All of renato's Comments + Replies

I'm looking forward to read it, because I think one of the current bottlenecks that limit how many predictions i do is that i cannot easily compare how i'm doing week after week, and i have been looking for a model that help me check how i'm doing for several predictions.

1Jan Christian Refsgaard2y
you may be disappointed, unless you make 40+ predictions per week it will be hard to compare weekly drift, the Bernoulli distribution has a much higher variance compared to the normal distribution, so the uncertainty estimate of the calibration is correspondingly wide (high uncertainty of data -> high uncertainty of regression parameters). My post 3 will be a hierarchical model which may suite your needs better but it will maybe be a month before I get around to making that model. If there are many people like you then we may try to make a hackish model that down weights older predictions as they are less predictive of your current calibration than newer predictions, but I will have to think long and hard to make than into a full Bayesian model, so I am making no promises

Because 8.86×10−30>2.76×10−30 Person A is also a slightly better predictor than person B.

Wait, i got confused by the function you used to assign the calibration score. It worked in that case, but it will yield higher values for those who make more 'correct' predictions, not those who are more calibrated. For example, person A predicts 100 things with 60% confidence, 61 of them turns out to occur and person D predicts 100 things with 60% confidence, 60 of them turns out to occur. Person D is more calibrated, but gets a lower score than person A, ~5.9e-30 vs ~8.86e-30 (and person E who made 100 predictions with 60 % confidence, which all turned out to be true, would score ~6.53e-21).

2Jan Christian Refsgaard2y
I have tried to add a paragraph about this, because I think it's a good point, and it's unlikely that you were the only one who got confused about this, Next weekend I will finish part 2 where I make a model that can track calibration independent of prediction, and in that model the 60% 61/100 will have a better posterior of the calibration parameter than then 60% 100/100, though the likelihood of the 100/100 will of course still be highest.
2Jan Christian Refsgaard2y
You are absolutely right, any framework that punishes you for being right would be bad, my point is that increasing your calibration helps a surprising amount and is much more achievable than "just git good" which is required for improving prediction. I will try to put your point into the draft when I am off work , thanks

Does anyone think that Anki is better than real life use for learning? For rote memorization, yes. For other more complex tasks, no, and it is not even its intended use.

Or is it perhaps more of a (possibly imperfect) substitute for when one cannot avail themself of a real life usage setting to apply what they have learned?

As others mentioned before, it is a complement for other study tools. It won't make you fluent, and probably it will not even help you directly to read/write/listen/speak, but it will make shortcuts to help you comprehend faster and ma... (read more)

I'll have to look into abilities to view and slow the playback. I do watch a fair amount of Korean language shows. Not sure if Viki's interface includes that or if I can setup PotPlayer (VLC always causes some problem on my system that I've never figured out -- and with all the options never willing to invest too much effort in solving)  to stream the shows and take advantage of the suggestion. Side note on watching the foreign shows. Since I'm also watching for entertainment value if generally have CC turned on. However, one things I have learned is I have to be careful about getting into the habit of reading and not really listening. When that happens I just hear English in my head and it starts drowning out the Korean! Turning CC off if I want to be in pure study/learning mode solves that but trying to be in that mode 24/7 is really hard on the brain ;-) Thanks for the Lang-8 suggestion, will look into that as well.

I'm glad you could already find the answer for it, and I hope your dad gets better.

I'm writing just to say something that has worked for me (I'm also allergic). When I was stung by a wasp, I found that applying a cold compress reduced the swelling considerably, and I recovered in a matter of hours. When it happened before and I didn't use the cold compress, I remember it taking days to recover, and I could not walk properly while it was swelled. Both times I was stung on my ankle.

The paragraph that starts with "Fourth, ..." is shown twice.

1Filipe Marchesini3y
Thanks, corrected it now.

Hurting knees can be due to riding in too high a gear.

