[tw1010]

This article reminds me of how it felt to be on the internet as a child. It seemed like this magical place where around every corner there was hidden an alchemic recipe to achieve something magical, if only you knew how to search in the right places.

[Waterluvian]

Well said. For me the 90s and early 2000s internet was completely magical. My fondest memories of my teens was finally having a computer in my room and the freedom to stay up late and explore the online universe. I met a lot of people and saw the advent of a lot of new things.

I wonder if that's going to happen for my kids. Today's internet feels so highly commercialized. All the new exciting discoveries are in the form of profit-driven silicon valley products with no humanity or soul.

I may very well be wrong. This is all emotion based and I really don't have concrete evidence. I'm probably falling into a fit of nostalgia.

[NeedMoreTea]

The sad thing for me is that a lot of those strange, eclectic sites are still there, only they became really hard to find. Thanks to Google and other search engines prioritising recency and Brands (with a capital) in all results, the eclectic is penalised to the point of near invisibility.

There was never a point to them, but they were hugely interesting. Today there's no point bothering as it would be gone forever a month after posted to twitter or a Facebook group now bloody everything has to have a stream or timeline. So akin to a chilling effect it invisibly promotes things that have releases. Products.

Nostalgia is some of it I'm sure, but I really do think there's something more concrete there too.

[pishpash]

I don't know that Google was ever the thing to find them. Various curated lists and word-of-mouth were how they spread. But yeah, the signal-to-noise ratio is pretty terrible now for anything that isn't a ho-hum "conventional" query. On the other hand, the sheer volume of information has to count for something. You can reach a greater breadth and depth for directed learning online than in the 90's where it was mostly undirected novelty.

[PhasmaFelis]

> I don't know that Google was ever the thing to find them. Various curated lists and word-of-mouth were how they spread.

There was a period in the mid-'90s at least when you could find really interesting personal pages by punching any given topic into your search engine of choice. It may actually have been petering out by the time Google arrived in '98, though.

[gbhn]

google.com/search?q=human+echolocation brings up a ton of related stuff.

The thing is "what do you type into the search engine." I, for one, would never think to type in any of probably tens of thousands of topics that, having seen, I think are awesome.

[mycall]

Which are your favorites?

[skykooler]

I like this page, on creating 3d 'holograms' by hand:

http://amasci.com/amateur/holo1.html

[MrEldritch]

Oh man, amasci.com was one of my favorite Magical Internet Places back in the day; I can't believe I'd totally forgotten about it.

[wbeaty]

"the time for jacking around with Tesla coils and ball lightning in the garage is over." -Abe, PRIMER

Few additions http://amasci.com/news.html The Trafficwaves article made the WSJ, even with video. And Crown Flash (leaping sundogs) is now all over the place (see my upcoming bit on TV show Strange Evidence.) Magnetic analogy: https://www.youtube.com/watch?v=tH-CAhtXBfQ

Fortunately (unfortunately?) I only posted examples of my somewhat-kid-friendly magic. Nobody died in microwave oven accidents. That plus a couple of minor inventions (made public as a test to see how fast the idea-thieves would pounce on them.) None of my frightening demos using vacuum pumps and Spellman supplies. Well, I guess I did write about our hypergun which fired a small plastic slug through a clay block twelve inches thick (a spinoff of the very first quarter-shrinker capacitor bank,) and no little kids yet tried building one: http://amasci.com/amateur/capexpt.html#shrink

An unexamined backwater is at http://amasci.com/weird/wtext.html

I still haven't tried out the giant version of Lord Armstrong's water-bridge, from an eyewitness report where a liquid "snake" several feet thick was extending itself thousands of feet across desert land during a thunderstorm. If such things are possible, then an HV supply of a few hundred watts and a couple hundred kilovolts should produce something similar (the recent water-bridge experiments use 10 to 20KV supplies at under ten watts, a 2cm bridge, a mm or two in diameter.) Rainwater or ddw on very clean sand? Needs a big Cockroft-Walton diode stack http://amasci.com/freenrg/wasser.html

And the infamous invisible wall at the 3M factory, recently someone claims to have seen an example caused by electrodes connected to a reversed pole-pig transformer, where coins would bounce off the empty space. This is silghtly conceivable: a corona in air may develop a stable annular double-layer with a significant space-charge and pressure-gradient of air. But would a metal object bounce off? Or just short it out? Plasmas do tend to avoid cold bulk metal surfaces (Tesla managed to make an opaque "neon sign" with a metal tube rather than glass, but skeptics refused to believe it was real. The plasma in near-vacuum can travel many inches through thin metal tubes without leaping to the metal! Try it, it's really weird: pipes full of glowing "electricity" rather than water.)

