His Youtube videos are gold. This one, in which he aims to take the imprecision of floating point numbers to extreme applications, such as training neural networks with linear activation functions or even implementing cryptologically-safe functions, is superb.
The first of those 12 is Emperador, a city of 692 people, so you have 692 people living very densely. Meanwhile, Paris, two ranks down, has more than 2 million people living at basically the same density. You'd need to account for that. The arbitrary nature of municipal and regional boundaries has always been the bane of comparisons of population density.
You could weigh the density by population (effectively giving you population²/area?! I'm not saying this is a good idea), and you'd get a top10 of Paris, Barcelona, Madrid, Bucharest, Berlin, Athens, Milan, Brussels, Vienna, Naples, which despite the slightly bizarre metric seems a more sensible ranking (Emperador is at the bottom rank), and which, to be fair, also features two Spanish cities.
But again, it's kind of a pointless endeavor, because of the arbitrary nature of the boundaries chosen -- why Paris and not Paris metro? etc. I guess ideally you'd have a function density(person) giving you the population density of any given person and you'd want to look at the distribution of that function, specifically the average per country of that function.
Hey! I work on the ML Feature Infra at Netflix, operating a similar system to Chronon but with some crucial differences. What other alternatives aside from Starrocks did you evaluate as potential replacements prior to building Chronon? Curious if you got to try Tecton or Materialize.com.
We haven’t tried materialize - IIUC materialized is pure kappa. Since we need to correct upstream data errors and forget selective data(GDPR) automatically - we need a lambda system.
Tecton, we evaluated, but decided that the time-travel strategy wasn’t scalable for our needs at the time.
A philosophical difference with tecton is that, we believe the compute primitives (aggregation and enrichment) need to be composable. We don’t have a FeatureSet or a TrainingSet for that reason - we instead have GroupBy and Join.
This enables chaining or composition to handle normalization (think 3NF) / star-schema in the warehouse.
Side benefit is that, non ml use-cases are able to leverage functionality within Chronon.
FeatureSets are mutable data and TrainingSets are consistent snapshots of feature data (from FeatureSets). I fail to see what that has to do with composability.
Join is still available for FeatureSets to enable composable feature views - join is resuse of feature data. GroupBy is just an aggregation in a feature pipeline, not sure your point here.
You can still do star schema (and even snowflake schema if you have the right abstractions).
Normalization is a model-dependent transformation and happens after the feature store - needs to be consistent between training and inference pipelines.
> Unemployment rate tends to hover above 20% and I think it has the highest rate of students dropping out of basic education.
The first claim is true, but there's a lot of under-reported economic activity, and a lot of fraud (people claiming they don't work when they actually do), so the unemployment rate figure can't be trusted.
Your second claim around the highest rate of students dropping out of basic education needs a citation.
> but there's a lot of under-reported economic activity,
I'm sorry, but that's just hand-waving over the problem to pretend it isn't there.
> Your second claim around the highest rate of students dropping out of basic education needs a citation.
I learned that fact because the Spanish Government posts tv ad campaigns on how Spain has the highest rate if students dropping out of basic education, and how they should instead explore options such as pursuing trade education.
If you really are interested on the topic, you could start by checking what OECD has to say about Spain's education problem, including how 27% of its 25-34 yo cohort (i.e., the latest generation entering the workforce) did not complete basic studies and does not have postsecondary education. Spain also boasts a NEET rate of 17%.
It's also noteworthy that Spain's horeca industry started to rely on heavy influxes of immigrants from South America to meet their demand, in spite of Spain having one of the highest unemployment rates in Europe.
Keep in mind Barcelona and Madrid are some of the densest cities in the whole Europe. It's one of the reasons why COVID took such a big hit at the beginning of the pandemic.
Barcelona is pretty aerial though. I’ve been there twice in my life and I’m very impressed by its original urbanism. At first sight by looking at a map it looks really American inspired but once you dig a little, it’s in fact a unique "modular" urbanism (of course I’m not talking about the historical city center, that’s another interesting topic).
What amazed me is how that city manages to be so densely populated but is designed so you never feel packed. My only "complaint" would be that there could be more trees but there were still more than in most French cities (which I’m from) so …
I'm a Spaniard, and what I've noticed is that many people in the U.S. have a limited understanding of what food in Southern Europe is genuinely like. They often overlook how it excels in areas that matter to them: health benefits (thanks to the Mediterranean diet), quality of produce (including vegetables, cheese, meats, fish, fruits, and nuts), and cooking techniques that result in unique and varied flavor profiles.
In Southern Europe, food is about enriching your life, not merely impressing your palate. People's lives often revolve around food, and they are discerning critics. Not only have they tasted a wide variety of dishes, but they've also cooked many of them at home. They're constantly sharing tips about which restaurants to try, which markets to shop at, which Spanish chefs to follow, and even which places to visit primarily for the food. Social circles frequently gather on weekends in restaurants, bars, or homes where food is considered another guest at the table. They critique dishes, challenge each other to improve, and often decide to cook themselves, confident they can do it better. As a general rule, the culinary skills found in an average Spanish home surpass what you'd typically find in an average American restaurant.
