rm999

It's mostly the rent but it's also Amazon.

I recently took part in a commercial real estate search in Soho, probably the biggest shopping district in NYC. The prices in this neighborhood have gone up insanely: one storefront (about the size of a large Banana Republic) recently sold for ~250M dollars, which is obviously absurd. The broker told me that a lot of these stores were taking losses on their commercial leases with the (mis)calculation that the exposure and presence were worth it. This miscalculation is everywhere, you e.g. see the same thing with banks opening locations every 5 blocks. Where the real bubble lies, IMO, is that a lot of the places are going to see competition from online and will no longer be able to justify the losses they've been taking. NYC needs bodegas, bars, and coffee shops - things we can't do online - not Bank of Americas and Banana Republics.

Yeah this was my first thought. This video feels a lot like a lesson on how not to build a great tech team.

Apple was a super-hyped company in the 80s with a very successful IPO, so hiring the best wasn't exactly difficult for them. Instead, it feels like they were filtering out people who may have brought diversity of thought and perhaps helped them avoid some of the critical mistakes that led to the failures of V1 of Apple.

I just signed up for the nytimes a couple days ago, and was exactly in the same position (I wanted articles + crosswords for myself and my SO). You want All Access which gets you two accounts with full access including crosswords (including mobile app). For the first year it's 8 dollars a month all-in, which seems like a pretty amazing deal. I think this may be a temporary sale.

https://www.nytimes.com/subscription/multiproduct/lp8XKUR.ht...

Thanks for the response!

I'm curious, what would be the interaction between Batch and Fargate? Right now I use Batch to run a container and then exit out with as little thought about the underlying machine as possible. Is Fargate a push further towards serverless?

One concern I had with Fargate from the product description is around the configuration options. Our models require more than 100GB to build, but I'm seeing "Max. 30GB".

Have you tried AWS Batch? My team moved a couple of our batch machine learning modeling jobs to it earlier this year and it's worked out great.

https://aws.amazon.com/batch/

Instead of market cap/GDP, wouldn't market cap/earnings (p/e ratio) more directly capture "market overvaluation" and "how much the stock market costs vs how much it is providing"?

If so, it seems like we're at a fairly average place for the last 30 years:

http://www.multpl.com/

https://seekingalpha.com/article/4056553-25-p-e-s-and-p-500-...

This is good advice for someone with financial security and a good salary. Which is sadly a fairly small % of people in their 20s. I'd add two higher priorities for people who don't have financial security:

1. Take stock of your finances. Track all your expenses and income for a month or two. If you find your debt is growing make changes to bring your costs down.

2. Pay off high interest debts like credit cards. Once this is complete, build up a rainy day fund of at least 3 months of expenses in a savings account. Consider this money hands-off except in emergencies.

After all this is implemented the person's finances are "stable" and they have a budget, and I think your advice is solid (but should be modified depending on salary). I'd also blend in some safer investments with the s&p 500 - perhaps using a robo advisor to manage the risk for you.

You're looking at average cost, which includes half-ton pickups and large SUVs. Look at page 6, which contains small and medium sedans.

Both Uber and Lyft cover additional liability insurance as long as the app is open. They also cover comprehensive if a passenger is in the car.

https://www.nerdwallet.com/blog/insurance/best-ridesharing-i...

My point is the fixed costs (the ones you list) don't significantly change if you already have a car and decide to become an Uber driver. Mr Mustache didn't pay 18 dollars to drive those 36 miles, he was already going to pay for his insurance and registration.

This article rests on the premise that operating a car costs a fixed 50 cents a mile. But several uber drivers have told me the economics work out pretty well for them because they already need their cars, so IMO using 50 cents a mile is an unfair assumption.

Looking at the operating cost of a car is complicated with many variables, but the standard 50 cent estimate is NOT a marginal cost, i.e. you don't actually spend an extra 50 cents to drive a mile. The annual AAA driving costs brochure puts the per mile cost (gas + maintenance) at 14 cents a mile for a small sedan, and 17 cents for a big sedan, plus ~3 cents a mile for depreciation (see "decreased depreciation"). Moving from 50 cents to 20 cents a mile brings his estimate pay from 7 dollars an hour to 12.40.

Brochure link: http://exchange.aaa.com/wp-content/uploads/2017/08/17-0013_Y...

edit: to be clear this math is if you already own a car and are driving on off-times for additional income, which (in my experience) is what a large % of uber/lyft drivers do. It's also what Mr. Mustache was doing in his test.

I think you may be underestimating the importance of this article to US politics. One of the biggest issues in the 2016 election was stagnating blue-collar wages - it was one of Trump's main platforms, and one of the main factors that made him President. If the economy continues to do well across all income levels, regardless of why, and regardless of how unpopular Trump remains, there is a good chance we will see a second Trump term. As much as I dislike him, Scott Adams made this exact argument recently: http://blog.dilbert.com/2017/10/14/low-public-approval-of-pr...

