30-minute binary training

So this morning I sat down for a bit of marathon practice. 30 minutes of memory followed immediately by 60 minutes of recall. The problem I always have when I haven’t done this for a while is just keeping concentration over that long period - my mind wanders, and I find that I’m just looking at the numbers and not remembering them at all while I think about something else entirely. It’s a real effort to keep my mind on the job, but I know from experience that if I get so I’m back in the habit of doing this every weekend I’m not busy, it soon won’t be a problem any more.

In 30-minute binary, I draw lines down the page, dividing it up into groups of 4-3-3-4-3-3-4-3-3 (each row for me is three images, each composed of a consonant made of the first four digits, a vowel from the next three and another consonant from the final three - when it comes to binary, perhaps because they’re spread across a wider area of the page, reading them takes a little bit longer, and I still find myself doing the intermediate step of spelling-out the word, which I don’t really do with decimal numbers or cards). A lot of people now use pre-prepared transparency overlays, and perhaps I should start doing that, because it really would save a little bit of all-important time. I just have trouble shaking off old habits.

Another little bit of time-wasting is that I do one journey at a time, and since a journey for me is 26 locations, that equates to one page (25 rows to a page) and one line of the next. It would make more sense to just use 25 locations in each journey, but again, that’s not the way I’ve always done it. I draw new lines when I turn to a new page, so I draw my lines, read the next 25 rows, turn the page, draw lines again, memorise the first row of the next page, then revise it. Then I read the remaining 24 rows of the second page, turn to the third, draw lines down it, memorise the first two lines, repeat that journey, and so on.

What I normally consider a good time for a journey is four minutes, but I was a bit over five minutes for each of them today. Again, I know I’ll get faster with practice. So I did five journeys and the first 15 rows of a sixth before I ran out of time (unlike numbers and cards marathons, I only look at each one twice, I don’t go back at the end to revise everything again. This might be because I take a little bit longer to memorise them in the first place, so they stick in my mind better. I’m not the only one who finds this - Gunther told me, years ago, that he reads numbers and cards just twice, and binary only once, without revising at all). So I looked at 4350 binary digits in total.

I normally expect to have a lot of gaps in 30-minute binary. I consider 80% success to be the target to aim for; that is, if there are twenty lines on a page that look correct and five that are wholly or partially blank, I’m satisfied with that. I used to have even lower percentages than that, before I started linking the last image on each location with the first on the next - I would normally attempt well over 5000 digits to get a score in the mid-3000s.

Today, though, I certainly didn’t come anywhere near to the 80% - I haven’t checked them yet (it takes a long time, I might not bother until I’m getting scores I’m not so ashamed of), but I know it was a score maybe in the early 2000s. But I can build on that! The important thing is just doing this and not getting bored and wandering away to watch TV half way through! :slight_smile:

Edit: Okay, marked them while watching some wrestling - 2125. I’m sure I can beat that next time!

1 Like

This isn’t related to binary, but since this is a sleepy thread…is the info hosted on memocamp.de right, that in 2006 and 2007 you scored your best national N&F scores to date, and that they are 53.5 for 5 min national N&F and 96 for 15 min national N&F?

Here are the scores of four competitors at the 2013 USAMC:

Saira Kothari:
Speed numbers: 40 digits
Speed cards: 24 cards
N&F: 124 names

Kelly Kohlman:
Speed numbers: 40 digits
Speed cards: 15 cards
N&F: 114 names

Kathryn Ruszkowski:
Speed numbers: 47 digits
Speed cards: 27 cards
N&F: 102 names

Mallika Kodavatiganti:
Speed numbers: 32 digits
Speed cards: 2 cards
N&F: 100 names

This may be a premature assessment, but they might be in a different competitive bracket.

Perfect; the point has now been belabored. :smiley:

Is the N&F you do a little bit different from the one at the USAMC? Is it boring to train for?

Boris also decided not to review his binary digits at this last WMC. Wonder if it contributed to his improvement?

That’s right, I’m famously bad at names and faces. Everyone makes fun of me for it.

But in my defence (just a little), the names at the US Championship are ‘American’ names, while in other competitions they’re ‘international’. The rule is that they are a random pairing of a first and last name which could come from anywhere in the world and don’t have to share an ethnic origin with each other or with the face they’re under.

