Fluent Forever (2019 Book #2)

Summary: Fluent Forever is a book about how to effectively learn new languages quicker. My review is that - it’s a good book, but I wouldn’t recommend. One thing that would be an interesting metric is the number of times you will re-read the book. Some books are so good that you need to reread it. This book’s reading count is 1. I would not recommend a buy unless you need a light read before bed time or need a break from other things. The book should be cut to 50 pages.

Naval Ravikant warns about books of this type. Books that talk about the same thing until cows come home. Effectively, it’s a book that can be summarized into one sentence. Here it is: use your visual system to learn the language.

Key Takeaways:

  • Learn Pronunciation First
  • Don’t Translate
  • Use Spaced Repetition Systems(Anki)
  • Never settle for safe when you can have fun instead - something to think about.
  • More accurately you learn pronunciation, the better you remember it.
  • ** Don’t use 1-1 word to word flashcards. Try to put an image and even better, if using computerized SRS systems, use sound as well.
  • Recalling is more effective with learning than reviewing.
  • Take advantage of the forgetting curve. There is an optimal point of forgetting, and you should review when you are just about to forget.
  • Whenever you are under stress/performing, your brain will act in gear. Put yourself in test/rating mode.
  • Images are more powerful than words. Hmm. But don’t we not see until much later? Are sounds more powerful than images? I found out from an HCI book that vision accounts for 80% of information content people absorb - so yes! This is true!
  • ** You need to look at mouth positions, and also try to emulate the mouth position when you are saying words.
  • Try to have a list of most frequent words and study them first.
  • The minute you get bored, your mental filters come back. You need to be able to spice learning up, to not get bored. This is very important.
  • Masculine nouns - fire. Feminine sounds - exploding. Assign attributes to gendered pronouns.
  • The fastest way to learn another language, is to ignore all other languages and absorb an insane amount of the language you’re trying to learn at the maximal rate. At least, that’s what kids did right?
  • You should use a grammar book as a shortcut to train yourself on how grammar works before diving deep into other territories.
  • Google Images is your friend.
  • Need around 15,000 - 35,000 words to be fluent in a language. That’s around 41 words per day.
    • Question 1: I wonder what the range of words that people use in a given country would look like? Would there be a correlation between the number of words used to gdp?
    • Question 2: One experiment I’d like to do is go nuts in one language. So study Japanese for 1 month, but you don’t speak English to anyone else unless you really have to. No english music. No english text. Only Japanese. You can’t think in English either. Japanese only. So you want to optimize just to learn the language I also wonder how many words you can memorize per day. You need 30k - 45k words to be fluent in a language. But how many words can people memorize per day? If you want to memorize a thousand per day, that’s 42 per hour. Maybe this is doable if you practice?!

Good Resources to Learn the Language:

Lang-8.com

Italki.com - conversation practice site.

** forvo.com -

Verbling.com

LiveMocha.com

MyLanguageExchange.com

Book Finished May 2nd, 2019.

Voice Lesson Notes

During the past month, I took 4 group voice lessons. It was a fascinating experience!

Here is how the format worked. We came in, and we sang the songs that we prepared. The instructor aka Mike would give us things to work on, and we’d take notes, practice, and bring our song back for next class.

Singing is hard. It’s easy to make fun of people on audition shows who flop, but man - it’s HARD. I practiced for a month and I have to say, I suck. Here are some of the things I need to work on:

REACH - I get hammered on this one. I don’t speak up. I was constantly reminded that I have to sing like I’m singing to the back of the audience.

Breathe control - Need to know when to breathe.

Dynamics - Lot of the music gets power from sudden chance in style, whether that be volume, tempo, etc. So when you’re going up a few notes, you need to emphasize the transitions.

Theme - There’s usually a theme or a message in one song. Usually the best movies are like that too. It’s important to focus on the theme or message of the song when you sing. I tend to focus more on the music and not as much on the mood or the lyrics.

Consistency - This is what I struggle with in general. I’m more a bulk kind of person, where I have sessions of intense focus. But I don’t think music practice works like that. Better to do 10 minutes a day than have 1 long session. Week leading up to my last class, I wasn’t able to practice as much. Yeah, ultimately they’re excuses. I went in to class on the last day. Mike heard me sing, and then he glared at me and said, “You’re back at where you were last week. You’ve been a very bad man.” =/.