Another common mistake is having the saddle height too low, forcing you to bend the leg too much and putting more stress on the knees to move the pedals down. Even a few centimeters (~5cm) can make a big difference, but it will feel strange at the beginning and it take some rides to get completely accustomed to the different movement.

I found a dropper post to be a great help with that. It's much easier to figure out the right height while riding and not having to dismount to adjust it. And anecdotally, it sometimes feels better to adjust it up or down by 5-10 millimeters, maybe due to different clothes or shoes or posture or surface grade. Note: even the cheapest dropper posts costs around 100 euro (from a cursory Google search). People who aim for cheap bikes often don't consider them. If you can afford it, consider if it would be a small investment into your comfort and longer-term health.

I could stop biting my nails using some TAPs.

Installing the TAPs took at most a few days, and I think I might have failed and bit the nails some times. It took around a month to get the nails long enough that I could cut and polish them, and after a while the TAPs disappeared completely because I didn't need them anymore. According to my journal, I started using the TAPs on 20180612 and cut my nails on 20180714, and there is also a register on 20180703 that I was not using the TAPs because their trigger had disappeared.

I think that what made me bite my nai

... (read more)

The software I'm using is AutoKey. What do you use? Are you happy with it?

I'm using LXDE as a desktop environment and it allows you to set the shortcuts in its config file. If I'm not mistaken, other flavors, like GNOME and KDE, had a GUI to bind a certain key sequence to a script.

It seems that it catches the keybindings instantaneously and any lag is due starting a program called by the script, but it is usually around 1 second, which is similar to launch the program independently.

Autokey sounded very promising, but it is too slow. I wanted to port my

... (read more)

It is really nice to have someone reporting on those little things that make a good (unseen) impact on their workflow and that we usually don't even consider the possibility of having them. I have something like that implemented and I also noticed that reducing those small frictions result in much more notes and less disruption of the current task, you think of something, a note is added in a few seconds and you can continue working on.

They are also surprisingly easy to implement, and I got surprised how fast it took me to have something functional. I thin

... (read more)
1Sunny from QAD3y
I upvoted for this snippet because it's an important aspect of the situation that I forgot to call out in the main post. Sure! This one is actually a one-liner: it's simply "gedit ~/Documents/lists/$1", which you put in a file called "l" in your ~/bin/ directory. If you prefer a different editor, you can swap out "gedit" for "emacs" or the command used to launch whatever editor you like. (This advice is directed at others reading this comment chain, you probably already know how to do that.) That's a good idea. I currently have a piece of software that I use to type diacritics (for Toaq) but I'm not super happy with it — it kind of bugs out on occasion and can be slow to insert the characters I want. The software I'm using is AutoKey. What do you use? Are you happy with it? This is also a good idea. I'm pretty fast at typing and pretty slow with the mouse, so I'd probably instead make a macro for "prompt me for a search key, open a new tab, search that thing, then take me back to the tab I was in before".

Portuguese uses the same vowels terminations for genders, but our articles are a simple 'a or 'o' (instead of 'la' 'lo') and we also use 'e' as the and connector. It means that the vowels 'i' and 'u' would still free for the third gender, but we do some vocal accommodation orally (I'm not sure about the correct linguist term) and often the sound 'e' becomes 'i' and 'o' becomes 'u' (it does not happen the other way). Because of that, all of our vowels are already "taken" with just two genders.

I found it fascinating that it works so well in Spanish and not at all in Portuguese, even with both languages being very similar (I feel that Portuguese is slightly more gendered than Spanish).

I liked how your idea resembles the aggregation used in prediction markets. But, I'm not sure how it would reflect in the number of cases judged by judge after considering that they receive the same ones to review. If the ratio of KBCJ to normal ones is high, there will be a waste from too many people review the same cases, and you are trading efficiency for more predictability, but it might cost too much in money terms. If it is low, and the pay rate is higher than the normal ones, it can create two different judging modes, the normal where they will stil

... (read more)

They are not too complex nor rare, so I suggest that you use some related words to increase your exposition to them in a slightly different way. More concretely what I usually do for hard word is:

  • Pick the noun that derived the adjective/adverb, or the other one.

