barber5

8 ms·

1.
by barber5 4y ago
Users should protect themselves from this. With the expectation that they will, there will be tons of tools available and this scenario will be a nonstarter. Decentralized defenses against bots will be far more effective than the central pl
2.
by barber5 4y ago
Yes for the end user that will be the main difference. From the curation of the training data to the model itself a number of things have been put together that make the generations substantially more aesthetically pleasing imho
3.
by barber5 7y ago
Exactly. The use of ad spend to talk about Russian influence in 2016 is negligent in focusing on a metric of convenience. I suspect the real channels through which they exert influence don't lend themselves to easy quantification.
4.
Ask HN: Tips on managing/being on a remote team
9 points by barber5 7y ago|2 comments
5.
by barber5 7y ago
Thanks for posting this. I'm doing research in this area and this is fantastic
6.
by barber5 7y ago
That's because a large discussion broke out underneath a comment that may not have been the most high quality?
7.
by barber5 8y ago
This is a fair point under some ways of looking at the problem. It could be that advertising is just nasty no matter how it is implemented, and thus any system that uses advertising will be tainted. I'm sympathetic to that analysis and
8.
by barber5 8y ago
I think what he's saying is that they do the inferences from localhost, more or less (see e.g. https://www.microsoft.com/en-us/research/publication/serving... ) There are a number of proposed systems unde
9.
by barber5 8y ago
I'm not aware of any evidence that shows that directly. But there a good number of surveys that show that people don't want advertising to target them based on their interests. https://papers.ssrn.com/sol3/pap
10.
by barber5 11y ago
Nice
11.
by barber5 13y ago
It's worth saying that at a number of firms that 2% goes into the pockets of the partners. One thing to like about a16z (full disclosure, I worked there over the summer), imho, is that by using this fee to scale their organization to h
12.
by barber5 13y ago
Really, it wasn't that much better, at least not to me at the time. I'd switch search engines every few months, or more accurately have 3 or 4 search engines to deploy against difficult queries. Google was _slightly_ better than
13.
H-1B: The indentured servitude visa that big boring companies love(brbr5.co)
2 points by barber5 13y ago|0 comments
14.
Reasoning under uncertainty: philosopher or scientist?(brbr5.co)
5 points by barber5 13y ago|1 comments
15.
No, your role at a big company is not going to make you a better entrepreneur(brbr5.co)
2 points by barber5 13y ago|0 comments
16.
by barber5 14y ago
This _IS_ a solution for payments outside the USA. Do you know how many countries with a rapidly growing middle class that Visa, MasterCard, etc won't touch because their crappy credit models can't deal with these people? This is a game ch
17.
by barber5 14y ago
It probably won't actually matter but it could for some who do something similar to you, but are you sure you actually own your research and that the IP doesn't belong to the university? That's another reason not to do a PhD that was relev
18.
by barber5 14y ago
I don't know if I need this, but bravo for being crazy