Switch from NF to FL

Anything to do with minesweeper...
Post Reply
Zlix
Posts: 38
Joined: Fri Jun 13, 2014 12:29 pm

Switch from NF to FL

Post by Zlix » Tue Sep 29, 2015 10:37 am

Hi guys.
After several years of NF, I decided to learn FL. It's quite overwhelming to keep track of flagged mines for now. Also, hard to decide which mine to flag. Any advice from experienced players is welcome. My current records are 64,7s and 2,76 3BV/s. Those records are almost identical to the ones I had when I decided to join the rankings. It will be interesting to see how fast will I improve my FL records. Also, current averages are 76,3s and 2,29 3BV/s.

Goran

User avatar
Tommy
Posts: 223
Joined: Mon Dec 01, 2008 9:22 pm
Location: Vienna

Re: Switch from NF to FL

Post by Tommy » Tue Sep 29, 2015 6:25 pm

First off, by playing NF, you probably completely got rid of using flags for orientation.

Do not reacquire that.

Some non-obvious things about FL (and very fast sweeping in general):
  • If you're fast, you probably have an "information pipeline" where you think about three things at the same time - you are executing your current plan, processing the new information you are getting, and planning what you will do next based on what you saw and processed half a second ago. This sounds way more rigid, complicated and thought-based than it actually is - I think that most people on the world ranking do this. It is the reason why you sometimes can't stop yourself from blasting on a square you now realize to be a mine - it would take reaction time to stop your current plan.
  • By placing flags, some of the time, you are taking actions that will not give you new information.
  • Therefore, flagging one mine and opening two fields with a chord is worse than opening them NF - you see the first square after the first click, and can plan accordingly.
  • Chording on single flags you have used before is absolutely safe. This compensates for what I just wrote, as it means that you can always "spam"-chord into an area as you process the information you are only now getting.
  • Flags do not mark mine positions. They are a game mechanic that allows you to open multiple squares (a complement) at once, and some of the best unforced guesses are chords on unsafe flags.
  • On edges, mines are deeper in the field than the squares that you will chord on. This means a more jagged mouse path.
  • While flags don't give you new information, they are perfectly safe to place. The same thing that hurts your information pipeline is a bonus here, as you can stop yourself from executing a bad plan. While a hypothetical perfect player doesn't accidentally misclick, we all know that all of us do that sometimes. Therefore, FL players are a little bit safer from human error and can risk a little bit more.
So, what I'd advise is:

Only flag when it is significantly more efficient to do so (you should be saving an absolute minimum of one click, if FL needs as many clicks as NF, always NF - you will get information faster, and FL mouse paths tend to be less smooth).

You probably never want to place more than two flags at once. Otherwise, you aren't opening new squares and processing information - as a consequence, even if your chord is efficient, you'll probably need to pause and think.

Some flags can be used more than once. Factor this in your FL vs NF decision, and prefer "deep" flags where reuse is more likely (more adjacent uncleared squares).

It is easy to fall into a FL rut where you start flagging everything out of laziness. Keep playing NF to prevent that.

And probably some other things :D
Don't anthropomorphize computers - they don't like it.

pauv
Posts: 78
Joined: Wed Nov 20, 2013 6:06 am

Re: Switch from NF to FL

Post by pauv » Wed Sep 30, 2015 6:02 am

Just pick up one of Kamil's skill:
flag and chord one side, then flag and chord another side, meanwhile analyze the pattern around the 1st flag, come back and chord more. You need fast moving and clicking ability for this.

Xian-Yao Zhang learnt another skill. NF a square(that is save) adjacent to a mine, flag and chord no matter what number it is. No consideration needed here, if the number proved to be 1, you gain a lot. If that mine has been marked, just chord after the leftclick. This skill is easy to do and you can observe even other area at the same time. After that (if it is 2) guess flag and chord again is a possible choice.

You can practice these two skills after sub45, IOE is not important if you are very fast. Follow tk's advice before sub45.