    For example, dangerous (gefährliche) <-> danger (gefähr ??) I rely mostly on the suffix like 'lich', because it preserve the original word. There are other ones like in the case of imagine(d) -> imagination, and as your vocabulary expands and you are able to recognize them easier the mem

... (read more)

A little unrelated to you original question, but it seems useful.

I could add "learn to write" to my goals but that seems significantly harder to self teach, because checking my work is harder than looking up the same sentence in the English version of the book.

You can try Lang-8 where some native speakers correct your writing while you do the same for some other users. I used it in the past and the community was very good, the texts I submitted were corrected very fast (it depended on the time I have posted them, but usually less than a day, and someti

... (read more)
Lang-8 looks incredible, thank you.

Can you give an example of the words that you cannot memorize and the sentence they are coming from (with a translation)?

Are those basic words or probably some rare ones? Depending on the frequency of those words appearing again I would suggest either:

  • If they are fairly common, add some related words to force some redundancy.

    I use Anki to memorize Japanese vocabulary, and in my case I pick words that share the same kanji than the hard word. For example, the word 図書館 (library) can be made redundant and easier to remember by adding 図書係 (librarian) and 図書

... (read more)
Sure. These are the first 5 that came up on Anki that I didn't know erreicht/reached Sie hatten das Haus erreicht/They had reached the house gefährliche/dangerous Sie begeben sich auf eine gefährliche Reise/They began a dangerous journey bedrohlich/threatening vorgestellt/imagined aber es ist weitaus düsterer und bedrohlicher als Quentin es sich je vorgestellt hatte/ but it is much darker and more threatening than Quentin had ever imagined bereit/ready Aber dazu war er noch nicht bereit/But he was not yet ready. It's chance that no nouns came up, I have just as much trouble with them.

A very interesting analysis of an interesting article. I'm not familiar with AI development and because of that my questions may be too elementary.

Its major strength is its ability to use mixed strategies... to do this in a perfectly random way and to do so consistently. Most people just can't.

It amazes me how much of the advantage from AI and other computer programs are derived from their lower bias than humans.

Because poker is played commercially, the risk associated with releasing the code outweighs the benefits. To aid reproducibility, we have

... (read more)

The images are not working anymore.

The wayback machine has a stored copy here.

The original website is down. This is an archived version around the same time the article was published:

Well… not necessarily. We’re talking here about a wheel, with weights on it, rolling down a ramp. Mathematically, this system just isn’t all that complicated. Anyone with an undergrad-level understanding of mechanics can just crank the math, in all of its glory. Take no shortcuts, double-check any approximations, do it right. It’d be tedious, but certainly not intractable. And then… then you’d understand the system.

I guess the idea was to emulate a problem without a established theory, and it was chosen to provide a simple setup for changing the para... (read more)

Thank you for the information.

I've finally finished some kind of mandatory material I had to cover for some exams.

Even as I probably failed them, I still learned some stuff in the process, and now I'm almost free to pursue something more interesting.

Thank you for the invitation.

Although, the artificial intelligence and related topics interest me a little, I'm still lacking several of the basic requisites.

I will try to focus on them first, and then decide what to do next.

I understand your first equation 3*30= 90 hr (3 hrs/day* 30 days = 90 hrs to complete a book). The second one is a bit confusing. 3*eff*90 = ~30 hours. (3 hrs/day * 1/9 efficiency * 90 ? = ~30 hours) Was the second 90 supposed to be 30 to make efficiency 30%?

Yes, I put the wrong number.

