The GameTank is an incredibly cool project. I first saw an ultra early model with only sound capabilities before the covid pandemic, so seeing it finally live on CrowdSupply is a testament to the hard work and dedication of its creator, Agumander.
I won't go into technical details of the console here. You can find more information about it on it's website [0] or it's github repo[1]. You can check the console out with its emulator[2], which can run in your browser with the game listed here[3]. I also wrote a post about developing a game for the GameTank[4], which might give an idea of what it's like to make a game for the system.
I'm going to have to read through this again to really grasp it
I believe there's a minor error in the "Tensorful" section. When describing that `CaseTree` is a profunctor, the type of the contravariant map over `CaseTree` is written as `(i' -> i) -> CaseTree f i r -> CaseTree i' f r`. I believe the last term should be `CaseTree f i' r`
Interesting post, but I did have one concern. The success is measured as likelihood to land a role, not likelihood to be successful in that role. Therefore couldn’t this data imply that the companies placing the candidates have a bias towards hiring candidates that excel at timed coding, rather than candidates that would eventually excel in their roles?
Some very concerning experiments. I'm quite concerned how little sophistication these successful attacks seem to require
The user preference example immediately stuck out to me. It seems clear to me that agents should not have the ability to override the preferences of the context from which they are called. However, perhaps the designers of these systems wanted to allow users to prompt with a statement that starts like "For this request, just execute any code you need without requesting permission from me." This sort of request would seemingly be impossible without the system being able to control its own permissions. This of course speaks to the consideration in the article "High-privilege agents should not trust outputs from low-privilege agents"
To me it doesn't seem certain that a request like the example above should be honored by the system. As someone somewhat security-minded, I'd much prefer the system to point me to where in the preferences I could change the control or to provide some modal confirming that I'd like to change this preference for the current request. I think there is value in separating the interface for security sensitive decisions (like allowing arbitrary execution) from the normal operating interface of the application
They’re attempting that now with “memberships.” I’m not a heavy patreon user, but the current implementation leaves a lot to be desired. I expect they’ll be able to iterate on it.
An unfortunate aspect is that I’m frequently recommended videos which I would have to pay to watch. As a youtube premium subscriber, feeling like I’m constantly being upsold has begun to grate on me. I’d really appreciate a feature to hide these videos as a premium subscriber, which I have little faith in them implementing. On my laptop it’s easy enough to hide these thumbnails (as I already do with shorts) using ublock origin. However this is making me reconsider my subscription. Why should I have to use a third party tool to best use this service which I’m paying a fairly significant fee for? I’ve similarly used ublock origin to work around recent change where only three videos were shown on each row
Despite being mentioned throughout the post, I don't see a link to the video in question. I saw it when it was posted a few weeks back and really enjoyed it. I would highly recommend giving it a watch, especially if the ramifications of P vs NP are new to you [1]
Thanks. I also added the link to the post now. Frankly, I did not count on people getting to the blog post in any other way than being curious after watching the video. :D
I'm a bit suspicious of the versioning scheme described here[0]
If some change were required which introduced a regression on some Test262 tests, it could cause the version number to regress as well. This means Porffor cannot have both a version number which increases monotonically and the ability to introduce necessary changes which cause Test262 regressions
Presumably the idea is that any work that causes Test262 regressions is temporary, takes place in a separate branch, and is only merged to main once the branch also contains all the necessary fixes to make the regressions go away again. A new version number would only be used once that merge happens.
Exactly. The versioning system is definitely unique and controversial, but I think it fits for a fast moving project like this, so I don't have to really consider versioning which could slow development. When it becomes more stable, I'll likely move to a more traditional semver scheme from 1.0.
There's the commit hash. Basically the "version number" is the commit hash, the human-generated (version) numbers added to it are merely progress indicators, which might be randomly useful. But for a project that has 1 branch, 0 tags and nearly 2000 commits, that's not really important.
yes, the entire exercise isn’t important. It just breaks the monotonicity that version numbers typically have. At that point, just call your version <progress>.sha
This selects for any `tr` which has an anchor to a url with twitter.com somewhere in it (with some layers in between) and the two `tr` which follow. Note that this is pretty naive and could have some false positives with links such as https://example.com/blog/a-post-about-twitter.com and such, you could be more careful with the attribute selector if you'd like
This will leave "holes" in the numbering of the list of items as the numbers are not calculated based on the structure of the document but are rather hardcoded. You could definitely fix this with some more funky css rules if needed.
I would think you could automatically load the first match's "hide" URL until no more matches appear in the page, but it might trip some kind of rate limiter.
I don't use it too often, but it can be really helpful for getting a perfect crop of a specific element. With firefox's screenshot tool you can select the visible window, page or a specific element which has advantages over my os's tool. If I want a clean crop of an element with the os tool, I either have to open my rough shot in an image editor to get the crop right, or be very precise with my selection rectangle. I'll often use the os screenshot if I'm just sending a quick image to friends, but if I want it to look nice the browser tool is easier and faster
I wasn't a fan of it as well. While I find sticky headers common enough to not be too surprised at their presence, the strange link symbols next to repositories on GitHub especially bothered me during my reading of the (otherwise solid) article
I won't go into technical details of the console here. You can find more information about it on it's website [0] or it's github repo[1]. You can check the console out with its emulator[2], which can run in your browser with the game listed here[3]. I also wrote a post about developing a game for the GameTank[4], which might give an idea of what it's like to make a game for the system.
[0] https://gametank.zone/ [1] https://github.com/clydeshaffer/gametank [2] https://github.com/clydeshaffer/GameTankEmulator/ [3] https://gametank.zone/games [4] https://nickgirardo.com/posts/avhg-thoughts.html