tags. They might seem "formal" or "corporate" but they're also inclusive (no cliques!). I'd rather feel awkward asking someone to wear a name tag than feel awkward forgetting their name.
- Consider hosting on what I call "non red-level days," aka days that are NOT socially competitive. Socially competitive days or "red-level days" in America tend to be Thurs, Fri, and Sat nights. Also holidays and long weekends. People schedule big stuff on these nights. Make your party easy to attend: host it on a Monday, Tues, or Wed night.
- Set a start AND an end time, and mention both when you collect RSVPs and send reminder messages. End times help get people to show up on time. They also give people an easy out to leave.
- 2 hours is the best length of time for an event like this. I like 6-8P or 7-9P.
- Get a group photo! You'll be proud of your event. And you can use the photo when you invite people to your next party.
- Don't forget to send reminder messages to everyone who RSVP'd leading up to your event. I like sending my reminder messages 1 week prior, 4 days before, and then on the morning of.
Good luck!! I think more people should host parties and happy hours. It changed my life and helped me build a network and relationships to launch my last company, Museum Hack (sold 2019).
I recently self-published a book of every little tip and trick from hosting events to teach you how to host your first party. The book is called 'The 2-Hour Cocktail Party: How to Build Big Relationships with Small Gatherings' and it has 230+ reviews on Amazon and Audible here-- https://amzn.to/39rfb2V Happy to give a satisfaction guarantee for any HN readers. You can Venmo request me @nickgray and email your receipt to [email protected] if you don't think my book is filled with actionable, tactical, extremely practical advice for hosting events. You can read the first few chapters on my site here https://party.pro/book-readnow/
OK good luck!! You should host a party!! Your event will be awesome!! We could all use some new friends these days.
Readers may also enjoy Steampipe[1], an open source [2] tool that uses Postgres FDW's for live SQL queries against Jira, Confluence, GitHub, GitLab, AWS, Slack and many more tools [3].
I have a bunch of scripts that use tool calling. Qwen-3-32B handles everything flawlessly at 60 tok/sec. Gpt-oss-120B breaks in some cases and runs at mere 35 tok/sec (doesn't fit on the GPU).
But I hope there's still some ironing out to do in llama.cpp and in the quants. So far it feels lackluster compared to Qwen3-32B and GLM-4.5-Air