Back in 2006/07, there wasn’t such a hard and fast rule, but the names were still more obscure and unfamiliar than the names at the American competition.

Oh yeah. And you keep em out of your object as much as possible.

But finding the discipline so difficult, you spend a lot less time working on it, right? Fun fact: yesterday and this morning, I set personal bests on both 5 and 15 minute national N&F that are almost identical to yours: 54/60 in 5 minutes, and 105/110 in 15. 600 on every event is a nice standard because it gets you a score of 6,000 overall, which is the new IGM requirement as you know.

To score 600 in these 2 events, a person has to get 60 and 120 names…we’re pretty close! My resource for faces is memocamp. Are you on there? If not, the limitations given are 72 names total for the 60/72, and 130 names for the 120/130. 600 points; a kind of not-quite arbitrary standard. If you have access to faces, we could see who gets there first. It’s just lucky we are equals here. FYI, I’ll probably do 3 5-minute runs and 1 15-minute run per week. What do you say? :slight_smile:

I like the sound of that! I used to practice on memocamp a while ago, but gave it up when my subscription ran out, because I was trying not to spend money unnecessarily.

My finances are still precarious (I’m currently trying to work out who it would be least embarrassing to ask for a loan of the money I need to get to the Extreme Memory Tournament), but I suppose you’ve got to speculate to accumulate, and I’m sure memocamp isn’t all that expensive nowadays, right? Some online practice would do me a lot of good with the XMT coming up, and a challenge to motivate me in names and faces is just what I need!

Okay, sounds good! I’ll stick to the practice schedule above. I got a 6 mo. subscription because it would luckily end on April 1. It cost me $10 usd per month for the subscription for that long. It is kind of a lot of money for training for the USAMC only, but is a more justifiable expense if training for Abstract and Auditiv Ziffern as well.

Please share your opinion about this if you would, Zoomy.

There’s an enormous disparity between the number of images used in the 5 minute binary world record, (1080 digits = either 108 or 120 images depending on whether Mallow uses a 9 or a 10 digit system), and the 5 minute digits world record (501 digits = 167 images). The gap ostensibly narrows in longer disciplines, but that actually is just the effect of reduced consequences of errors when there are 3 images per line instead of 13. When memorizing the 0-9 digits, an MA has to think of damage control. With binary, he can be reckless. But even with this extra care, far more correct images are recalled in the 5 minute digits versus the 5 minute binary.

A string of 10 binary digits is not that it’s too much information for the brain to handle. It’s just three chunks; the problem is the eye. The physical length of the string of digits makes reading it as a chunk difficult and unreliable, so it is absorbed in the form of 2 or 3 chunks. Does this sound right in your experience?

We could get fix that by encoding binary in blocks instead of lines.
Though any single error with a 3x3 code would cost a whopping 90 points instead of just 30, it would absolutely be worth the risk provided that
a) that danger were mitigated by adopting the digits strategy of eliminating or greatly reducing errors, and
b) the binary digits could be read and encoded and/or reviewed at the same or very nearly the same rate as 3 digit images. If closing the gap is possible, then that’s what has to be done.

If you believed that 3x3 blocks had a definite advantage in reading speed, would you agree that it would be worth the trade-off of more points taken away per mistake?

The very, very short explanation of my idea is to organize 512 images into “families” based on similar attributes of the appearance of the 3x3 block, so they are learned primarily by the right brain and read heuristically as opposed to depending on the left brain to convert the chunks into the digits of meaning that we are used to. The numbers that the 0’s and 1’s describe would be meaningless unless you wanted to incorporate the major system or some modification of it. It’s more like Chinese. As such, the 512 images might be studied more slowly and carefully than we would do with a typical card or digit system; more like a foreign language. So for instance, the most obvious image-family is one with nine members; the “single 1” family, which would be:

1 0 0
0 0 0
0 0 0

0 1 0
0 0 0
0 0 0

0 0 1
0 0 0
0 0 0

0 0 0
1 0 0
0 0 0

0 0 0
0 1 0
0 0 0, etc.

