boomzilla

Because we expect more from YC. No one doubts that YC has been a hugely positive impact in the startup world. It seems below YC's standard to share selective metrics that might not tell the whole story.

They deliberately avoid answering that question. The point is you feed "the machine" with all the data you can, maybe from all the sensors in every Android phone around the world, and "the machine" will tell you what philosophy is (or should be).

Yeah, that bit of information is more interesting. Did he get the access illegally? Why was that not a crime, but sending the messages was? Something is off here.

Does this mean YC is shifting from an incubator to a more full fledged VC pipeline? Or is it just a lead generator for the YC companies to meet more investors?

First I agree with you that I don't get a lot of YC startups, including the food selling machines at their current state. I do think that one of the more inefficient aspects of life is how humans handle food, from producing, transporting, storing to cooking/eating. There are just too much waste in every step of that chains. More depressingly, this is waste that doesn't necessarily go back to support the environment like the salmon leftover from the bears. Improving efficiency in any step in this chain is huge for humankind, and potentially worth a lot of money too. I don't know if the YC food startups have that goal in mind, or they are more going for the "oh this is cool" aspect.

That's an interesting angle that I've not thought of. Yes ignorance is a bliss. However my original point was more in the line of: if the chimps had a choice of getting extinct, vs the option of getting conscious enough to realize their fates are very much on the hands of a more technologically advanced race, and there is absolutely zero chance to catch up, which option would they choose?

I feel very good: they are tools and the only way they can intentionally harm us is for all of them to collectively commit suicides.

Simple economics: they get some value from having their content on Pinterest. There is no value (that they could see) from having their content scraped by other bots.

Would it be better for them to get extinct, or to become advanced enough to understand that there is a much more technologically superior race in the same planet and that their lives and deaths totally depend on that race?

It seems mysql (and hopefully postgresql sometime soon) with custom sharding logic in the app layer still hits the sweet spot for scaling to the order of 100M users. With some thoughts going into designing an appropriate data model and sharding logic, certain join queries can be delegated to the databases too.

I mostly read Reddit while not logged in, so I've never seen any latency. I guess it's mostly because of authentication/personalized content stuff, which should be easy to fix, by doing less of them or lazy loading.

Most people I know just use Google for site search by adding the site name to the query. More sophisticated queries can use site:reddit.com syntax, but usually adding reddit to a query should have the desired effect.

If any DuckDuckGo guy around here, is there a command for reddit search, like !rd or something.

What are the technology challenges is Reddit facing? Clearly not scaling as it's been pretty reliable. New features? What more feature does one need from Reddit? It's pretty good for what it's doing.

Someone seriously said: "that's not satisfactory" in an interview? That's so unnecessary. The strongest language I've ever said when interviewing is "I think there might be some perspectives that you've not considered". Interviewing is a two-way street. You'd want to hire good candidates and usually good candidates have multiple options, so you'd also want to make the candidate want to work for you too.

This depends on your current infrastructure, your expected growth over the next few years and obviously your software stack. In my experience, unless you are doing something with extreme latency or scaling requirements, AWS is likely cheaper in TCO, at least in the beginning where your fleet is less than a few hundreds machines. Also note that AWS is not just EC2. Other services like S3 or Dynamo can be life savers. The downside of AWS is that you can become over reliant of it.

Do you know if that's because Tesla is not selling enough cars, or the market demand is just not there? China used to be thought of as the big market eventually but with everything going on there, it seems that's not a good assumption anymore.

The utility of a hybrid is just much higher. You get the gas saving in city drives, and the range on the odd weekend trips. The latest Prius can be charged too: http://www.toyota.com/prius-plug-in-hybrid/ and it cost less than 50% of the cheapest Tesla. The only thing you don't get in these Prius is the look factor, though they look cool enough for many people.

The big problem with Yelp is how you make money off the reviews? Advertisement doesn't work well because people will just go to the best reviewed shops. Checkin discount, Groupon style, is too small a market, and a failed biz model anyway. So right now Yelp makes money by strong arm (yes, strong arm) restaurants into paying to remove bad reviews and/or rerank the reviews favorably.

Essentially, Yelp provides a lot of utility to the users, but can't make money off them. It provides leads for the shops, but it doesn't control the demand side. Its review data is scraped by Google, so the value of its data is diluted. I think Yelp is a failed business model by itself. The only reason I am not shorting it right now is I think someone will acquire the company soon. In the hands of a big company, it could act as a huge loss leader for other value added services.