I would say my average was 4 hours during almost two months. I could notice it slowly increasing, but every time I need to change to a new subject/different kind of activity (from reading a non-math book to one that requires more exercises) my time plummeted. After a few... (read more)

1Logan Riggs5y
I understand now! haha My study time was a non-interrupted 1-3 hr block. This made it easier to get in the zone and have relevant details in my working memory. Going for longer than that (4-6 hr), I'd predict I would need to take a walk outside and just think of nothing to let my subconscious do it's thing. I haven't done that more than once or twice this summer, so I'm not sure what would be normal for me. TurnTrout has a lot more experience doing that than me, and he'd be a great resource for any of these type questions. We actually have a discord server with several people studying miri-related materials if that's something that interests you.
I interpret that to mean "how many pages per day"/"How many chapters per week"/"How many books per month". If that's correct, then I would say I could (right now) learn a subject/book a month. Like I could read Linear Algebra Done Right in less than a month and Tao's Analysis I & II in less than two months, while doing all of the exercises.

It wasn't exactly that, but it still a very good information.

I will try to describe it with my own experience.

When I read Nate's post, he said that he studied 5~7... (read more)

2Logan Riggs5y
Thanks for the clarification! I understand your first equation 3*30= 90 hr (3 hrs/day* 30 days = 90 hrs to complete a book). The second one is a bit confusing. 3*eff*90 = ~30 hours. (3 hrs/day * 1/9 efficiency * 90 ? = ~30 hours) Was the second 90 supposed to be 30 to make efficiency 30%? I definitely think efficiency plays a huge role, and I'd say (Hours/day *days * efficiency = total actual hours for a 100% efficient person) would be my answer. (hopefully) Related: an overall good frame to tackle this is focusing on increasing quantity and quality of time spent. Increasing quantity: * Taking time from other activities and put it in this. * Gradually building up to longer and longer times. (6-7 hours on a good day is fantastic, and I applaud you! So what about on average? 4-5 hours?) * Noticing why you stop. Does your mind want to quit? Are you physically tired or have tired eyes? Increasing quality: * Making sure you have a good textbook. * Simply reading faster w/o losing comprehension (Imagining hypothetical "Gun to the head, read as fast as you can w/o losing comprehension for 5 minutes, you will be quizzed" might be a good exercise to play around with) * Not wasting time between activities (like going from one exercise to the next) but also!: * Knowing how to make use of your subconscious by letting your mind wander (I don't have a gears-level model of this one yet, but I have intuitions that it's important) May I ask why you would like to compare study paces?

Congratulations on finishing this 80 days journey.

It is always nice to read some feedback from people after they finished a long project like that.

I have been reading some of Nate Soares' post recently, and I'm still fighting with myself to start trying to do something similar. I hope I can start something like that soon. Your post gave me an extra motivation to do it.

Can you detail a little more how much time you spent on each activity you set on the other post? If you didn't followed the set amount, when did you decide to change them?

About... (read more)

The most potent motivational blend I have found is a mix of emotionally updating to x-risk being real, and of delightful curiosity about discovering how the mathematical world ticks. Echoing Nate: once you get started, the learning part is really not that hard (in my opinion).

Thanks renato!

Regarding your first set of questions:

Reading: originally 3 hr. This changed to 0-3 hours depending on when I woke up in the mornings, which meant going to bed around 9-10, which meant making a habit of trying to fall asleep when I get in bed. I did try to read in the evenings as well, but my eyes would glaze over after working for the day.

Tensor flow: originally 2 hr. I dropped this after the first week due to work taking up 2.2x as much time as I initially predicted. I also felt like I wasn't actually learning anything while going thro... (read more)

The vast majority of people who were on a job site, seeking a job, were not capable of tasks like: Write a profile page in English without major mistakes. Respond promptly when an employer contacts you to respond to your application. Talk politely on the phone and sound like you are listening and care about the job. Show up to your interview.

If there are several jobs open, and the chances of being hired after jumping through the hoops is 50%, some of the candidates probably found something better, and stopped wasting their time on a lower position.

Ther... (read more)