Speaking of PRIMER, did they actually base the plot on meissner maglev "null coils," placed in a circle rather than parallel rows? Then, put it all inside a "One Atmosphere Plasma Globe" full of argon? But then, where do platinum catalytic converters figure in that? Or maybe they were just huddling inside a styrofoam version of the USS Eldridge hull, covered with active magnetic shielding coils, but not turning invisible.

:)

[hnzix]

I remember Fjord Of The Net Vikings. Dan's Gallery of the Grotesque. I was too young for BBS culture but then I found the Hotline protocol, which gave me a new horizon of musical styles and I even made a music video with some random German dude by mailing CDs across the pond.

Just last night I remembered micromusic.net which introduced me to chiptunes and continues as a ghostship.

[gpm]

It looks like it hasn't been updated in years, but http://www.powerlabs.org/index.html has some cool content

[Gracana]

FWIW, everybody from his forum eventually made their way to 4hv.org.

[nyolfen]

not OP but this is what i thought of instantly:

http://margo.student.utwente.nl/el/microwave/

and this page in particular:

http://margo.student.utwente.nl/el/microwave/mladen_story.ht...

[prolikewh0a]

wiby.me is fun.

[Alex3917]

> I wonder if that's going to happen for my kids.

I've got a four month old. By the time she's in third or fourth grade, I expect that VR will give her the same kinds of experiences that I had with the web at that age. What made the web exciting wasn't just the fact that it was run by amateurs, it was the fact that there was just an enormous amount of possibility waiting to be explored where we genuinely didn't know the answers.

I actually think the web itself may go back into that kind of phase again after a couple more iterations of Moore's Law make programming easy enough for people who wouldn't be able to do it today, especially when combined with blockchain opening up new kinds of relationships that people can have with one another. The web currently has some issues, but I think a lot of people have given up on it prematurely without taking into consideration the new kinds of experiences the recent and upcoming improvements to the underlying infrastructure are going to make possible.

[andrepd]

Please don't be upset that I say this, because you seem to be earnest, but your comment reads like a parody. "VR" and "blockchain" and other buzzwords are new tech o ligues so they necessary are going to be great and revolutionary (in some ill specified way) and therefore will provide "experiences" and "revolutionise the way we do x". Why, you sound like an SV "evangelist" :^)

[nikofeyn]

is this a serious comment?

i am not that old and not that young, and VR has been "around the corner" for 10 to 15 years already. even with all the hype and the amount of people that have been feverishly working on it, there's basically nothing to show for it. it will come eventually, but not in any useful form anytime soon.

what does moore's law have to do with ease of programming?

blockchain has no relevance to the common person.

[blhack]

>is this a serious comment?

Is this a serious comment? VR isn't around the corner, it's here. I'm sitting in my living room right now looking at the vive lighthouses on the wall, and my wife and I frequently use VR for all sorts of things.

[Alex3917]

Exactly. Based on what already exists, my best guess is that it will actually be good after another 2 - 3 iterations of Moore's Law and another 10 years of software and hardware architecture improvements.

[blhack]

It's better than good now. It's incredible now.

[nikofeyn]

> my wife and I frequently use VR for all sorts of things.

for what? you didn't give any examples except the demonstration of the existence of VR products and VR "stuff".

[MrEldritch]

I think you're totally right that it feels like that wonderful old Web, the one that felt like a magical place, a weird place, seems to have vanished.

For me, some of that's moved on to YouTube. Even today, I find channels full of crazy stuff, brilliant stuff, beautiful stuff, educational stuff, stuff I'd never even heard before, just randomly stumbling around. It's hard to find them, what with the suggestions throwing only the lowest-common-denominator version of the stuff I get baited into clicking on, but sometimes I manage to run across it.

[samstave]

>* My fondest memories of my teens was finally having a computer in my room and the freedom to stay up late and explore the online universe*

This is something i think about regarding my kids... specifically that my experience was free and unfettered - theirs is contolled, walled and full of ads and bullshit.