Are things that bad at Dropbox? I know its user growth has been slowing down, but I thought it's more because they've been focusing on monetization. It does seem to me that Dropbox is a feature, not a complete product. If it goes away tomorrow, all I need to do is to move all my files from Dropbox folder to either OneDrive/GoogleDrive/AmazonDrive, and all is good.

I've seen some data sharing problem that's critical for business and I don't think Dropbox can help there. For example, we need to share (and work on) some very confidential and sizable datasets (think hundreds of GBs). After looking around a bit, we decided to go with an AWS VPC (two factor login required), and set up big Windows machine in the VPC, and everyone who need to work on the datasets have to remote desktop in. It's not ideal, but at least we can be reasonably sure that no one needs to make a copy of the data (one can not accidentally do that, and it's not trivial to do deliberately).

It really depends on the mindsets of users. My guess is that Github users tend to browse/read code in multiple repos more while Bitbucket users tend to work on their own private repos more. So it's logical that Github defaults to source tree while Bitbucket defaults to commit history.

What's the current state of team strategies implemented in this competition? Every robot going for the goal on its own? Simple passing forwards? Blocking? Double team on the ball carrier? What's the most number of intended passes in sequence that a team carries out in practice?

Thanks.

More details in Gawker's article: http://gawker.com/here-are-the-internal-documents-that-prove...

Bitcoin newbie here. How do we know that's him?

No I am not denying the value of the brand and the network, but I don't know how to valuate these assets. I don't see how they are worth 20B, but maybe some others do. In terms of the engineering platform, I am quite confident a similar platform can be built with much less than 1B. Twitter platform is not that special in terms of engineering challenges.

I am not so sure. They actually have negative earning, GAAP-wise, caused by a huge amount of stock based compensation and acquisition costs. Even with all these spending, Twitter is not showing any growth. Twitter will lose more engineering and product talents as the stock price goes down. They can't also use stock to acquire growth either. It's a dead spiral. Twitter has been leveraging Wall street goodwill and hype but I think it's the start of the end.

Believe it or not, you can freeze binaries for python apps too, if you like: https://wiki.python.org/moin/Freeze

This is a key point. If you three all started from a coffee shop, why are they co-founders and you are the first employee? Even if the other two persons worked on the company a few month before you started, after 4 years, you'd been there (and hopefully contributed) 90% of the founders time and contribution.

Startup life is hard, and it's a lot easier to sacrifice with significant ownership and a sense of purposes.

That is because the taggers are not trained on the same data. You can't expect taggers trained on wikipedia data to do well in anything but other wikipedia articles. On the other hand, if one has access to Amazon review data, (with links to the product catalog), I am pretty sure a tagger that does well on Amazon data can be trained.

Neural networks, and the plain old trusted logistic regression :) handles raw, continuous data better than the other learning algorithms. For example, if your inputs are images or audio recordings, it's really hard to do classification with decision trees or random forests as you'd need to construct the features manually. What would be a feature: color densities, color histograms, edges, corners, Haar-like, etc.? The promise of multilayer neural network is that given a lot of data, the right network structures, an appropriate learning strategy, and a huge farm of GPUs, the network can automatically learn the right features from raw data in the first layers, and utilizes the features in later layers. The big advantage of this approach is that you abstract away the domain problems (hopefully), and focus on picking the right network design, the right learning strategy, collecting a good data set etc. Neural network training is also easy to parallelize, so Google and the like can leverage their huge infrastructures.

Now if the features in the domain problem is more well defined, like credit ratings, and data is sparse, and domain expertise is available, decision trees are perfectly valid options.

Yes, and even when you run your own hardwares, it's still far easier to just KVM up your virts and "bash bootstrap.sh". For your developers, tell them to "vagrant up" with the same "bootstrap.sh". This setup and a Jenkins server for build artifacts solves all my devops needs.

The Walton family hold a bit more than 50% of Walmart shares. This is by design. As they control the board, they usually make sure that Walmart buys back enough share to keep them in control. Their income are mostly coming from dividend, which amounts to a few billion a year :) It is possible to argue that this basically prevents Walmart from taking risks in more innovative business lines.

Jeff Bezos holds less than 20% of Amazon now, but he basically "owns" Amazon for all practical purposes. I'd also say that at least 20% of Amazon valuation is based on Bezos being the CEO/President/Chairman of the Board.