Challenge: Build the fastest glider (and win something!)

[canidae]

The glider should run for a certain time before scoring starts, this to eliminate any bonus the glider may gain by the initial setup.

I think I understand canidae's evil setup plan.

Actually, I have no plan behind this idea, I just noticed that it's likely possible to take advantage of the scoring starting immediately

[Nemek]

The glider should run for a certain time before scoring starts, this to eliminate any bonus the glider may gain by the initial setup.

I think I understand canidae's evil setup plan.

Let's say we have a glider that is 200 tiles wide using max spaced roboports.

I logged in to ask the same: if there was f.e. a width limit to the glider in order to avoid some exploits.
Otherwise the best glider will just be the 2-days long one. We could make different challenge categories: fastest glider that fits in a chunk, fastest glide with no size limitations, etc.

EDIT

A simple rule suggestion to prevent the initial setup exploit:
"Gliders that will have an higher average speed in the first days because of their initial setup, will be disqualified, at NiftyManiac's discretion."

That's part of Creative Mode mod.

Thank you, now everything makes sense.

[DaveMcW]

Thinking some more, a 200 tile glider already gets a 200 tile penalty to the final score. So this is only exploitable by, say, a 50-tile glider that temporarily extends to 200 tiles at the start.

[mattj256]

Thinking some more, a 200 tile glider already gets a 200 tile penalty to the final score. So this is only exploitable by, say, a 50-tile glider that temporarily extends to 200 tiles at the start.

Is that true? This is the relevant part of the rules:

Final score is similar, but based on your right-most entity at the start, and left-most entity at the end: the full distance you crossed. This will penalize really wide gliders.

Anyway I'm not going to waste my time on something I consider cheaty, and I hope NiftyManiac will update the rules if anyone's entry gets an artificial bonus from taking advantage of a quirk in the scoring function. (If the speed after one full day is significantly different than the speed after two full days that may be an indication that something is wrong with the scoring function.)

[Yoyobuae]

I logged in to ask the same: if there was f.e. a width limit to the glider in order to avoid some exploits.
Otherwise the best glider will just be the 2-days long one. We could make different challenge categories: fastest glider that fits in a chunk, fastest glide with no size limitations, etc.

My glider is 10x4 at start. While running it's about 20~25 tiles or so wide (unless bots "forget" to pick some stuff up then need to go back and pick it up -.-).

Anyway I'm not going to waste my time on something I consider cheaty, and I hope NiftyManiac will update the rules if anyone's entry gets an artificial bonus from taking advantage of a quirk in the scoring function. (If the speed after one full day is significantly different than the speed after two full days that may be an indication that something is wrong with the scoring function.)

I don't think there's anything wrong with the scoring function.

Everyone can do the same. This is a competition of figuring out tricks to get whatever slight advantage is possible. It is also a competition of making it work reliably. Reliability has been my worst enemy so far. :-/

Theories are nice and all, but taking them to practice is a whole different story.

[hansinator]

My glider is 10x4 at start. While running it's about 20~25 tiles or so wide (unless bots "forget" to pick some stuff up then need to go back and pick it up -.-).

What, so small? But you've got more stuff in a chest? I can't wait to see your solution when the competition is over
I wonder if it might be an advantage to start with one gilder and construct the first copy from chests instead of starting with two and deconstructing the left one first. My first thought was that both ways should end up with equal score.

This is a competition of figuring out tricks to get whatever slight advantage is possible. It is also a competition of making it work reliably. Reliability has been my worst enemy so far. :-/

I agree. There are numerous things to try to get an advantage, I really enjoy it. Making gliders that work reliable hasn't been a problem for me so far. From my perspective it is reliability versus speed and the challenge is to maximize both.

[Yoyobuae]

My glider is 10x4 at start. While running it's about 20~25 tiles or so wide (unless bots "forget" to pick some stuff up then need to go back and pick it up -.-).

What, so small? But you've got more stuff in a chest? I can't wait to see your solution when the competition is over

It starts with a bunch of stuff in chest.

[justarandomgeek]

Well, it's not technically a glider (it's still only semi-automatic, and requires periodic exchange of items and a power tether), but I think you'll all enjoy it anyway:



A rolling wall, for clearing land on a waterless ribbon world.

[canidae]

My glider is 10x4 at start

That's HUGE!
Mine is 6x4 at start. Size doesn't matter!
And yes, it's a fine balancing between speed and reliability. I have one trick to make it pretty much 100% reliable, but that reduce speed with about 7-8%.
I kinda hope there will be arranged a new contest where the task is for example building a glider that lays railroad tracks (like FARL). I know it's possible to submit such gliders to this contest, but I think we need a real competition to come up with clever designs.
Actually, let's not have another such competition. I've already spent way too much time on perfecting the racing glider

[hansinator]

Well, it's not technically a glider (it's still only semi-automatic, and requires periodic exchange of items and a power tether), but I think you'll all enjoy it anyway:
[..]
A rolling wall, for clearing land on a waterless ribbon world.