EWQMinesweeper
Posts: 410
Joined: Sun Nov 30, 2008 11:50 pm

Re: Switch from NF to FL

Post by EWQMinesweeper » Wed Sep 30, 2015 4:20 pm

Tommy wrote:Therefore, flagging one mine and opening two fields with a chord is worse than opening them NF - you see the first square after the first click, and can plan accordingly.
I disagree. With NF we have 2 mouseLeftDown and 2 mouseLeftUp events. With FL we have mouseRightDown, mouseLeftDown, mouseLeftUp, which is bound to be faster. Also the mouse path is usually slightly shorter and the risk of misclicking is lower (only 1 click which opens squares and could potentially lose the game). The planning you mention has to happen before you start solving a pattern: is it worth throwing another chord in the dark right after the first one? This is much faster than looking at the number you just opened whilst clicking elsewhere. Some of my pre37 games and I think some of Ian's games show this quite nicely.
Tommy wrote:Chording on single flags you have used before is absolutely safe. This compensates for what I just wrote, as it means that you can always "spam"-chord into an area as you process the information you are only now getting.
Use this with caution. While it is fast, it is not always efficient. Use it for a while, then don't use it for a few games and compare your IOE, 3BV/s and cl/s.
Tommy wrote:Flags do not mark mine positions. They are a game mechanic that allows you to open multiple squares (a complement) at once, and some of the best unforced guesses are chords on unsafe flags.
Depending on how much you play you can afford to take more unforced guesses. Think about it like this: "Where does the mine have to be so that I can chord most efficiently?"
Tommy wrote:Only flag when it is significantly more efficient to do so (you should be saving an absolute minimum of one click, if FL needs as many clicks as NF, always NF - you will get information faster, and FL mouse paths tend to be less smooth).
I'd say use flags whenever it doesn't cost additional clicks. A mine that is already flagged and has been used for a chord can be used again for another chord. That's where FL becomes so efficient.
Tommy wrote:You probably never want to place more than two flags at once. Otherwise, you aren't opening new squares and processing information - as a consequence, even if your chord is efficient, you'll probably need to pause and think.
Try to never place more than 1 flag between chords. There may be some exceptions where placing 2 flags at once is faster, but they are very rare. The usual sequence would be something like this: flag-chord-flag-chord-chord-flag-chord.
Tommy wrote:It is easy to fall into a FL rut where you start flagging everything out of laziness. Keep playing NF to prevent that.
NO! Instead play for IOE and do some slow-clicking.
pauv wrote:Just pick up one of Kamil's skill:
flag and chord one side, then flag and chord another side, meanwhile analyze the pattern around the 1st flag, come back and chord more. You need fast moving and clicking ability for this.
Keep it simpler and you have what I refer to as "sweeping" in my solving style.

I haven't watched Kamil's latest replays but in his older games he hardly ever used this. His solving was more straightforward with small hesitations in between bursts of lightning fast clicking.
pauv wrote:Xian-Yao Zhang learnt another skill. NF a square(that is save) adjacent to a mine, flag and chord no matter what number it is. No consideration needed here, if the number proved to be 1, you gain a lot. If that mine has been marked, just chord after the leftclick. This skill is easy to do and you can observe even other area at the same time. After that (if it is 2) guess flag and chord again is a possible choice.
Doesn't sound very promising. The additional chord mentioned above is much quicker.
pauv wrote:You can practice these two skills after sub45, IOE is not important if you are very fast. Follow tk's advice before sub45.
The faster you get, the more important IOE becomes. I'd say learn both skills from the start: solving with high cl/s and efficient solving.
„Das perlt jetzt aber richtig über, ma sagn. Mach ma' noch'n Bier! Wie heißt das? Biddä! Bidddää! Biddddäää! Reiner Weltladen!“

Zlix
Posts: 38
Joined: Fri Jun 13, 2014 12:29 pm

Re: Switch from NF to FL

Post by Zlix » Thu Oct 01, 2015 12:48 pm

Thank you for all your replies.
I'll read them again from time to time and I'll try to implement as many suggestions as possible over time. Today is my 6th FL day and I'm improving way faster than expected... New hi-scores are 56,42s and 2,90 3BV/s with today averages of 69.62s and 2.39 3BV/s on 10 games.
Best advice for the beginning is EWQMinesweeper's last line. Rest will be useful after I get some routine..

How many flags is optimal on expert?

EWQMinesweeper
Posts: 410
Joined: Sun Nov 30, 2008 11:50 pm

Re: Switch from NF to FL

Post by EWQMinesweeper » Thu Oct 01, 2015 3:48 pm

How many mines are optimal on expert? This is impossible to answer. It depends on the board and the sweeper.