Focus - If you try to fix multiple issues at once, you become super scatterbrained. Work on one issue as you sing. Similarly, focus on parts that you’re messing up. Because if you’re singing the entire song - especially the parts that you can nail, you’re wasting your time. Also, if your mind wonders when you sing, then the consistency of the sound becomes weak at certain times of the song. What should be happening, in my opinion, is you should be focusing on your vocal cords so that they always resonate.

Pitch Deafness - I don’t know. I haven’t been criticized for missing pitch, but it’s more on an intuitive level. If someone told me to start with an F, I might fumble around with a piano. Something to work on. There should be an app to help with this.

The songs that I worked on were:

Moonriver - https://www.youtube.com/watch?v=mZ-KPwD1eno

Stranger’s in the Night - https://www.youtube.com/watch?v=h5h_EW4odWw

Fools Rush in - https://www.youtube.com/watch?v=lBAkOQtuWMk

You are so Beautiful - https://www.youtube.com/watch?v=E7VkSGqXdUU

It was a great time! But I decided not to continue for now due to other priorities that I want to focus on, and uncertainty in my living situations. But if I were to do it again, I would just focus on private lessons. Because marginal utility -.-.

Week specific notes:

Week 1:

No notes taken. First time.

Week 2:

Don’t get sentimental/lose focus.

Enunciate.

Try to pretend you’re singing to someone.

Focus on the voice.

Make one song perfect, incorporate the techniques into next song.

Overpronouncing vs. underpronouncing.

Week 3:

Voice should be crisp, not soft and flabby.

Focus on the voice.

Side note: Are uttered words without feelings and truth lies?

Get the thought, then sing.

Be on the look out for drop out.

Higher pitch -> happier. Lower pitch - sadder.

Work on one thing at a time.

Mike told me to drill a hole through a wall with my voice XD.

More mouth movement.

Always focus on throat.

Think character voice.

Emphasis on the word, also emphasis on vowels. Sometimes you want to each vowels faster in the word.

If you get hoarse, rest.

some parts need to be sang softer, some parts need to sing with more force.

my mouth needs to move more.

Week 4:

Tension/pressure ruins singers. This is the biggest problem singers face.

Concentrate on the beauty of the voice.

Focus on transitions.

A new person came to check out the class. When she made a mistake, she laughed. I don’t know - people who do that are amazing. There’s something so charming about that, and it just dusts off the seriousness of a mistake.

Gym Over

I was at the gym yesterday - the gym closes at 10 PM. I got in around 9:20 PM, and did a 3 mile run. I don’t know what it is for me, but I feel like I go through a cleansing every time I go on a run. It’s great.

But that’s not the story. I was at the gym, I go through the shower and everything. I’m in a rush to get out because I feel sorry that I’m making the workers stay late because of me. I run out.

The minute I leave the gym doors I check my bag for all my clothes, and I realize that I left my underwear and shorts in the shower. The doors lock after closing time. I screwed up.

So I’m hoping the store managers would come back to the front desk near the door.

They don’t.

I time-box myself to wait 10 minutes. They don’t show up in the front. I start my walk back home, cursing underneath my breath to have slowed things down and double checked rather than rushing out, because… the workers also need to go home! It’s been a long day for them. The clothes I left may or may not be there tomorrow - the cleaner may just throw out my dirty clothes. I doubt there is a lost and found where they keep all dirty gym clothes that are festering with germs and sweat.

At this point, I realize two things.

  1. There is a backdoor to the gym. That’s why the employees never came back out. They must have a back entrance they prefer. This can mean a lot of things - I really think that there is a backdoor for almost everything. When going to a concert, trying to meet someone you really admire, or even a job application. There is definitely a backdoor somewhere to meet them. You just have to find it and crack it open.
  2. Once the door shut behind me, gym time was done. No turning back. It’s over. Done. Game Over. Gym Over. There’s no point for me to wait there until the cows come home. What’s done is done, and it’s time to move forward. What I’ve been struggling with is the fact that during the past few months, I never took initiative and got to know someone I was interested in until they left. There’s been a lot of regrets on that.

    But what’s done is done. There’s a reason why we don’t have eyes on the back of our heads. We move forward. We don’t let what’s behind us bother us for every step we take. Every second we spend looking at the past, we sacrifice seconds that could be spent moving forward.

    Time to move forward!

P.S. I went back to the gym the following day. Fortunately, the clothes were on top of the shower curtains, and the cleaning crew probably failed to notice. I got my clothes back!!! =D.

First drafted May 11, 2019.