Haha nice! Maybe this one is useful in a multiplayer game.. Just launch some of these on the enemy bases
Edit: Or launch 4, one in each direction and tweak it with ConMan such that in each iteration the width increases so it always forms a closed square. Make sure it leaves railway paths from the center so you can build mining outposts in the newly cleared areas and constantly feed in supplies. Then watch it conquer an exponentially growing area. Muahahaha *evil laughter*

My glider is 10x4 at start

That's HUGE!
Mine is 6x4 at start. Size doesn't matter!
And yes, it's a fine balancing between speed and reliability. I have one trick to make it pretty much 100% reliable, but that reduce speed with about 7-8%.
I kinda hope there will be arranged a new contest where the task is for example building a glider that lays railroad tracks (like FARL). I know it's possible to submit such gliders to this contest, but I think we need a real competition to come up with clever designs.
Actually, let's not have another such competition. I've already spent way too much time on perfecting the racing glider

The size of mine is more like the OP's sample submission. Hopefully it'll get scored soon so I know how it compares to yours.. I've too spent some hours trying different stuff and optimizing speed. I could optimize further but first I want to know if I need to or if it is enough . I wonder if I can build a glider that shoots out smaller gliders like a moving glider gun.

Yes a real competition is a good thing. All those clever people building these clever and or complicated designs suddenly have a common goal they find interesting and try to beat each other. Maybe the Factorio staff can be convinced to host such competitions on a monthly or quarterly basis. It suspect it's a good way to keep the game interesting and challenging and spread the word.

[justarandomgeek]

Well, it's not technically a glider (it's still only semi-automatic, and requires periodic exchange of items and a power tether), but I think you'll all enjoy it anyway:
[..]
A rolling wall, for clearing land on a waterless ribbon world.

Haha nice! Maybe this one is useful in a multiplayer game.. Just launch some of these on the enemy bases
Edit: Or launch 4, one in each direction and tweak it with ConMan such that in each iteration the width increases so it always forms a closed square. Make sure it leaves railway paths from the center so you can build mining outposts in the newly cleared areas and constantly feed in supplies. Then watch it conquer an exponentially growing area. Muahahaha *evil laughter*

Well, it currently can only copy the wall (and control zone) exactly as it is (it's reusing the same blueprint to copy&paste areas across), but i did tweak it some last night so it now has auto-advance mode. Can't decide now If I should hook in the circuit i've got to craft chests from the waste wood and shit out boxes of trash, or if I should hook it up with an auto-extending rail line for item exchange!

[hansinator]

Well, it currently can only copy the wall (and control zone) exactly as it is (it's reusing the same blueprint to copy&paste areas across), but i did tweak it some last night so it now has auto-advance mode. Can't decide now If I should hook in the circuit i've got to craft chests from the waste wood and shit out boxes of trash, or if I should hook it up with an auto-extending rail line for item exchange!

I'm missing the big picture and have problems to follow your description and to put it together in my imagination to understand what it means exactly. It sounds interesting . I thought it is advancing automatically already. I do not yet fully grasp what items need to be exchanged. It does not consume a lot of items, right? I imagine it needs power poles or substations to extend the power tether. Bringing those in via robots, belts or train over a long distance consumes more materials to extend the supply line. I think it's wisest, bandwidth and speed wise, to bring those things in via train. Wood, stone and alien artifacts seem to be the only trash it produces. Leaving it behind in automatically crafted chests sounds like the best idea.

Maybe we can team up to add features. I am good at logic and combinators. If you have a specific sub-function in mind and can give a clear specification of what it must do I could develop it independently and share the result with you. I have lots of other projects in Factorio already, but a small side-challenge sounds like a good idea

[justarandomgeek]

Well, it currently can only copy the wall (and control zone) exactly as it is (it's reusing the same blueprint to copy&paste areas across), but i did tweak it some last night so it now has auto-advance mode. Can't decide now If I should hook in the circuit i've got to craft chests from the waste wood and shit out boxes of trash, or if I should hook it up with an auto-extending rail line for item exchange!

I'm missing the big picture and have problems to follow your description and to put it together in my imagination to understand what it means exactly. It sounds interesting . I thought it is advancing automatically already. I do not yet fully grasp what items need to be exchanged. It does not consume a lot of items, right? I imagine it needs power poles or substations to extend the power tether. Bringing those in via robots, belts or train over a long distance consumes more materials to extend the supply line. I think it's wisest, bandwidth and speed wise, to bring those things in via train. Wood, stone and alien artifacts seem to be the only trash it produces. Leaving it behind in automatically crafted chests sounds like the best idea.