[andrepd]

No, I think that very sadly you're 100% right. They won. The internet is commercial. The internet is the corporation's, not the people's. The early, magical, futuristic Internet is gone for good. All we have now is a bastardized version of that ideal of global network.

[acct1771]

It's not gone. Just underground/waiting.

[closed]

So true! This reminded me of the first time I came upon Conway's Doomsday Algorithm for telling what day of the week a certain date is.

http://rudy.ca/doomsday.html

[tzs]

The method given on that site for computing the component that comes from the last 2 digits of the year can be made a lot easier for mental calculation. I'll give some known improvements below, and then offer one of my own that is noticeably simpler.

In the following, let Y = the last two digits of the year (e.g., 18 for 2018), and let N be the value we are trying to be compute. We only actually need N mod 7, but I'm going to leave out the reduction mod 7 in the equations to reduce clutter.

The simplest way to compute N is simply:

  N = Y + Y//4

I'll use Python3-like arithmetic and pseudocode in this comment, so "//" is integer division (with x for times to avoid accidental italics).

To keep the numbers smaller during mental calculation, people have developed alternatives. The one given in that link is this:

  N = Y // 12
  N += Y % 12
  N += (Y % 12) // 4

An interesting alternative, called the "odd+11" method, is given on the Wikipedia article [1]:

  if odd(Y): Y += 11
  Y = Y // 2
  if odd(Y): Y += 11
  N = -Y

For the last step there, N = -Y, it will usually be easier and clearer to reduce Y mod 7 before doing that N = -Y. Also, given Y, sometimes the simplest way to get -Y mod 7 is to just note what you have to add to Y to get to a multiple of 7. For example, if when you get to that step Y = 20, note that adding 1 to Y gives a multiple of 7, so -20 mod 7 = 1.

Anyway, here's my method. It keeps the numbers smaller--if you reduce mod 7 aggressively never more than 12--at the cost of slightly more branches in the logic.

Let the last two digits of the year be T and U, so Y = 10 T + U.

  N = 2 x T
  if odd(T): N += 3
  N += U
  if odd(T):
    N += (U+2)//4
  else:
    N += U//4

That if...else is taking into account the number of leap years that have occurred in the current decade (not including the year T0). When doing mental calculation, it is probably easier just to remember that if T is odd, at 1 if U >= 2 and add another 1 if U >= 6, and if T is even same except at 4 and 8.

Here are examples, using some years from the link, with parenthetical explanations for some of the numbers:

2018: T=1, U=8. 2x1(T) + 3(T is odd) + 8(U) + 2(U>=2,6) = 1 mod 7.

1929: T=2, U=9. 2x2(T) + 9(U) + 2(U>=4,8) = 1 mod 7.

1999: T=9, U=9. 2x2(T%7) + 3(T is odd) + 2(U%7) + 2(U>=2,6) = 4 mod 7. Note that I reduced U and T mod 7 inline when using them. You can do this as long as when checking odd/even you use the original T, and when adding in the leap year correction you use the original U. E.g., for 99, you could compute it like it was 22, except you have to add the 3 for odd T, and use 9 for the U>=2,6 check.

1982: T=8, U=2. 2x1(T%7) + 2(U) = 4 mod 7.

1969: T=6, U=9. 2x(-1)(T%7) + 2(U%7) + 2(U>=4,8) = 2 mod 7.

Some might find changing the leap year handling to this a little easier instead of just remembering the 2,6 or 4,8 +1 points. Change the if...else to this:

  N += U//4
  if odd(T) and N%4 >= 2: N += 1

E.g., compute the leap year adjustment with U//4 regardless of whether T is odd or even, and if U is in {2, 3, 6, 7} and T is odd, add one more.

[1] https://en.wikipedia.org/wiki/Doomsday_rule#The_%22odd_+_11%...

[tzs]

Bonus: here's how to compute the day of the year from month/day. First I'll show how independent of the Doomsday stuff, and then show how you can use the Doomsday month number stuff to reduce memorization for the day of year stuff.

The basic idea is to compute the year day for the "0th" day of the month, then add the day of the month. For example, today, October 13th is the 286th day of the year, because October 0th is the 273rd day of the day, so 13 days past that is the 286th.