As Bill Clinton said, "it's the economy, stupid".

That's almost exactly how they found the connection, from the article:

> Early epidemiologic research showed that incidence and death rates for certain cancers were lower among individuals living in southern latitudes, where levels of sunlight exposure are relatively high, than among those living at northern latitudes

An obvious problem is confounding variables, this isn't a controlled correlation in any way. The article goes on to say:

> However, additional research based on stronger study designs is required to determine whether higher vitamin D levels are related to lower cancer incidence or death rates.

First, that's a feature, not a product.

Second, there's no barrier to entry there. Facebook's ~10000 engineers can recreate features with amazing swiftness, do it better than the first movers, then spread the feature quickly through its 1 billion+ user social network. If Facebook wants to beat snapchat at the AR game, they will do it when they feel like it.

https://techcrunch.com/2017/04/13/instagram-stories-bigger-t...

Sure, but that's not a database running in Docker, it's "external storage to store state". Which is the right way of doing things if you're ok with your data persisting on the server.

Containers are designed to be non-persistent - if an instance of a container goes down you should be able to replace it with another instance with no loss in state. Most people want their databases to be persistent (unless it's just temporary local storage for your process), so containers are a bad idea here. What you want is some sort of external storage to store state. This also enables better horizontal scaling.

> Mr. Inman [author of The Oatmeal], like many comedians seems to be fundamentally unhappy and finds an outlet through humor. Oddly enough his behavior is similar to Tesla's in that he reacts in a nasty fashion if someone challenges his ideas

Yikes, got a less biased source? It's 2017, it's sad the emotional feud between Tesla and Edison continues like this.

Full article without paywall: http://www.marketwatch.com/story/chevy-bolt-meet-the-first-p...

The story is about a near-instantaneous ~23 billion dollar drop in the market cap of Apple after reports of production slow-downs; at least that's the story several analysts are telling. Your comment is very insightful and puts this drop into perspective (i.e. it's not that big of a drop), but the reality is some major investors clearly found the news unexpected.

A few years ago I was at a failing company. The quick influx of detailed negative reviews (from people who were both still there and who had recently left) was perhaps the only public indicator that things were going poorly. I felt so terrible for all the people who were joining based on rosy promises while glassdoor reviews laid it all out.

So yeah, in many cases glassdoor is a strong signal. Don't just look at the average rating, read the reviews from the past few months and look for trends and red flags.

Sad - I believe AIM is the oldest living account where I still know my username/password. I just logged in for the first time in years and not a single other person was logged in. There were people on there I haven't spoken to in 18 years.

It's funny how 15 years feels so ancient in internet time.

>Dynamo gets expensive and it gets expensive quick,

DynamoDB's pricing scales sublinearly with volume; if it starts getting expensive it was an initial misuse of DynamoDB that got obvious with scale. There are a lot of factors that go into whether you should use DynamoDB and how you implement it. I recommend anyone who is considering using it very carefully understand this page first: http://docs.aws.amazon.com/amazondynamodb/latest/developergu...

DynamoDB is amazing for the right applications if you very carefully understand its limitations.

Last year I built out a recommendation engine for my company; it worked well, but we wanted to make it real-time (a user would get recommendations from actions they made seconds ago, instead of hours or days ago). I planned a 4-6 week project to implement this and put it into production. Long story short: I learned about DynamoDB and built it out in a day of dev time (start to finish, including learning the ins and outs of DynamoDB). The whole project was in stable production within a week. There has been zero down time, the app has seamlessly scaled up ~10x with consistently low latency, and it all costs virtually (relatively) nothing.

NYC has a great ecosystem of experienced people who know how to program, know math, and understand how businesses/product work. In this regard there are many amazing qualified data science practitioners.

But, the data science startup scene is not really thriving in NYC yet. Compared to the Bay Area, I see two main issues: a lack of data scientists with strong software engineering skills, and a lot less angel money (probably due to very few large successful exits).

I have a question that's been bugging me for a long time, ever since I worked with a couple of brilliant Canadians who informed me of the dismal offers they got before moving to the USA. Why does Canada pay tech people so little? Market effects are clearly harming Canada, with an exodus of really strong engineers from really great Canadian schools. It seems like a lose-lose for Canada and its citizens.

I'm mixed on this article. While I strongly agree that data analysis trumps math for applying machine learning, I think there's a middle ground the article is missing.

Like programming, machine learning has its 10xers, and I've worked with several. There's one thing they all have in common (beyond experience): they can think like the model. They've read a bunch of ML papers, get the intuition behind them, and envision from start to finish how data can be captured by various models. Beyond a small amount of hyperparameter optimization they often build a great model on the first or second take, and can put out production models in days (which would normally take novices weeks or even months). This requires math knowledge, because ML models are built on top of math. Even in higher level ML packages that do a lot of the work for you, novices get stuck in dead ends where their models don't work and they don't know why. The 10xers drive right through it because they know why their model isn't working.

