2018-02-10.links

10 Feb 2018

Stopping childhood lead exposure responsible for a 56% drop in violent crime in the 1990s. Link.

Despite what they say, men and women have a strong automatic preference for monogamy. Link.

This study used the implicit-association test, which is somewhat hard to fake, but not impossible.

How mangoes became popular in China. Choice quote:

Alfreda Murck writes that when a mango celebration came to a small Fulin village a local dentist didn’t see what was so special. He exclaimed that it just looked like a sweet potato and, for his insolence, “he was arrested as a counterrevolutionary.” The man was found guilty and executed.

Are we in a bubble that’s about to burst? I don’t know, and after reading this, I still don’t know.

NOTHING IS SAFE.

I just learned about vDSOs (virtual dynamic shared objects). Basically, relatively safe kernel functionality like gettimeofday() can bypass the whole syscall context switching by using a shared library. (If the kernel doesn’t have vDSO support it will fallback to a traditional syscall).

2018-02-03.links

03 Feb 2018

Baremetrics shows their own real stats in their demo.

Recovering audio from recording the vibrations of objects.

The history of economic growth.

Mike Bostock, creator of D3.js, launches notebooks for Javascript.

What do you know. The basic tenets of economics works: building more housing in Portland lowers rent costs.

2018-01-26.links

26 Jan 2018

I’ve been reading a lot of things that are just compilations of links lately, like Benedict Evans’ newsletter and SSC links posts, and I’m inclined to start my own, partly so I can start posting regularly again. I’ll try to do this every week. It’ll mostly be stuff I found interesting the past week.

From SSC: Two basic at-odds political meta-theories: conflict theory vs. mistake theory. Conflicts theorists view politics as a zero-sum game, where it’s a constant struggle between those with power and those without. Poor political decisions are poor because they benefit the oppressive class. Mistake theorists think poor decisions come from poor decision-making/priors instead of power struggles.

The original rsync paper is surprisingly short, easy to read, and elegant.

A very good list of 57 startup lessons from the founder of RethinkDB. It’s hard to pick a favorite, but I’ll try:

Minimize complexity. The simpler the product, the more likely you are to actually ship it, and the more likely you are to fix problems quickly.

An oldie but a goodie: Steve Yegge’s Platforms Rant. Illustrates the importance of having a platform. You can’t expect to consistently build what users want. Let other developers do the work for you by building a platform instead and opening it up. AWS and Zynga-era Facebook games are great examples of this. So was Twitter before they added developer restrictions.

Why does SF have a huge homeless problem? Here’s why: 1) Insane rents due to NIMBYism, 2) Easy weather (you won’t see too many homeless in Montreal), and 3) Mental illness, although this may be circular: becoming homeless for an extended period of time probably makes you go crazy.

50 insane icebreaker questions by Chuck Klosterman: my favorite is Canadian Football Future — reminds me of DFW for some reason.

A $1.50 Michelin Star meal in Singapore. Particularly relevant cause I’m in Singapore right now and I really want to go try this place out, but it’s raining pretty hard right now. Maybe tomorrow.

Shameless self-promotion: I wrote a post on backtracking.

I used Elm to create the visualization. It was such a pleasure to use — like Haskell, but more front-end oriented. Here’s a good beginner’s book on Elm.

What I’m listening to this week:

Jarvis' 100 most asked reminders

27 Feb 2017

Well, it’s midterm season, which means I’m doing anything except studying for my compilers test tomorrow. I thought it would be interesting to go through all the reminders Jarvis has received and find the most commonly asked reminders. Thanks to our very liberal privacy policy, I used Heroku Dataclips to download a JSON-formatted list of reminders, and then wrote this short script to count the 100 most common reminders:

from collections import Counter
import json

with open('reminders.json') as reminders_file:
    data = json.load(reminders_file)
    reminders = data['values']

c = Counter()
for reminder in reminders:
    reminder_id, user_id, message, timestamp, completed = reminder
    c[message] += 1

print(c.most_common(100))

Here are the results:

Reminder	Count	Percentage
to wake up	1349	1.779%
to sleep	463	0.611%
to eat	350	0.462%
Yes	289	0.381%
to study	269	0.355%
to get up	222	0.293%
to go home	219	0.289%
to get coffee	211	0.278%
to go to work	189	0.249%
to call mom	187	0.247%
to drink water	183	0.241%
to go to the gym	165	0.218%
yes	159	0.210%
to go to school	144	0.190%
remind you	140	0.185%
to do your homework	139	0.183%
to take out the trash	133	0.175%
to go to sleep	131	0.173%
to leave	120	0.158%
to walk the dog	119	0.157%
to go to bed	113	0.149%
meeting	109	0.144%
to shave	104	0.137%
to work	103	0.136%
to do your best	89	0.117%
to do homework	89	0.117%
Remind me	79	0.104%
to pay your rent	78	0.103%
to fuck you	77	0.102%
Remind	76	0.100%
to call	73	0.096%
to go for a run	70	0.092%
No	70	0.092%
to pay rent	68	0.090%
to kill myself	67	0.088%
to go out	67	0.088%
to call your mom	66	0.087%
to take a shower	64	0.084%
to buy milk	63	0.083%
to buy toilet paper	62	0.082%
to eat lunch	61	0.080%
to take a bath	60	0.079%
to do something	59	0.078%
to take your medicine	58	0.076%
wake up	58	0.076%
to do laundry	56	0.074%
to go to gym	55	0.073%
to go	52	0.069%
to kill you	52	0.069%
to have lunch	52	0.069%
to go to meeting	51	0.067%
to say hi	50	0.066%
to pick up laundry	50	0.066%
to leave work	50	0.066%
to check metrics	50	0.066%
to brush your teeth	49	0.065%
to call dad	46	0.061%
to read	46	0.061%
to fuck	46	0.061%
to die	45	0.059%
to exercise	44	0.058%
to run	42	0.055%
to shower	42	0.055%
to pray	42	0.055%
to workout	41	0.054%
to pour tea	41	0.054%
to take medicine	41	0.054%
to poop	41	0.054%
to say hello	41	0.054%
Nothing	41	0.054%
to eat dinner	40	0.053%
to wake you up	40	0.053%
to go to the bank	39	0.051%
to call Mom	38	0.050%
to go to	37	0.049%
to test	37	0.049%
to get ready	36	0.047%
to leave the office	36	0.047%
to take a break	35	0.046%
to have sex	35	0.046%
to work out	35	0.046%
to go to the team meeting	35	0.046%
to go to class	34	0.045%
appointment	34	0.045%
to take your pill	33	0.044%
to pee	33	0.044%
to go to the dentist	33	0.044%
to wakeup	33	0.044%
to return the library book	33	0.044%
to go to leave work early	32	0.042%
to send email	32	0.042%
to have dinner	31	0.041%
to call mum	31	0.041%
to check	30	0.040%
to drink coffee	30	0.040%
to watch tv	30	0.040%
to fill out your visa forms	30	0.040%
to hand in math assignment	30	0.040%
to masturbate	29	0.038%
to go to the DMV	29	0.038%

Some of those are pretty…interesting.

If you found Jarvis useful, don’t forget to donate!

Announcing Juicebox

30 Jan 2017

I’ve been working on something for the past few months that I’d like to share with you. It’s a website called Juicebox where you can stream music with people from around the world. Think of it like Twitch for music. Check out my box, for example.

I’m also working on the mobile versions. Stay tuned!

Previous Next