imh

31 ms·

241.
by imh 9y ago
Human computation! Not to be too #hailcorporate, but my job's pretty fun. It's using machine learning and humans together to solve problems. The fields pretty focused on labelling tasks, but there's much more to it. Having a
242.
by imh 9y ago
Why do you do that?
243.
by imh 9y ago
People here are reacting to the headline. The point doesn't seem to be that the best idea does win, but that the best idea should win. It's about listening to other people's ideas. >As Charlie Munger once said, “I beli
244.
by imh 9y ago
Is it? I don't mean that facetiously. If you and I are both selling the same product and you spend more on marketing, does that mean you can raise your price that much more, or is it just a little, where the primary spending comes at t
245.
by imh 9y ago
I once worked at a company that was having trouble finding women candidates. We were a tiny company that at that point largely recruited out of the people we knew, who were largely people we'd worked with before and people we went to s
246.
by imh 9y ago
D seems like a cool language. Do many people use it?
247.
by imh 9y ago
I'm thinking about this as less of a newsfeed thing and more as pub/sub maybe being a bad model for keeping in touch with friends. Especially when it's curated pub/sub.
248.
by imh 9y ago
One practical point anyone implementing bandits should be aware of is time correlation. Bandits are awesome if your data behaves the same going forward as it did before. But when you change your A/B split on Friday based on what you le
249.
by imh 9y ago
I agree that there are some oversteps in the current atmosphere (e.g. equating asking someone out for drinks with harassment [0]), but this doesn't seem like that kind of case at all. >One night after the conference talks were over,
250.
by imh 9y ago
For everyone who seems to be skeptical of this article or skeptical of overreaction, it's just a shitty article. Read the blog post it's reporting on. It's much more damning than 'he made off color jokes' or 'h
251.
by imh 9y ago
Partially, this is just bloomberg doing shitty reporting. This is a poor article about a blog post. The blog post itself is much more damning in my eyes https://medium.com/@kristianlum/statistics-we-have-a-problem... F
252.
by imh 9y ago
The word "bankrupt" is only present in the title :-\
253.
by imh 9y ago
I think it's a an practicality thing. Getting these systems powerful seems sorta solved, but getting them easy to use seems tricky. AFAIK it's similar to computer formalized proofs. We can do them, but it's super verbose and
254.
by imh 9y ago
My recommendation depends on what you'd like to do with it/what problems you are trying to solve. Huge field and all. Whatcha looking for? One resource I've started recommending to people is this little book http:/&#x2F
255.
by imh 9y ago
I'm sorry, but this is entirely unfair, especially points 1 and 3. First, addressing #3: >People who know statistics can't interpret results to laymen. This doesn't match my experience. Stats experts can sometimes be terri
256.
by imh 9y ago
I had a lyft ride in one last weekend and chatted with the driver about it. I agree it looks nice. The interior was great and the ride was comfy. The driver was very enthusiastic about how much he loved the car. Plug it in over night, drive
257.
by imh 9y ago
Whenever I feel substantially undervalued. For my first few jobs, that was relatively often (1.5-2 years) since most companies will offer tiny raises, while your skill growth will be raising your value much more substantially. My first few
258.
by imh 9y ago
That's subjective happiness, aka "Taking all things together, how would you say things are these days — would you say you’re very happy, fairly happy, or not too happy these days?” In the discussion section of the paper [0] that f
259.
by imh 9y ago
https://www.politico.com/story/2017/11/21/fcc-net-neutrality... Not only are they removing federal rules, they want to prevent states from making their own rules. That seems an extra special kind of awfu
260.
by imh 9y ago
Study higher level math! Category theory is a brainfuck that exercises that skill well. This book https://mitpress.mit.edu/books/category-theory-sciences doesn't have any real mathy prereqs but is fun. Added benef
261.
by imh 9y ago
If this is any indication, traditional venture capital seems to be doing pretty well: https://www.crunchbase.com/lists/recent-funding-rounds/f535c... And those are mostly just rounds raised today.
262.
by imh 9y ago
Yeah that's totally true, but I feel strangely about it, like treating the YIMBY's as a force of nature takes the morality out of it. If I owned anything else, like a bike, then it's mine because the law says you can't t
263.
by imh 9y ago
You did buy it if the law/zoning said it would stay that way. Hence people getting worked up when people talk about allowing big buildings where they currently aren't allowed.
264.
by imh 9y ago
When I left school, I found myself in a similar boat, and decided to set a goal of getting myself the knowledge equivalent of an undergrad degree in math. I already had a physics degree under my belt, so it wasn't as long of a path as
265.
by imh 9y ago
What do you think the mechanism is on the connection between ads and pageviews? Or the mechanism for the autocorrelation of pageviews? Have you been able connect the dots on something like sharing to sites like HN?
266.
by imh 9y ago
For myself and a number of friends, Community's D&D episodes were the cause. But the episodes, like Stranger Things, were probably a symptom of larger trends.
267.
by imh 9y ago
Well that ad's great. It has a footnote giving caveats to the very first sentence!
268.
by imh 9y ago
One controversy that has settled into the foundations of math was over what it means to use infinite sets, and comparing their cardinalities. I was unsettled when I first encountered Cantor's diagonal argument to prove the reals are la
269.
by imh 9y ago
"there is a 5% chance of the data being noise" is a statement on the probability of whether the phenomenon is not real given the data: p(it's just noise | data). A p value is a statement on the probability we'd see this
270.
by imh 9y ago
>The top funds can return a rate of over 20% while the median is close to 10% and the bottom venture capital firms may produce a negative return. If startups were all just lottery tickets then there wouldn’t be such a consistently differ

More ›