So a person would learn these images little by little, in groups of 4 usually, but in groups as small as 2 and as large as 16. 512 images later, you have a 3x3 “character” representing each image.

Do you see anything I don’t see? A fatal flaw?

1 Like

FYI, I followed up on the method above and learned it. Or at least I’ve been learning it. I really enjoy it, except for that Zoomy’s strategy above is unwise to use with it which is a shame because he’s the half hour binary master so a big part of me just wants to copy him. And my 30 min binary scores are no good yet. But just a moment ago I got a score of 720 for the first time, which was the goal that only weeks ago I set for the end of October! :smiley: :smiley:

Hey Lance, I was curious how you practice binary. The memocamp binary lines seem pretty difficult to read as 3x3 blocks. Do you use the binary stuff in the training tab on this website? Print stuff out? I know Ben mentioned that he marks the pages for 30 min binary. Do you do this? It does seem like it would help with the reading a lot, but I wonder if it’s too time consuming for 5 min binary…

One more random question: if you use a 3x3 method and 3 images/locus, you end up with one 3x3 block at the end of every 3-line chunk. Do you just go right over the line? As in, you make this last 3x3 block the first image of the next locus, along with the first and second blocks in the second 3-line chunk? Sorry if that was confusing. Thanks for your help!

If you haven’t seen it, I recently created this:
https://artofmemory.com/3.1415926/t/binary-grids/

If you have requests for new scripts or for improvements to existing ones, just let me know… :slight_smile:

Alex, I don’t mind explaining myself. It feels good to influence others’ directions when I believe in the path. I actually enjoy the questions on that account.

Alrighty. First of all, you will eventually observe that there is a difference in the kind of information that is forgotten when using linked objects from a large list. Especially when you are good enough to do what Zoomy mentioned above that is so helpful, and it really is so helpful, it’s been a main focus of mine since he mentioned it in his AaMC thread, which is to continue to link objects across loci. I have forgotten a string of objects because the object before the first in the string didn’t remind me of the next, but the object after the last object of the forgotten string reminded me of the object before it, and so on, and so on, and working backwards, I fill the whole gap back up. The point is that when stringing objects and linking across loci, assuming you are good at that, there is no major reason I can see that the contents of a given locus would be forgotten. It’ll just be an object here and there, and you’ll usually have some context (“it must have been some kind of projectile or moving object…”) but the whole thing is a string, so taking the tenth block and starting a new 3-image locus is not the risk it seems.

One time so far, I did forget an entire locus at the end of a line. And I lost 150 points from it, knowing with 10-digit binary, I’d have lost only 30. And it hurt. But mind you, that has only happened once.

On the drop box above the metronome, there is an option 3x3. Use this, and below it select “bounce” or “reflect” the object across several lines. When you do that, you’ll have a higlighted 3x3 block that you can move with te forward arrow or set to move forward with the metronome. How cool is that?! Memoriad is also free software that separates the blocks by lines.

Yeah, Zoomy draws lines in between palaces. You might want to try that too once you become the best in the world, because being the best in the world is probably the only way you could afford the loss in time and momentum.

I do not use the page on this site just because memoriad has a built in timer that you can set to 1 minute and then the page vanishes, so you can get an accurate score for how many objects you can read in a minute.

However, if I were able to somehow practice reading only the 64 digit group of my choice (there are 8 of these groups, those with the top line being 000, 001, 010, 011, etc.,), and I were able to randomize this group and read it and link it again, I would absolutely abuse that training script. If you practiced reading and linking just the same 64 blocks each week (along with regular binary practice), you would be such a baller at the 2 month mark. I’d take 64 blocks over 72 virgins any day, but we aren’t in heaven. If only there were some savior

Someone up above…

^ ^ ^ up there…above me…

I just got a printer so I guess I’ll start printing stuff out, I haven’t thought to do that yet.

Edit: I wasn’t completely fair above. I have had excellent results on card speed following a review of the entire list of images. When I am a little bit faster at reading the binary blocks, I will start to use the training link. For now, it takes a little bit too long to do. But it would be especially good to link these together at an increasing speed (metronome?) because it is a little over 4600 digits total, which is a little more than the 30 minute world record. So you get that reference on top of being able to increase your speed and review the whole list.