When I interview for my team, I usually have non-junior candidates describe a ML algorithm they would use for a specific problem, then ask why certain things could go wrong - e.g. if they say they would use a logistic regression model to predict an action, I ask why the model may be returning the same score for every test case. A good candidate would understand that this means the coefficients are getting pushed to zero, and would list off reasons why this could be occurring (too much regularization, underpowered data, constant target variable, etc). You learn these things from experience, but you understand them because you know the math behind the models.

>It isn’t a forward-thinking move like Apple’s decision to kill off the floppy disk, because no one’s playing along.

In an article about how too many phone manufacturers are removing the headphone jack? A lot of people are playing along. The market for bluetooth headphones has blown up, covering a range of prices and form factors - it's absurd and perhaps wishful thinking to say this isn't a real trend. I also don't believe there are no benefits to the manufacturing process.

I predict this article will feel very outdated in a couple years. I was very skeptical of the removal of the headphone jack, but six months ago I tried out a bluetooth headset, and I would never go back. It's easier and doesn't snag, and is easier to put away. This isn't to say everyone should get rid of the headphone jack, but I don't see why manufacturers shouldn't be able to.

I build AI/ML systems for a living and it's not as big of a deal as you think. Production machine learning systems built by experienced teams will almost always have business logic protections built-in to stop the machine from doing something totally stupid. For example, an auto-bidder may have capped velocities or auto-shutoff mechanisms (those examples of bidders spending billions of dollars in minutes are exceedingly sloppy).

This is not unlike the human body's deeply-evolved reflexes - before the neocortex can even process that a pan is boiling hot, the nervous system bypasses the brain and and tells the hands to drop it.

In the case of driverless cars, DannyBee is absolutely correct: truly driverless cars will not make decisions from a single data point. I have some past experience with unmanned aerial vehicles and those things had insane amounts of redundancy in almost every signal and decision making process; driverless cars are more complex and will come decades later and will probably take this even further.

The calculus answer is that gradient descent approximates the shape of your optimization space with the first order taylor expansion at your currently optimized point (first derivative information), and Newton's method approximates it with the second order taylor expansion (first and second derivative information).

The geometric answer is that gradient descent approximates the shape of your currently optimized point by a plane that is tangent to the current point, whereas Newton's approximates it with a quadratic "bowl" using the local curvature. In gradient descent you have to take short steps down the path, because you know your approximation cannot be true (you will not always be able to lower your error by moving in the same direction, if you assume there is some minimum error). In Newton's method you can assume that your bowl approximation is correct and go straight to the bottom of that bowl. If your space really does look like a bowl, you will get to the bottom super quickly.

The ELI5 answer: you are a ship captain looking for the lowest point in a part of the ocean. You cannot afford to send a submersible down, so your approach is to send an anchor down to check the depth and take a panoramic photo a few feet around the anchor. After analyzing the results of each anchor drop you can move your ship anywhere and try again. In gradient descent you look for the lowest point on the edge of the camera's vision and move in that direction. You can move as far as you'd like, but you will not go too far because you know the drop will not go on forever. In Newton's method you will look at the curvature in the pano and extrapolate it out assuming it is a "bowl" shape. For example maybe your anchor is on an underwater J-shaped cliff getting less steep as it goes north, so you map it out as a U shape that starts going up again if you go a mile north. So, you move your ship a mile north and try again there. In gradient descent you would have moved maybe 100 feet north then checked again. Newton's method can intuitively get you to the low point in very few anchor readings, but crunching the numbers of the bowl can be very time-consuming (this becomes relevant in higher dimensions). In this analogy, the long and lat coordinates in the ocean are your two parameters to be optimized, and the height of the ocean floor is the error in your optimization. You want to find the parameters that give you the lowest error.

For me, productivity is 90% about existing in a time and place without blockers. You can have all the right ideas (and even the least productive people I know are full of great ideas) but a single blocker can take your productivity from 100 to 0 indefinitely.

* For programming, this can be knowing a language very well, having smart coworkers who can help with issues, and a plan of how your software will look in the end.

* For getting in shape it can be living near your gym and going to bed at 10:30 instead of 11.

* For writing it can be about turning off message notifications and keeping an outline to prevent writer's block (honestly I'm still working on this one).

I can go on, but this realization has been tremendously useful to me, and lets me exist in "10x productivity mode" many times longer than in the past. While I don't agree with his politics, I strongly believe Scott Adams' ideas around systems vs goals is the right way of thinking of life. You shouldn't expend energy achieving specific goals, you should set your life up so it takes minimal effort to be productive and live a good life. http://blog.dilbert.com/post/102964992706/goals-vs-systems