It's also a mistake to speak of French and Spanish food as if they're a monolithic culinary style. Both Spain and France, often cited as the powerhouses of modern cuisine (for an authoritative source, check "Modernist Cuisine" by Nathan Myhrvold), boast multiple, diverse, and rich regional cuisines. These regional cuisines are as different from each other as the various regional foods in India. While there are common elements—like the use of olive oil, a focus on seasonal ingredients, a philosophy of highlighting the essence of each dish, and similar proportions of carbs, fats, and proteins—each region has its unique flavors and dishes.
So why do Americans have such a skewed perception of life and food in Southern Europe? One significant factor is the lack of immigration from Spain and France, except for Italians who have had a too much of a strong influence on American perceptions of their cuisine. Cultural and economic factors likely contribute to this. Spaniards and French people value their family and quality of life highly, making them less inclined to move to a country where they might earn more but live less comfortably. This is particularly true for Spaniards, many of whom return to Spain if they do emigrate for professional or survival reasons.
This lack of immigration also explains the scarcity of authentic Spanish restaurants in the U.S. There are only a few, and most that do exist are of subpar quality. Even in places with a higher concentration of immigrants and focus on fine dining, like Miami or Chicago, the quality often falls short. Another issue is the difficulty in finding skilled cooks in the U.S. who can handle the complexities of Mediterranean dishes. Many lack the necessary training (as they are often immigrants, at least in Florida/California), and those who are trained are often accustomed to a different style of cuisine. They haven't developed a palate and understanding of the cuisine. This is not an invention of mine; I've recently talked to a Spanish chef opening a restaurant in Chicago who had to change his whole menu because his staff would be unable to learn how to properly cook it.
The same issues largely apply to French cuisine in the U.S., which brings me to your point that French cuisine focuses on impressing the customer. Traditionally, French cuisine has always been associated with fine dining, and especially so in the U.S. I've always found this to be a parody; Americans are often bewitched by French culture and elitism, it's a self-reinforced belief.
Spanish and French high-end restaurants in America often feature Spanish and French dishes, but they focus more on impressing diners with innovation and surprise than on authenticity, often at the cost of flavor. Running such a restaurant in the U.S. is expensive, often requiring chefs to be brought in from Spain or France, which in turn requires a focus on "impressive" menus to justify the costs, and especially to convince chefs to immigrate to the U.S. There's also a demand for such restaurants, as Americans want to buy in into the luxury of dining at "great" places, and they want to attract business gatherings. Ultimately, any food connoisseur will tell you these restaurants are mediocre. They are focused on making you feel special, not in feeding you stellar food, and they are to be avoided.
The interesting paradox is that if you look at the the most awarded restaurants in the world, most of them outside the US, you'll find the top lists to be dominated by Spanish, French, or at the very least Mediterranean-inspired or Mediterreanean-instructed chefs (like René Redzepi from Noma in Norway, trained at El Bulli). Fine-dining experiences in Europe are significantly different than in the US.
I don't want to argue about which cuisine is the best; I appreciate great frood from anywhere in the world, and personally I truly enjoy the food from Japan, Peru, and China. However, I find that your views on Spanish food seem not only highly biased but also reflective of a broader American perspective that I consider flawed and self-centered.
I encourage you to visit different regions of Spain to experience the food as locals do - you'll need at least three weeks to truly enjoy and experience it. If you need recommendations, let me know and I'll happily share my personal list. And, if you're ever in the Bay Area, I'd be more than happy to share authentic Catalan and Valencian dishes with you, including a true paella.
Ultimately, take my word for it. The Spanish food you can find in the US is a caricature of what it truly is. It somehow pains me how inadequate the American perception of Spanish food is, given the merits it actually has.
This is the secret underbelly to the car-centric design of the US. People drive illegally all the time. They drive over legal BAC limits, they drive without insurance, they drive unlicensed, they don't pay parking tickets, they drive looking down at their phones and not at the road.
When you're poor and you live in an area completely unserved by public transit and you lose your license because you can't afford to pay parking tickets, are you really going to stop driving and lose your job and become homeless?
We have statistics to show what unlicensed and uninsured driver crash and fatality rates are like and they're a lot higher than the rest of the cohort, but there's still a sizable part of the US population that does all of these things and still uses the same public road infrastructure as everyone else, often out of lack of alternatives.
And to get your car registered in most states, you usually only have to pass an emissions test, have a valid license, and have proof of insurance at the time that you register the car.
This means that 11 out of 12 months, you get to drive around without insurance.
Pretty much illegal everywhere in the US except for a few weird outliers. I think there’s one southern state that lets you have a bond instead of insurance?
> There's a lot of research out there already that the OP doesn't seem to be aware of.
You don't need to undermine someone's position to add more context. I like the additional context you add, but I believe you'd be better off without the judgment. It's unwarranted, and being an avid follower of the OP's blog, I'm pretty sure the author knows these facts about procrastination.
reply