maksimum

23 ms·

61.
by maksimum 9y ago
In fact Nvidia cards are the only ones that are recommended for hackintoshes for OOB compatibility [1]. IME on Mac OS 10.12 getting a Nvidia GTX 1060 to behave properly has been much easier than a AMD RX 480. [1] https://www.tony
62.
by maksimum 9y ago
Define 'hard'. To me, almost anything that can be completed in a fixed amount of time with high probability isn't hard; it's tedious at most. But if it can be completed within one hour it's not even tedious, it&#x27
63.
by maksimum 9y ago
Is the ratchet a good design strategy for software updates (e.g. Chrome, Mac OS)?
64.
by maksimum 9y ago
> The average user's mindset must have changed from when I was younger Yes, for a simple reasons: the average web user today is basically the average person. The average web user 20-25 years ago was a technology enthusiast.
65.
by maksimum 9y ago
I think the point here is that we don't need to define a fitness function for "good music", but a fitness function for e.g. "reconstruction error" for auto-encoders. While the latter is still challenging, it seems s
66.
by maksimum 9y ago
What do you mean by "find" and how do you define "global optima"? For "find" you could discuss convergence rates vs. the points that are being converged to. If you add randomness at every iteration you're
67.
by maksimum 9y ago
Not that I necessarily agree with OP's "call to action," but as software people we have more potential for impact on software-related issues...
68.
by maksimum 9y ago
You're right, additional properties are desirable. The ACM has recently published a [set of guidelines]( https://www.acm.org/binaries/content/assets/public-policy/20... ) for algorithmic decision maki
69.
by maksimum 9y ago
> The particular set of results that Google shows you on Google.com are not a decision about you I think that's easily demonstrated as not true: when I search for something like `luigi` in a regular window, I get the github link 1st
70.
by maksimum 9y ago
I took a video of traffic flow on a NYC avenue: https://youtu.be/RXIyzdLl3vk I wanted to see whether the lights were being used effectively to let the motor-vehicle traffic flow without stopping, and what potential causes f
71.
by maksimum 9y ago
> as any big city resident can tell you. I used to take the bus every day in Prague (Czech Republic) for about 6 years. It was almost always on schedule in both the morning and afternoon. I don't know anywhere near enough to argue w
72.
by maksimum 9y ago
> If Bud and Sam Walton knew they couldn’t pass on their business to their kids, they would either have never done it in the first place or they would structure the business to close out well before they died How confident are you in tho
73.
by maksimum 9y ago
> make a more sensible amount which still attracts and sustains the best talent in the enterprise. Who should be in charge of how 'sensible' is defined? And how can you argue that it attracts the best talent? Most people outsid
74.
by maksimum 9y ago
You'd probably be fine, scariness overflow.
75.
by maksimum 9y ago
I'd like to stretch some ideas for the sake of exploration. Curious to hear what you think: I think you could interpret the US military's success despite "waste" to mean that a federal government can get away with a lot
76.
by maksimum 9y ago
The algorithm as expressed in the article is simply the result of observing simple mechanical actions. We don't know whether those mechanical actions are actually what the ants are 'thinking about' when they build a bridge. A
77.
by maksimum 9y ago
FHE and DP aren't concerned with the same definition of privacy. FHE aims to guarantee exactly what is revealed. DP tries to guarantee what can be inferred from what is revealed.
78.
by maksimum 9y ago
This comment implies that these companies going out of business would be a bad thing. In the short term, definitely. But jobs are just means to an end for most people. Why can't government aim to simply provide those ends? The obvious
79.
by maksimum 9y ago
Thanks, using DP to bound deviations of arbitrary functions of a model is a neat idea. I wonder if it it makes sense going from generalization error to an estimate of something similar in spirit to (but not as strong as) differential privac
80.
by maksimum 9y ago
I've seen this limitation brought up before, but I don't have an intuitive understanding of when it would apply. Is the essential issue spacing between floats? The scenario I'm thinking in is each user's entry is a real
81.
by maksimum 9y ago
> This can also provide a simple "upper bound" on how differentialy private your model is. This is a very interesting idea. Are you aware of any resources that go through the mechanics?
82.
by maksimum 9y ago
There are high rise "settlements" all over the former Eastern Bloc countries, and none of them have become ghettos. It's social inequality, lack of access to good health care, and education that lead to ghettos; not density a
83.
by maksimum 9y ago
There was a great article posted here a few months about the internal market maker that Facebook uses to decide which ads to display. Long story short unless a particular platform is the only way to deliver ads to some group of people, the
84.
by maksimum 9y ago
The fundamental issue is that it's relatively hard for us to reasonably account for future costs against present costs. I also think it's interesting that in more social equitable countries (like the Scandis) the public and politi
85.
by maksimum 9y ago
How does accountability to the public work with career civil servants? I'm not saying it's not possible, but I think for every country like Finland where civil servants hold themselves accountable you can find 2-5 countries where
86.
by maksimum 9y ago
Very interesting point. Any chance you can recall a specific example? I'm not doubting you; simply curious since the above seems like a good idea to me and I want to understand the pitfalls.
87.
by maksimum 9y ago
> I dont think that they really understand the fundamentals (A) the fact that the analyst you're talking about is paid for their opinions and you're (most likely) not, means that with high probability you're wrong. (B) wha
88.
by maksimum 9y ago
Having made multiple presentations in beamer, it's definitely not better at everything than WYSWYG presentation editors; some things like non-standard slide layouts are more laborious than they need to be. I think the take-away is that
89.
by maksimum 9y ago
If you're trying to be economically rational you're better off simply using the GPUs to mine while you're not using them. ROI time is around 6-9 months for most GPUs these days.
90.
by maksimum 9y ago
> you don't have to remember which ones to update if you changed one, it will update automatically. Although I see how that would enable rapid iteration and make e.g. creating nice visualizations much easier, I don't think it s

More ›