Data Analysis on Restaurants in Downtown Brooklyn

During my time in downtown Brooklyn, one of the things that drove me mad was looking at restaurant reviews. Friends and I would decide to go to restaurants, and we’d look at the ratings. When we did, we’d get an aggregate rating of 3.5 or 4. But what does it all mean? It really doesn’t mean anything unless we understand the distribution of the data.

It drove me nuts. I had to find out the average, and what the review landscape look like.

So I did. I cobbled together a program in Python using Scrapy, Pandas, and Matplotlib. I would have left out Sir Scrapy, but the review website’s api had this random feature where if you queried a restaurant for all its reviews, it would give you 3. Grr…

Process:

I used the a certain restaurant review website’s own filters to hone in on downtown Brooklyn within a mile radius. The website’s api request gave me two two longitude latitude pairs. They are shown below:

Interesting. So it’s either endpoints of a circle, or a rectangle. I think it’s a rectangle, so we’re going to refer to this area as the rectangle.

This search query gave me a starting point of all restaurants, but displayed only 10 or 20. I scraped them, and then used a program to move to the next page. I stored the set of restaurants in a directory. Then for every restaurant, I automated a http request, got to the main webpage which was an anchor point of a restaurants set of reviews. So I went through all those reviews and got them too.

Results:

Number of restaurants in the bounding box: ~500.

Number of restaurant reviews: ~70,000.

The average of all review ratings: 3.70

Standard deviation of all ratings: 1.32

Graphs:

So people in Brooklyn tend to rate 4s and 5s much more often than 1s and 2s. I wonder if you do this for every city, then could you gauge a friendliness metric for every city, and see if it correlates with the happiness ranking of every country? That would be AMAZING.

This is if you plot average rating of restaurants, and plot them by rounding them to the nearest 0.5.

This is simply the number of reviews and their restaurant names. Seems like it follows some power law distribution, but I’m not quite sure. It may be just one outlier.

Top 20 restaurants by review count:

I guess this can be interpreted as popularity or to some degree how much people care. One can perhaps use this information to extrapolate the length of time the restaurants have been around.

Restaurant Number of Reviews
Grimaldi’s Pizzeria 4440
Juliana’s Pizza 1955
Junior’s Restaurant 1573
Joya 1259
Rocco’s Tacos & Tequila Bar 1196
The River Café 1038
Habana Outpost 962
Shake Shack 788
Mile End Delicatessen Brooklyn 744
Clover Club 728
Yaso Tangbao 703
Ki Sushi 671
Hanco’s 658
Vinegar Hill House 634
Alamo Drafthouse Cinema Downtown Brooklyn 621
Forno Rosso 604
Two 8 Two Bar & Burger 603
Dekalb Market Hall 589
Bedouin Tent 586
Sottocasa Pizzeria 586

Top 50 Highest Restaurants by review average (ignore just 5s – probably insufficient number of reviews. Perhaps I will add review count next to the rating later…)

Restaurant Average Rating
Moshman Dental 5
Pipitone’s Pizza 5
Bird’s Eye Vietnamese 5
VALENTINE’S CAFE 5
First Wok 5
New Fresco Tortilla Plus 5
Smith Gourmet Deli 4.952380952
Thai on Wheels 4.888888889
Lillo Cucina Italiana 4.870588235
GMC Temaxcal Deli & Grocery 4.833333333
Simple NYC-Downtown Brooklyn 4.80952381
Yumpling Food Truck 4.790697674
Cafe Gitane 4.75
Sunny Delicatessen 4.75
Pret A Manger 4.666666667
Ashland 4.625
Govinda’s Vegetarian 4.621118012
Dariush Persian Cuisine 4.580645161
Grand Canyon Restaurant 4.577777778
dot & line 4.52173913
Rice & Miso 4.519230769
dumboLUV 4.5
Kazi Halal 4.5
Saint Julivert 4.5
Yemen Cafe & Restaurant 4.452173913
E-bite 4.444444444
Sanpanino 4.444444444
ACE Thai Kitchen 4.414634146
Sushi Gallery 4.4140625
Bread & Spread 4.412698413
Lavatera Grill 4.409090909
Forcella Fried Pizza 4.407407407
Chicks Isan 4.404761905
Doner Kebab NYC 4.401360544
Mr. Fulton 4.4
W XYZ Bar 4.4
Yossi’s Cart 4.4
Juliana’s Pizza 4.396930946
Shawarma & Grill 4.375
Makina Cafe 4.375
Koji Izakaya 4.358974359
Daigo Handroll Bar 4.333333333
Metro Buffet 4.333333333
Warung Roadside 4.333333333
Taiki 4.327586207
Sultan Restaurant & Cafe Lounge 4.3125
Espresso Me 4.306122449
Piz-zetta 4.305220884
Downtown Natural Market 4.304347826
Sottocasa Pizzeria 4.298634812

