advael

69 ms·

931.
by advael 5y ago
Why do you think that defining accuracy in relative terms works in favor of this model? This pairwise relative measure should give you less confidence that the model generalizes, because now we don't even have an idea of what the relat
932.
by advael 5y ago
Nah, deep dream was computer-aided pareidolia. This is just regular old computer-aided prejudice
933.
by advael 5y ago
Yes, and this further demonstrates how ridiculous the reporting on this result was. Their sample population was tiny and skewed, and this paragraph is a great example of why you can make your results look better to lay readers by reducing
934.
by advael 5y ago
Duh. My point isn't that someone designing some awful kafkaesque "performance algorithm" is somehow better than some human manager voting on which employees they like the most, it's that you couldn't even tell the d
935.
by advael 5y ago
Yes, when you look at phenomena in retrospect, you can decide that all kinds of things are the sole unique cause of other things that happened afterwards In general, it's very easy, starting with a model and a bunch of phenomena, to ex
936.
by advael 5y ago
Have you read the actual paper? That figure is at best inflated, and their methodology is riddled with confounds, including but not limited to the role of social networking profile pictures as dating signals, intentional signaling such as g
937.
by advael 5y ago
Sure, you're going to get some people gaming the metrics, (as happens with all metrics). Currently, by keeping the metrics opaque, Amazon (et al) prevents people from explicitly doing this, although it is using said metrics as a strong
938.
by advael 5y ago
Seeing a lot of "Well but what if it works" here, and I'll weigh in as someone who does AI - including CV-related applications involving convolutional neural networks - professionally and has participated in academic research
939.
by advael 5y ago
Unfortunately this sort of thing is pretty much inevitable when a large org like Apple makes itself the single bottleneck for all apps. No org that big can actually serve everyone's needs (Cue people whose needs it serves coming to tel
940.
by advael 5y ago
Absolutely a fair point
941.
by advael 5y ago
As a knowledge worker, I know for a fact that the market rates for my work have everything to do with the calculated need the company that hired me for has, and any intellectual property that may get produced by that work is only captured a
942.
by advael 5y ago
There are several problems with both the implementation and the principles underlying intellectual property laws, some of which pertain to who owns what, many of which pertain to how ideas can be separated out from each other, still yet mor
943.
by advael 5y ago
Ah, the ol' "This is part of my idea of the status quo, so everything that's happened since that could roughly pertain to its stated purpose serves as evidence that this practice uniquely produced all the outcomes I like&quot
944.
by advael 5y ago
Yes. Under the current laws, people can be used as tools for pay. Also, without intellectual property, people could be used as tools in the same way. But under the current laws, people can be used as tools to produce something that might pr
945.
by advael 5y ago
What consumers currently use isn't evidence of what they want, it's evidence of what they want out of what's available to them. The widespread "piracy" epidemic was also pretty good evidence of revealed preference,
946.
by advael 5y ago
The provisions you describe are a minor inconvenience compared to the sweeping provision that criminalizes circumvention of anything that could be construed as DRM
947.
by advael 5y ago
Irrelevant. The comment I was responding to said the following: "...I don't think many people here would be familiar with the day-to-day, mundane aspects of IP law which probably, for the most part, works fairly well." I am r
948.
by advael 5y ago
I don't think there's strong evidence that piracy even was an existential threat in practice, but regardless the new business models that have emerged are in fact far worse value propositions for consumers that want to be able t
949.
by advael 5y ago
I don't buy it. I know people who have worked for patent firms, and independent artists who have tried in vain to enforce the rights they're supposed to have when someone sells a T-shirt with their art on it without asking, attrib
950.
by advael 5y ago
Perhaps, if said author is able to produce the whole book such that it can be copyrighted, find a way to publish it independently, and defend their copyright if a publisher likes the idea and gets something similar enough to not quite be co
951.
by advael 5y ago
I posted some explanation as a reply to AmericanChopper below that I think is a good summary of my take on this
952.
by advael 5y ago
How it already works: Arrange to get paid to do it by someone, or find a way to monetize it. The uniqueness of an idea is not what you're actually selling even in the current context, and the idea that you can go do some R&D as an
953.
by advael 5y ago
That's totally nonsensical. Yes, capital pays for labor to produce a thing, and then owns the thing. For "things" where there's an actual good being produced, this makes perfect sense under property rights. We do not nee
954.
by advael 5y ago
I don't think it can function, and I think trade secrets are frankly quite weak. In practice, trade secrets don't usually stand up to sophisticated reverse-engineering efforts, and patents frequently and consistently fail to discl
955.
by advael 5y ago
This kind of gets into philosophy, but I think that ownership of ideas is just an indefensible principle in the first place, which is why people generally only justify IP with rose-tinted ideas about its intended practical effects. Here are
956.
by advael 5y ago
I agree, and my strong stance against IP as a legal principle in general is extremely unpopular in just about any circle I talk to. I nonetheless think it's both correct and important. I think people have bought pretty heavily into the
957.
by advael 5y ago
This seems like the antebellum for an IP reckoning similar to what resulted in the DMCA in the United States in the late 90s. The technology is getting to the point where it makes the law as it stands pretty nonsensical in the face of what
958.
by advael 5y ago
Solutions that mandate companies to continuously actively maintain software can't and won't work. Even if they have to set aside a budget, they could obey the letter of the law and fail to do this, no matter how the law is framed.
959.
by advael 5y ago
I agree that enterprise-level collusion has locked you and many others into proprietary platforms. Again, I advocate using whatever tools work best for you, if your priority is not control over your tools. Every year the FOSS ecosystems get
960.
by advael 5y ago
To be frank, I think the most parsimonious theory for why these trades involve so much violence is because we've made them black markets and forced the entities that successfully operate in those trades to be equipped to contend with

More ›