How do we prevent miraculously true DC results from manual submit?

leonardyan96 · 2017-05-13, 15:16

Well, after a first LL check for a exponent the residue will be publicly shown with the last 2 digits hidden.
If someone makes up many fake residues with correct first digits and random last hidden digits, without any actual DC, and then repeatedly submits these fake DC results manually for the same exponent, I think it's possible that, after he accidentally submits the "right" residue exactly the same as the first LL result, the fake DC result will be accepted.
So how do we stop that? Are the we8 and AID in the results submitted designed for this use?

retina · 2017-05-13, 16:00

If someone is submitting so many results like that then it won't last long. Mismatched DCs are found easily and quickly and it would be blindingly obvious to see a flurry of results all with only the last two digits differing.

TL;DR Don't worry about it. everything is fine.

kladner · 2017-05-13, 18:21

It might also be considered that the amount of work required to cheat might be spent with more rewards by doing legitimate work.

leonardyan96 · 2017-05-14, 03:03

Quote:

Originally Posted by kladner

It might also be considered that the amount of work required to cheat might be spent with more rewards by doing legitimate work.

But if anyone insist on ruining the results on purpose, the effort to make might not stop him.

leonardyan96 · 2017-05-14, 03:16

Quote:

Originally Posted by retina

If someone is submitting so many results like that then it won't last long. Mismatched DCs are found easily and quickly and it would be blindingly obvious to see a flurry of results all with only the last two digits differing.

TL;DR Don't worry about it. everything is fine.

So are such abnormal submission prevented by some software mechanism or just by administrator's manual review to the database?
I'm just afraid of missing a prime due to an error LL although I know the probability is extremely small...

retina · 2017-05-14, 03:17

Quote:

Originally Posted by leonardyan96

But if anyone insist on ruining the results on purpose, the effort to make might not stop him.

But it also wouldn't work, it would be caught immediately.

There are much more effective ways to "ruin" the results (although for what purpose I can't figure out). And who could possibly profit from such an effort? Where is the threat coming from? I think you are over imagining the worth of the results here.

retina · 2017-05-14, 03:21

Quote:

Originally Posted by leonardyan96

So are such abnormal submission prevented by some software mechanism or just by administrator's manual review to the database?

Both. Like I said above, such a glaring set of anomalies would stand out like dogs balls.

Here is a suggestion: Try it yourself and see how long it takes you to get banned.*

* I'm not really serious there, but I'd be curious myself to see how long it takes to get spotted.

Mark Rose · 2017-05-14, 03:41

Quote:

Originally Posted by retina

* I'm not really serious there, but I'd be curious myself to see how long it takes to get spotted.

Nobody's caught me so far hehe

kladner · 2017-05-14, 07:22

Quote:

Originally Posted by Mark Rose

Nobody's caught me so far hehe

GOOD! Probably. Possibly. Maybe. I guess. conceivably....what kind of rewards are you ~~raping~~ reaping from these efforts?

rudy235 · 2017-05-15, 04:07

There are 256 possible combinations of 2 hex digits. So lets say someone is "lucky" and reports 100 "manual" results of which 1 is a correct match and 99 others aren't.

How is he/she going to prevent the other 99 wrong results not sticking out like a centipedede's sore thumb?

Siegmund · 2017-05-15, 15:58

As soon as you submit a doublecheck, matching or not, both residues get displayed full length. (At least on your own screen; I assume on everybody's.)

While there is nothing stopping you from immediately submitting a third result for that exponent, which actually matches - what's the point? As soon as a real live person sees that there was a triplecheck, he'll be suspicious if 2 of the 3 came from the same person.

By comparison, I've mis-typed the residue (or mis-cut-and-pasted it) a time or two when submitting a manual result, and then submitted the correct result a moment later. Thus far George has believed me when I've written to say "hey, I had a typo in my manual result, the machine really did get the right answer."

2017-05-13, 15:16	#1
leonardyan96 "Hazawa Tsugumi" May 2017 Northern China 17 Posts	How do we prevent miraculously true DC results from manual submit? Well, after a first LL check for a exponent the residue will be publicly shown with the last 2 digits hidden. If someone makes up many fake residues with correct first digits and random last hidden digits, without any actual DC, and then repeatedly submits these fake DC results manually for the same exponent, I think it's possible that, after he accidentally submits the "right" residue exactly the same as the first LL result, the fake DC result will be accepted. So how do we stop that? Are the we8 and AID in the results submitted designed for this use?

2017-05-13, 18:21	#3
kladner "Kieren" Jul 2011 In My Own Galaxy! 2×3×1,693 Posts	It might also be considered that the amount of work required to cheat might be spent with more rewards by doing legitimate work. Last fiddled with by kladner on 2017-05-13 at 18:21

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Automatic submit results + fetch assignments for mfaktc?	DuskFalls	GPU Computing	5	2017-12-02 00:34
GPU id/name for manual results	preda	GPU Computing	15	2017-08-16 17:34
MLucas, submit results?	Sleeping_menace	Mlucas	17	2015-06-13 03:12
manual results	ramgeis	PrimeNet	8	2013-05-30 06:33
Only submit part of ECM results?	dabaichi	PrimeNet	5	2011-12-07 19:27

2017-05-13, 16:00	#2
retina Undefined "The unspeakable one" Jun 2006 My evil lair 1100000110010₂ Posts	If someone is submitting so many results like that then it won't last long. Mismatched DCs are found easily and quickly and it would be blindingly obvious to see a flurry of results all with only the last two digits differing. TL;DR Don't worry about it. everything is fine.

2017-05-15, 04:07	#10
rudy235 Jun 2015 Vallejo, CA/. 1111100010₂ Posts	There are 256 possible combinations of 2 hex digits. So lets say someone is "lucky" and reports 100 "manual" results of which 1 is a correct match and 99 others aren't. How is he/she going to prevent the other 99 wrong results not sticking out like a centipedede's sore thumb?

2017-05-15, 15:58	#11
Siegmund Mar 2014 2⁴·3 Posts	As soon as you submit a doublecheck, matching or not, both residues get displayed full length. (At least on your own screen; I assume on everybody's.) While there is nothing stopping you from immediately submitting a third result for that exponent, which actually matches - what's the point? As soon as a real live person sees that there was a triplecheck, he'll be suspicious if 2 of the 3 came from the same person. By comparison, I've mis-typed the residue (or mis-cut-and-pasted it) a time or two when submitting a manual result, and then submitted the correct result a moment later. Thus far George has believed me when I've written to say "hey, I had a typo in my manual result, the machine really did get the right answer."