Maybe we can team up to add features. I am good at logic and combinators. If you have a specific sub-function in mind and can give a clear specification of what it must do I could develop it independently and share the result with you. I have lots of other projects in Factorio already, but a small side-challenge sounds like a good idea

What I posted last night was advancing with a button push. It now has a toggle to "advance when ready". It leaves a tail of Substations behind it to keep the power tether going, and will accumulate wood, stone and aliengoo over time which need to be removed periodically or the storage will clog up. It will also eventually need laser, conbot & repairpack replacements probably, as the biters do occasionally eat a few. I have a circuit already that will spawn pre-filled assemblers (and decon them later to claim the product - no logbots in this net) to turn some of the waste wood into chests, but that doesn't solve the problem of getting new stuff.

I think I can handle the circuits for now

Edit: and "it can only copy as-is" was in response to the suggestion of extending the wall as it runs to make a growing-box. I do plan to do an any-side expanding box at some point, but that will be a program for the computer,not a standalone circuit.

[nielskool]

Darn, hadn't any time to improve my glider anymore.... my latest design(not submitted) was still 0.3m/s slower in daytime than canidae's final score

love to see more of these competitions.

[canidae]

Darn, hadn't any time to improve my glider anymore.... my latest design(not submitted) was still 0.3m/s slower in daytime than canidae's final score

Meh. For some reason I thought there was one more day of the competition. I think I figured out how to beat Dave's item count (by 2) and planned to give it a go today after work, but I see I'm too late
I kinda wish the competition lasted a bit longer so I could complete my final glider, but two weeks really should be enough. The last glider I submitted contains at least 12 more items than necessary (possibly 14), but I didn't have time to finish my next glider. Teaching the construction bots to behave properly is tedious work
Oh well, I still had tremendous fun with this competition. I'm looking forward to the final score of this competition, and I hope to see more of these competitions in the future

Also a big thanks to NiftyManiac for hosting this competition. I can imagine this was quite a lot of work and I think [s]he did a very good job!

Spoiler, mostly for Dave

While I don't know your design, I suspect you're using two constant combinators. I think you can drop one constant combinator (and thus also one construction bot) by sending Available construction robots signal from a Roboport as a Construction robot signal and use this to deploy the blueprint.

Edit:
Just after I wrote this I noticed that Nemek now claims the #1 spot!
Now I'm really anxious to see the final results. I've been fairly certain that at least Dave, Yoyo and myself have quite similar gliders, but now I'm beginning to wonder if Yoyo and Nemek came up with a completely new idea.

[DaveMcW]

Nope, I already used the 1-combinator trick. Sadly, even a perfect 2-part design is not enough to win.

I tried a few tricks to bring the chaos of 3-part designs under control, but ran out of time before I could find one that runs 50000 ticks without breaking.

[canidae]

Nope, I already used the 1-combinator trick. Sadly, even a perfect 2-part design is not enough to win.

How about deployer chests, are you using two of those? I think it's possible to both construct and deconstruct using the same deployer chest (I thought about this before the 1-combinator trick, it didn't occur to me at first that you can send conbot signal with a value greater than 1 and it'll still work)... I didn't have time to test this out properly though.

[Nemek]

I agree with canidae: this competition was incredibly fun and I learned a ton about factorio, logistic system, bots behaviour. I'm looking forward to see all the designs and who the final winner will be. Yoyo and I are very close so the result will be decided by the smallest tweaks done and a ton of luck. Gliders love to go bananas.

Anyway I had still room for improvements, and I think we should keep working to make the perfect glider together. And keep doing competitions like this one: the best tree cutter glider, smaller, etc. Of course not only related to gliders.

Thanks in advance to NiftyManiac for hosting this competition in such a clear, very well-thought way.

You don't need combinators in order to have fun with gliders.

[nielskool]

Spoiler, mostly for Dave

While I don't know your design, I suspect you're using two constant combinators. I think you can drop one constant combinator (and thus also one construction bot) by sending Available construction robots signal from a Roboport as a Construction robot signal and use this to deploy the blueprint.

haha nice, didn't think about that one.... so obvious!

[offtopic]

I think I can handle the circuits for now

ahhh ;( you already made a computer... was about to make a cpu. Looks awesome though.
Will have to think of an other challenge then.

[Yoyobuae]

Finally it's done.

I had a final glider which I coudn't fix reliability problems due to bot randomness so I didn't enter it into the competition (only like one run in five can make it thru without errors):

Screenshot_2017-03-06_10-34-24.png (205.95 KiB) Viewed 7865 times

Score also varies wildly, I got 5.9 m/s on a previous run. :-/

Savefile with glider ready to run:
https://www.dropbox.com/s/espg9c31rbkfg ... 3.zip?dl=0

To run take the second (from the left) blueprint from hotbar and place it on requester chest.

The one thing I learned from all this: Bots are SOOOOOO annoying. I don't want to deal with them at this level of detail ever again. I like inserters and belts better, they are fully deterministic at least.