Skip to main content

Timeline for answer to What's the origin of the term 'seed'? by Robotnik

Current License: CC BY-SA 4.0

Post Revisions

26 events
when toggle format what by license comment
S 3 hours ago history suggested Elements In Space CC BY-SA 4.0
improved formatting for list in footnote (with blank line after `<sub>`), some re-wording, changed “do a seed” to “use a seed”, removed some odd commas, upgraded some hyphens to em-dashes, added missing colon
4 hours ago review Suggested edits
S 3 hours ago
17 hours ago history edited Robotnik CC BY-SA 4.0
Whoops. I smort, do numbers good. But yeah, I changed the formula before posting but missed a couple of values. How did noone call this one out lol.
18 hours ago history edited Robotnik CC BY-SA 4.0
deleted 18 characters in body
20 hours ago comment added fyrepenguin @AustinHemmelgarn regarding the strings of crits needing to not be random to feel "random", Moze in Borderlands 3 has a skill called Short Fuse which actually uses a fully predetermined sequence to determine whether the skill will proc or not. From what I found, the start of that is "11000000100000101110", which is actually pretty neat since it actually means that the first two instances of damage after loading into the game are guaranteed to activate the skill (and this is utilized to do consistent one-hit-kills of bosses).
21 hours ago history edited Robotnik CC BY-SA 4.0
Clarify footnote based on feedback
22 hours ago history edited Robotnik CC BY-SA 4.0
Typo. We use 101 in the formula examples
yesterday comment added murgatroid99 @AlexanderThe1st The fixed point is not a general property of PRNGs or even LCGs, it is a consequence of the specific parameters chosen for the example. The Linear congruential generator Wikipedia article describes how the choice of those parameters affects the behavior, and it specifically notes that there are choices of parameters for which every seed has the full period.
yesterday comment added Alexander The 1st @Robotnik: Regarding footnote 2 - is it always true (Either for LCGs or PRNGs) that the "fixed point" behaviour is always present? That is, is it impossible to make the seed be able to be 16 in this case without it returning the next value as 16? That is, it kind of comes across as an "anti-seed" of sorts for that LCG, for lack of a better term to come to mind.
yesterday history edited Robotnik CC BY-SA 4.0
Clarify its not impossible to get, you have to use a specific seed value
yesterday comment added voided @Robotnik Very nice answer :)
yesterday vote accept voided
yesterday comment added pinckerman Just a note, if you are doing x % 100 you are not generating numbers between 0 and 100, but between 0 and 99
yesterday comment added Joachim 42, yet no digression into RNGsus? I thought all computer science was inherently mystic.
yesterday comment added Austin Hemmelgarn @murgatroid99 ‘Good’ in the context of a PRNG is at least partly dependent on the use case. For example, a mathematically ‘good’ PRNG is actually pretty terrible for random selection of the next song to play in a music player because it leads to far more frequent repeats than a typical user will want (and writing a good shuffle algorithm without just randomizing playlist order is in fact a difficult project). Similarly, a ‘good’ PRNG to use for calculating whether an attack is a critical hit in a game should ideally not produce long runs of crits or no crits.
yesterday comment added pinckerman 42 mentioned, yay! Also +1 for computer science degree
yesterday history edited Robotnik CC BY-SA 4.0
Clarify that random values are then used to generate more random values
yesterday comment added infinitezero @Murgatroid99 depends on what you call slow. Compared to a PRNG? Yes. But at least this website claims a homebrew setup with a rate of 18Mbit/s of randomness. So depending on how much random numbers you need, generating them asynchronously might always be an option. Definitely enough for gaming.
2 days ago comment added Raven Dreamer @murgatroid99 I think I overloaded my terminology - I agree that pesudorandom number generators should be statistically indistinguishable. My larger point was - sometimes "mostly-random" makes for better gameplay. (e.g. Fire Emblem's various 2RN systems)
2 days ago history edited Robotnik CC BY-SA 4.0
I ended up not needing point 3
2 days ago history edited Robotnik CC BY-SA 4.0
fix a few things - expand e.g. to 'for example' so that the preceding formula is all on one line. Range is actually 0-100; Add more info on this specific PRNG
2 days ago history edited Robotnik CC BY-SA 4.0
Clarify that computing determinism is not always the case
2 days ago comment added murgatroid99 Good pseudorandom number generators generate sequences that are statistically indistinguishable from truly random numbers, so if it feels "more random" than actually random numbers, that is a design failure.
2 days ago comment added Raven Dreamer Conversely, there's also the saying, "The generation of random numbers is too important to leave to chance!". Since Humans are bad at estimating "randomness", pseudorandom numbers can lead to better feelings of "randomness" than actually random numbers! That's the exact kind of gamefeel that's relevant in games.
2 days ago comment added murgatroid99 One thing I would add here is that it is technically possible for real computers to produce numbers that are for all intents and purposes random by using other hardware, such as by measuring thermal noise. However, those methods produce random sequences very slowly, so they are impractical for most applications, and when they are used, it is usually just to generate a PRNG seed.
2 days ago history answered Robotnik CC BY-SA 4.0