To-do list:

  • I need to ask someone who is really knowledgable in statistics, or do research on if there’s a law that correlates the number of reviews to its true rating. What should it converge on? For example, a review with 1000 reviews with a 3.5 should be weighted differently with a review with 10 reviews with a 3.5.
  • It would be interesting to plot the average words per review or aggregate words per restaurant.
  • What would be the most commonly used and the most least commonly used words? I want to run it through a basic NLP program to stem the word, remove stop words, etc.
  • What if we have a list of words, give them a raw numerical value of positive and negative numbers, and average that out. Would the rankings be different?
  • What if you plot out the datestamps of the reviews, and use some metric for happiness/economic activity, and see if there’s a correlation between that and the stock market or the consumer sentiment index? Would there be a correlation?
  • See what another review service from a search engine’s data looks like.

Examining all this data is a lot of fun! But for now, my experiments are on hiatus. There’s already enough things to read and build! But if anyone wants some Brooklyn data, let me know!

Tips:

  1. You should probably use this dataset to get interesting insights: https://www.yelp.com/dataset/download
  2. DO NOT SIMPLY SCRAPE ON YOUR LOCAL NETWORK. If you mess up, it’ll cause your ip to be banned. Your ISP will refresh your ip within a month, but it’ll annoy the people who share the internet connection with you. Either use a vpn, or spin up a cloud computing instance and use that to download your data.
  3. Rate limit your scraper. Don’t try to download all this data within a span of a minute. Pace your scraper to download a reasonable amount in a minute so you don’t overload their apis. Or get caught. There is no need to rush.
  4. You should learn regular expressions. I was talking to someone who was doing data collecting with DOM traversal. That maybe more painful than building a reasonable regular expression. My two cents.

I Know Nothing

Humbled. Today is the 11th day of June of the year 2019. I’ve felt this somber melancholy that I really don’t know anything. Compared to the vast knowledge that have been built, and will be built, I know so little. There’s no need to be giddy that you’ve solved problem X, when it’s one among the infinite ones out there. Similarly, maybe I shouldn’t be extremely stressed out about mistakes and mess ups, when those actions are cosmic dust in a spectrum of all available decision. I reflect back on the younger days, when I thought I knew more, and I was wiser than someone else. There’s some shame in that. I feel like an idiot for ever thinking that way.

You think you understand, then you take a second look. And you’ve realized that you’ve never really internalized nor grasped. Whether that be a simple story or a math equation, I feel that there is always deeper meaning underneath it all. It’s like scratching the surface of an iceberg. You can keep going down forever.

It’s a two sided coin. On one hand, you will never run out of things to learn, explore, and think about. How exciting! But on the other hand, you will never learn everything. Impossible.

Other things I struggle with:

  1. You can call someone out for their bs, but I find myself in similar situations and under the same conditions I can easily make the same mistakes. I.e. criticizing someone for not willing to do something then a week later I find myself not willing to do something easy for a stranger because I was in such a terrible mood.
  2. It’s hard to strike up conversation with people when you know that the relationship will never be deep. What’s the point of striking up a conversation with a person at the subway? You may never see them again. But then again, you may never know where conversations end up. They may offer you a book recommendation, or something super insightful and change the way you think. But you don’t want to annoy them either.
  3. Making good choices. For every decision in the decision space, there is a theoretical best decision. For yourself, there is a theoretical best decision that you can take at every time t. I want to take those optimal decision, or decisions close to them. This is hard. The big recommendation I keep hearing is to study philosophy deeply. It has characteristics that cannot be defined my structured math and logic.
  4. Becoming a more efficient coder. One of the things I’ve realized I need to work on, still, is coding a little and testing a little. Knock on your steps before going to deep. Dipping your toe before jumping into the ocean. Not digging myself into a cobble-web of complexity.
  5. Becoming a better communicator. Lately, I feel like I’ve had a string of.run on sentences. People can’t keep track. I should cut, verify they understand, cut, and perhaps get more feedback. Usually shorter sentences are better than longer ones Ugh. But at the same time, it annoys people if you keep asking for feedback. Maybe the better approach is to study people’s so I can detect confusion.

Drafted June 11th, revised June 13th.

Site built with http://lanyon.getpoole.com/ template. Thanks to @mdo for the original template!