The simplest way, short of memorizing the 0th day of each month directly (or the 1st day...if you are going to take this approach might as well do the 1st, not the 0th), is to memorize how far off the actual 0th day is from what the 0th would be if all months were exactly 30 days.

Those differences are, for the 12 months:

  0, -1, 1, 0, 0, 1, 1, 2, 2, 3, 4, 4

If F[m] refers to the the m'th entry in that list, 1 <= m <= 12, then the 0th day for month m is 30(m-1) + F[m]. E.g., for October, m = 10, so 30x(10-1) + F[10] = 30x9+3 = 273.

If it is a leap year, add one more day for dates after February 29.

The F table has enough structure that it is not unreasonable to just memorize it.

For m >= 4, F[m] = (6x(m-4))//10. If you round toward -∞ instead of 0 when dividing, that works for m >= 3. It's probably harder to remember this than to just remember the table.

Let M[m] = the month factor from the Doomsday algorithm:

  3, 0, 0, 4, 9, 6, 11, 8, 5, 10, 7, 12

Then F[m] = -(M[m] + 2 x m + 2) mod 7. Just remember that -1 <= F[m] <= 4, so you can pick the right range when reducing mod 7.

For example, let's do July 20th, 1969, the day of the first Moon landing. July is the 7th month, and the month factor for that in the Doomsday algorithm is 11, so we have F[7] = -(11 + 2x7 + 2) = 1 mod 7, so F[m] = 1, and July 0th is 30x(7-1)+1 = 181, and so July 20th is the 201st day of the year.

BTW, if you find F[] easier to memorize than M[], you reverse the formula given above for computing F from M, to go the other way.

M[m] = -(F[m] + 2 x m + 2) mod 7.

While I'm here, one more Doomsday observation.

If a given year, Y, has year number N, then the next year with the same year number is:

• Y+6 if Y%4 == 0 or 1

• Y+11 if Y%4 == 2

• Y+5 if Y%4 == 3

(but only if that does not cross a century boundary).

The day of weeks within a century go on a 28 year cycle. Within any 28 consecutive years in a century, each of the 7 possible year number values occurs 4 times, once on a year with Y%4 == 0, once on a Y%4 == 1 year, and so on.

If you start with Y%4 == 0 year, the gaps between consecutive years with the same year number are 6, 11, 6, 5, which brings you back to the Y%4 == 0 year for the next 28 year cycle. You visit the years in the order 0, 2, 1, 3 mod 4 stepping this way.

If you memorize this pattern, it is easier to answer questions about when years or dates with specific properties happen. For example, let's say you want to know when there will next be a Friday the 13th in October.

Assuming that there will be another one this century, so using the century factor of 2 (for 20xx), you can work out that in this century Friday October 13 occurs when the year number is 0. Such a year was 2000, and 00 is a Y%4 == 0 year, so Friday October 13th occurs in 2000, 2000+6=2006, 2006+11=2017, 2017+6=2023, and then 2023+5 brings is to the 2028, the next 28 year cycle. So the next Friday October 13 is 2023.

BTW, the first Y in a century with year number N and with Y%4 == 0 is (3xN%7)x4.

If you don't care about starting from a Y%4 == 0 year, then it is easiest probably to just memorize the first year with year number 0, 1, 2, ..., 6 in a century:

  0, 1, 2, 3, 9, 4, 5

and then add multiples of 28 to get to where you want to be. E.g., if you were trying to find the first time your birthday falls on a Thursday after 2040, and figured out you need year number 4, that table would tell you 2009 is such a year, and the 28 year cycles would tell you 2037, 2065, and 2093 are other such years, so you'd probably go to 2037 and work forward from there. 36 is a 1%4 year, so the next with the same year number is 2043, and you are done!

[FigmentEngine]

or Zeller's congruence https://en.m.wikipedia.org/wiki/Zeller%27s_congruence

[closed]

Ah, I had no idea about this one!

Bonus: I just learned the word proleptic...

[rubyfan]

That’s why I like HN. I visit only a few news sites per day and HN usually brings a treat that I never would have stumbled on myself.

[turdnagel]

I do like the commenting on this site but is there a way to collapse comments without downvoting? The fact that I have to scroll halfway down the page to find comments actually relevant to the article is a major flaw with HN's threading system.

[grzm]

The [-] in the comment header is clickable and will collapse the comment and any replies in its thread.