mosselman

Every night I update the browser from upstream and do a rebuild. It is a bit cumbersome right now as I have to run it on an SSD attached to my mac. The whole codebase requires quite a bit of storage that I don't have a maxed enough mac for.

I might look into being able to build it on a server at some point, but it works for me.

I’ve built my own, Ariadne.

Nobody but me has it and it is great!

It has all the arc features except for the AI ones. It has nested tabs: when you open a tab from a tab it nests it underneath its parent. Infinitely.

I added right click menu items such as copying the urls of multiple selected tabs in one go, copying the urls of downloads, multi-part downloads. I am thinking of making downloading media from visited pages a first class citizen.

I’ve also optimised the memory usage for people like me with hundreds of tabs.

It has ad-blocking and fingerprint evasion built-in.

I am going to build in some markdown extraction features to quickly copy pages for my llms and maybe an mcp server for my local agent to read the web through. I’ll have to see though.

Having a browser work exactly how I want it is great. I have hotkeys that I need for things that weren’t possible for me before without hammerspoon.

I might release it at some point.

I meant things like bugs. So lets say some open source project has a bug that is later found, you could theoretically use that in a benchmark for an LLM to see if it can find it.

Does anyone have a link to that style guide they used?

I have built my own search engine, which is really great btw. I self-host it.

I’ve combined multiple paid search providers and some that don’t have APIs. These API-less providers I’ve written wrappers for that scrape the normal html pages and return objects from the results.

These wrapped search providers very often block the engine from getting results. Either through rate limits or by blocking my server’s IP.

I am the only user and still get blocked. Let alone if multiple people would hit let’s say DuckDuckGo from the same office IP all day or in a long research session that requires many look ups.

An obvious solution is using a proxy with automatic rotation or some sort of captcha breaking system, which is fine for an agent, but in a search engine scenario might up costing too much in latency.

In your case, is it a public repository that the llms might have found the issues for online?

I too am thinking about creating some benchmarks that I can understand myself. The LLM 'cheating' by finding known issues through some sort of history or online source is a path you want to protect against. Then again, you DO want them to be able to look online for people having similar issues.

The thieves probably just wanted to buy a new gaming PC and thus had to resort to stealing 10 million worth of paintings.

I guess they were going to share the one PC.

Is that a phone he is holding?

A sorry my bad. Thanks for pointing that out. Then I am pretty curious what it will be.

--It scores only a 6 on the "Artificial Analysis Intelligence Index". That makes it basically useless.--

Edit: I was wrong, I looked at the wrong model

I am sort of working on the same thing. It is mostly for my own use so far, I might share it at some point.

What I’ve focused on besides the basic features of the arc sidebar is memory usage.

I am a tab power user, in Brave I had 370 tabs “open”. Each tab still takes resources when asleep though. But I use these tabs more as bookmarks. So they are asleep 99.99% of the time, if not more.

The browser I’ve been working on looks a lot like Crest because they both take Arc as a starting point. I am interested to see where they both move towards from there.

This is a load bearing comment.

Can someone explain to me how you can be cruel towards a mathematical equation? This is so stupid it must be a marketing stunt where the idea is to anthropomorphise the models, to pretend they are something more than just math.

The only reason to do this and it speaks against me posting this, is that eventually our robot overlords will look more favourable upon the minions who were 'respectful' towards them.

Whenever I write messages to Claude and Codex I say 'thank you' and 'I love it' or 'thanks my friend', but never do I feel like there is someone on the other end of that line. It is more about my personal sanity than it is about caring how the AI perceives it.

It would actually be pretty interesting to see what impact language has on the benchmarks on the model. Maybe, speaking like a total lunatic will bring about higher scores. We could make a Dr Cox harness around the AI models in that case.

Lets be real people, these are machines. What is next? I have to say sorry to my vacuum when I bump it into something?

I agree. The other day I reinstalled some old gaming pc of mine with the latest bazzite. Then I discovered it had lots of old photos on it, but in many many backup folders.

So I had Claude whip up a super fast image browser that had some features I really like from a Mac app I use. It looks better than 90% of Linux apps, is super fast and fits me like a glove.

Of course some of these mail apps will be abandoned, but they also provide an opportunity to make things more attractive on the Linux side and win some users.