I use 3x3 blocks of binary when memorising, it is extremely efficient and effective

Awesome! So you have 512 images then. Do you have 1000?

How do you guys solve the problem of “25 rows”? Think about it. If you memorized binary digits in chunks of 3x3, then you would end up with the last line unmemorized.

My idea is to memorize the last row and maybe few rows of the next page as a string of digits in a row, and then you would switch back into chucks. Everythings depends upon the system and number of images per locus. You simply have to memorize the entire rows in order to have a “clear” start for the next chunk.

I use 2 digit system, 2 images per locus… I think I’m going to try 3x2 chunks of binary.

Anyway, good topic. :wink:

Oh my god, I can’t believe that I never thought about it! ¯\ _ (ツ) _/¯

There are no rules about which lines must be memorized, so you just skip it and get 0 points for every 25th line.

It’s a good thing too, because I’m a little bit of a perfectionist. At first, the idea that I should learn the 1024 objects just for every 25th line and then switch back to 3x3, which is the easier system to learn, made me want to pull my hair out.

2 images per locus is the way to go if you’re comfortable with it, just to be extra safe about ending loci at the end of the line. Not worth switching to 2 images per locus, but a boon for those who are already used to it, with 3x3 or 3x2.

3x2 sounds nice, but Jonas has been incredibly successful and made very quick progress using PAO with the 6 digits in a line, which is much safer because an error is only ever worth a potential 30 points, whereas with 3x2 it would be a potential 60. What do you think are the benefits of 3x2 that make up for this disadvantage?

1 Like

Well, as I said above, I’m going to try it for fun, since you started this topic… You also posted an interesting comment above regarding the number of images for 5m binary numbers and 5m decimal numbers. The number is different…Maybe because of the “eye” problem. And with my serious eye condition I’m not able to memorize numbers in a row without skipping or memorizing the same number twice… That’s the reason I want to try 3x2… I could be able to look at them without some extra eye movement… Maybe I’m wrong. I’ll let you know.

But you are right, in terms of mistake making, I can definitely end up with far worse score…

sorry guys, I do a 3X2 not a 3X3

I use it for speed binary, you can get 40+ digits in under 4 secs quite easily
I’ve even hit the 30 digits mark in 1 sec

My phone died and I still have to get memory ladder onto the replacement so I am a bit out of practice but I was using memory ladder with 12-15 digits per line.

I found if I went beyond 6 digits per line the eye movement slowed me down, working downwards in blocks of 6 (3 across X 2 down) was much easier than working from left to right

That’s a very good reason. Doing binary like this is a lot of fun, memorizing from these patterns rather than numbers or cards we are so familiar, you know? The 1’s and 0’s could just be stars and hearts instead.

That is very, very interesting. Remember the caveat with what I’m doing is that I have to be able to hit my goals without making mistakes, just like with digits. From my using this system instead of a 10 digit one, it could be inferred I’m willing to ‘live dangerously’ for what looks to me like the potential for extra speed. Maybe that commitment could be taken even further, increasing the both the penalty for a missed locus and the overall speed by moving downward instead of across. Thanks for the idea, I’m going to play with that for a while.

Hey Lance, I’m trying to figure out what I’ll do with my 3-digit/2-card system as far as the number of images per locus, so I was wondering what you do these days. Do you have a pretty set standard now? I noticed in your blog you’ve experimented a ton with different options; have you settled on any settled on anything for decimals, cards, binary, words, etc.?

All I’ve used up to this point has been PAO and 3/locus for words. As far as I can tell, I’m pretty comfortable with 3/locus, but I’m thinking things may change without the nice order of PAO. Here’s what I’m thinking right now:
Decimals: 3/locus (I know this is what Ben does, and Mallow used to, but I think he’s switched to 2/locus)
Cards: varies (I’m using a 2-block system, so it changes, but I’ll think I’ll do what Mallow does; ie. images from block 2 end the locus)
Binary: 3/locus
Words: 3/locus
Spoken #: ? (Never even tried it; do people tend to adjust their approaches much in this event?)

If anyone else has experimented with # of images/locus or has some ideas on this, don’t hesitate to share :slight_smile:

1 Like