zwaps

85 ms·

811.
by zwaps 7y ago
But what is it? It says it is an API or module? Is it a DL framework? Is it an extension of TF/Jax with new models? Is it a set of tutorials? Are we to work through the code? Is a part of the doc for Jax? Is it a set of implementations
812.
by zwaps 7y ago
Is it just me or is there zero explanation to what this actually is? It somehow "helps" me understand deep learning but its tutorial / doc is one python notebook with three cells where some nondescript unknown API is called t
813.
by zwaps 7y ago
First, software should be efficient to use. Too many designers, maybe a majority, make "beautiful" software that is a pain to use. Are forms today better than they used to be? I don't think so. Most "web based" or &
814.
by zwaps 7y ago
https://scholar.google.com/citations?user=c9egBlsAAAAJ&hl=de...
815.
by zwaps 7y ago
Cause at this stage opting out included navigating many stages of dark pattern and deceptive pages, including semi hidden links and fake slow progress bars. Even on the biggest sites. If, and I say If, these gangsters ever get hit with maj
816.
by zwaps 7y ago
I mean, one may assume incompetence and corruption. And perhaps that's the case. But on the other hand, teaching languages are not the same as production languages. There are a number of languages that may be more powerful, but I think
817.
by zwaps 7y ago
It's interesting that in many countries of the EU, or even in the EU as a whole, all this is simply not possible. Not because of dark patterns being illegal, but rather because they would not be profitable. As an example, in Germany: F
818.
by zwaps 7y ago
Parallel programming in Python is hard. On Windows, it's hell. In Matlab it's a breeze. Just put "par" in front of the thing. Done. If you wanna shell out money to enable that functionality, that is.
819.
by zwaps 7y ago
Sonnenschuld is a beautiful word! Thank you!
820.
by zwaps 7y ago
The problem of recommending something (especially something genuinely new) is actually quite complex, which you realize by studying inferential/causal statistics. A statisticians would say that you are making "heroic assumptions&q
821.
by zwaps 7y ago
Wow those React examples are an absolute mess, compared to the few lines of C# and Elm. No wonder websites are so slow nowadays.
822.
by zwaps 7y ago
I love that I have to accept ALL tracking & cookies to even read that webpage
823.
by zwaps 7y ago
Probably the biggest issue is pre tenure, as there isn't even a possible way to do anything else. On the plus side, my work as PhD is self directed and I can most often come in whenever I like, and work whenever I like. Also, of course
824.
by zwaps 7y ago
At the cost of a crappy and sluggish user experience, of course
825.
by zwaps 7y ago
How could it be one? Loading text, or anything statically, is faster than many megabytes of JavaScript frameworks. I think web devs have collectively broken their brains if they think this.
826.
by zwaps 7y ago
I can say that I have yet to see ANY web app, that is fast and snappy. ANY. At some point, they all have problems. Even many frameworks for iphone and android that are essentially web apps are terrible and make every app slow and miss click
827.
by zwaps 7y ago
Depends on where you are. In the EU, parts of contracts that sre against the law, which includes labor without payment, are invalid and you are free to sign it. In this case, giving up IP you made in your free time requires a new contract,
828.
by zwaps 7y ago
No it is not. In Germany, you need to make a new contract with explicit payment https://www.urheberrecht-leipzig.de/urheberrecht-nutzungsrec...
829.
by zwaps 7y ago
Honestly, have you seen a rigid car hitting a crumply car going full force? In a situation where maybe both would have survived otherwise, now at least one party is goo. It's a Prisoner's Dilemma. If everyone drives fat, rigid SUV
830.
by zwaps 7y ago
Price discrimination itself is not illegal (for example in the EU), it is only so if it is based on certain parameters. And this you'd need to prove in court, which makes it unlikely that regulation will solve this issue any time soon.
831.
by zwaps 7y ago
One need not perfectly discriminate. Any discrimination eats consumer surplus. It is a major concern. It is used in booking sites, and it will be used in other sites. We can not rely on regulation. My point stands: User targeting is a threa
832.
by zwaps 7y ago
You probably would be well advised to read up on the economics of consumer differentiation and price discrimination. All of today's ad-tech is set up and developed for that purpose. Whether overtly or covertly, it is the singular purpo
833.
by zwaps 7y ago
The difference is that users can go to privacy, and switch a single option and they are opted out. In Android, and with Google in general, you have to go through many different menus and hidden option in different places. And then, you can&
834.
by zwaps 7y ago
This is why I bought a brand new iPhone last week. It is my first Apple product. I like this, and I am willing to spend a lot of money. Best I can do is probably let my wallet do the talking.
835.
by zwaps 7y ago
They have basically one opportunity to set the expected price point and as it turns out, they prefer to get all the positivity and hype on their side, than perfectly optimize to marginal demand. I am not sure it is a bad choice.
836.
by zwaps 7y ago
When talking about economics relations, there is a difference. In most cases, and I am sure that was meant here, there is an "outside option". And then things differ. If we analyze supplier networks, trade relations in old style b
837.
by zwaps 7y ago
Reading a bit more, it's really interesting that your result relies on Lemma G.4, which is a CLT based on independence or mixing (as you wish), whereas all theorems assume that anything put in is Gaussian. It "smells" like th
838.
by zwaps 7y ago
Good results often seem "obvious" in ex post. So take this as a compliment. Intuitively (I have never read a paper in this field), since you are talking about wide networks, I also expected that a CLT would be used. For "dens
839.
by zwaps 7y ago
>> Mitchell said. “They believe no third party can be trusted. We take a different position: that trust should be established directly between consumers and the brands, and publishers they trust, and with the third parties that those
840.
by zwaps 7y ago
Seems like multi-processing is already better than Python, which is a hot mess. For things that you'd actually need multi-processing in, say, deep learning - like preparing and loading and then saving data, Python is difficult to use.

More ›