My general rule of thumb is 1 mine per second. A bit lower, if the 3BV is low (sub130), a bit higher, if the 3BV is high (sup200), a bit lower, if your times are above 70s, a bit higher, if your times are below 45.
„Das perlt jetzt aber richtig über, ma sagn. Mach ma' noch'n Bier! Wie heißt das? Biddä! Bidddää! Biddddäää! Reiner Weltladen!“

Zlix
Posts: 38
Joined: Fri Jun 13, 2014 12:29 pm

Re: Switch from NF to FL

Post by Zlix » Thu Oct 01, 2015 6:53 pm

Thanks.
Forgot to ask what is good enough IOE at full speed? I've already improved my NF IOE hi-score...

EWQMinesweeper
Posts: 410
Joined: Sun Nov 30, 2008 11:50 pm

Re: Switch from NF to FL

Post by EWQMinesweeper » Thu Oct 01, 2015 8:59 pm

Forget your IOE at full speed for a while and focus on your throughput (which is basically the ioe of all those clicks that weren't wasted). Try to keep your throughput as high as possisble, preferrably always above 0.8 or even 0.9.

Looking at some of my games on youtube might give you a good idea of what you should aim for. Look at how the ratio of left clicks to double clicks is around 1:1 during slow clicking and close to 1:2 during fast clicking.


What do you think about posting two 5-10 minutes video of you playing at full speed and of you trying slow clicking. I'm sure tkolar and I could give some insights.

Another thing I used to do for other players was an in-depth analysis of a single game where I pointed out every single thing I noticed about their game.
„Das perlt jetzt aber richtig über, ma sagn. Mach ma' noch'n Bier! Wie heißt das? Biddä! Bidddää! Biddddäää! Reiner Weltladen!“

pauv
Posts: 78
Joined: Wed Nov 20, 2013 6:06 am

Re: Switch from NF to FL

Post by pauv » Thu Oct 01, 2015 11:32 pm

Take Kamil's 38.599(189) game for example. Time followed is rt. please add "http://www.minesweeper.info/"before each link, this forum doesn't allow too many URLs.
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=exp_3bv189%2838.599%29_Kamil%20Muranski.mvf

0.71s-1.33s: NF*2+skill 1, half example of skill 1
after two clicks on num1, the 1st 1.5click solved bbbv*3(number435), the 2nd 1.5click solved bbbv*1(number4), skill 1 ends since the number435 has implied no further space for fastsolving.

1.78s-2.62s: successful example of skill 1
the 1st 1.5click solved bbbv*3(number111), the 2nd 1.5click solved bbbv*3(number112), then 4 more chords followed since the number111 shows the space for fastsolving. After this, chording on number112 is one of the choice but solving the left side is better, skill 1 ends.

5.91s-6.37s: failed example of skill 1
the 1st 1.5click solved bbbv*3(number322), the 2nd 1.5click solved nothing since the flag is placed adjacent number3, the chording here has shown the reaction of the fail already. Skill 1 ends.

9.03s-9.48s: failed example of skill 2
the flag placed here is aimed at a 1.5click operation, but it's impossible since the number NF here must be at least 2, as the reaction no chording happens. Though it helps at later.

11.16s-11.63s: half-successful example of skill 2
luckily it's 1 here, unluckily it's an opening after 1.5click.

12.24s-12.62s: successful example of skill 2
solved bbbv*6

13.01s-13.83s: typical example of skill 1
the 1st 1.5click solved bbbv*3(number123), the 2nd 1.5click solved bbbv*3(number112), the 3rd 1.5click was to solved nothing since the flag is placed adjacent number2, but the 2 chording followed on the number112, skill 1 ends.

20.70s-21.52s: different form of skill 2
the real skill 2 begins at 20.87s, but the cell NFed at 20.70s is also possible to be number1(it proved to be) and the cell NFed at 20.87s is possible not to be number1, so the 1.5click here placed the 1st chording at the 1st cell, the 2nd chording placed at the 2nd cell. Skill 2 ends, and "spam" chording is to make sure nothing missed.

21.67s-22.53s: another different form of skill 2
no matter what number it is, NF*1, flag*2 and "spam" chordings followed. You can react when flaging, if it isn't number2 you can give up in time. You will find similar technic at 31.133(109) game(the expert time WR), around 15s-17s.
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=exp_3bv109%2831.133%29_Kamil%20Muranski.mvf

23.52s-24.16s: it the same with 21.67s-22.53s

25.17s-25.41s: failed example of skill 2
the place of NF is possible to be number1, so it's possible to solve bbbv*6 without reaction. The cell is proved to be number3 but it doesn't waste very much time and helps recognize the pattern later(around 28s).

29.51s-29.91s: failed example of skill 2

30.75s-31.53s: successful example of skill 2
solved bbbv*4

31.68s-32.47s: it the same with 21.67s-22.53s
what's different: with the followed chordings bbbv*9 is solved without reaction.

33.46s-34.55s: failed example of skill 2+successful example of skill 2

37.07s-37.50s: successful example of skill 2
-----------------------------------------------
What I want to express here is: these two skills are very very useful and you can find them everywhere in Kamil's replays. No other guy use skill 1 so frequently, and even Xian-Yao Zhang isn't so familiar with skill 2 like Kamil. Look at the beginner 3bvs WR again please, it's just the combination of these two skills and different forms of them.
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv25%282.598%29_Kamil%20Muranski.mvr

More examples of Kamil: (flag-chord-flag-chord-chord-chord)
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv13%281.415%29_Kamil%20Muranski.mvr
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv17%281.847%29_Kamil%20Muranski.rmv
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv18%282.07%29_Kamil%20Muranski.avf
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv22%282.472%29_Kamil%20Muranski.mvr
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv27%283.087%29_Kamil%20Muranski.mvr
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv30%283.31%29_Kamil%20Muranski.avf
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv33%283.44%29_Kamil%20Muranski.avf

Solving of other guys: (flag-chord-chord-chord-chord)
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv24%282.79%29_Yu%20Jin%20Yang.avf
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv28%283.47%29_Hong%20Yu%20Yi.avf
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=beg_3bv29%283.46%29_Hong%20Yu%20Yi.avf
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=int_3bv121%2821.01%29_Zhou%20Ke.avf
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=int_3bv110%2820.07%29_Zhou%20Ke.avf

@EWQMinesweeper your solving is much more like the f-c-c-c-c guys, but you have some command of skill 2.
Yu Suzuki has some command of skill 1, but most of time in a different form:
videoindex.php?dir=LinJinFan/Record%20of%20BBBV/World%20Record/All/&file=exp_3bv239%2850.92%29_Yu%20Suzuki.mvr
Pavel Mishin knows both of the skills but the most of time he belongs to f-c-c-c-c guys:
members/files/2717/3BVS-Pavel%20Mishin-47,784-219-4,681-020312.mvf
Last edited by pauv on Sun Oct 04, 2015 12:57 am, edited 1 time in total.

Zlix
Posts: 38
Joined: Fri Jun 13, 2014 12:29 pm

Re: Switch from NF to FL

Post by Zlix » Fri Oct 02, 2015 4:25 pm

My throughput is over 0.8 almost always.
It's too early to post videos. There are so many things I should improve before, and only time can help.
One of the main reasons I switched to FL is that I reached a level when I can solve a board faster than I can click. Yes, I was getting faster every month, but those were baby steps... When I come remotely close to that level, I'll post videos.
For the moment I play as fast as I can solve. There is no full speed or slow because I'm currently slow enough to play for IOE because I can't solve it faster..

@pauv
Thank you for taking so much time and effort to give me examples. I'll watch them tomorrow.

Zlix
Posts: 38
Joined: Fri Jun 13, 2014 12:29 pm

Re: Switch from NF to FL

Post by Zlix » Sun Oct 04, 2015 6:40 pm

Improved 3BV/s to 2.95, than to 2.98. Here is link with those 2 vids and one fail.. Tell me your opinion.
https://www.mediafire.com/?iee52ce36n5k1qa

pauv
Posts: 78
Joined: Wed Nov 20, 2013 6:06 am

Re: Switch from NF to FL

Post by pauv » Sat Oct 10, 2015 5:45 am

The blast has shown your pottential to be a good player. You have applied skill 1, try to be as fast as possible. ;) What tk said is enough so I think it is better to give you further advice after you have been familiar with FL operations. BTW I agree with EWQMinesweeper that Thrp is the benchmark you need to care about, in fact in my first reply here I mean you don't need to worry about spam chords so that IOE is not important, so don't hesitate.

Post Reply