TIL: Today I Learned

I found this place today: http://regexr.com/

For somebody who only occasionally (VERY occasionally) uses regular expressions, this place is a godsend. A nice little sandbox that tells you what matches, what doesn't, etc. The "quick help" is good too as long as you already know a bit about what you're trying to do.
 
TIL that, according to this site, there are only 139,713 women in the US who fulfill my preferred criteria when it comes to dating, or approximately 0.043% of the total population of the US. That seems surprisingly low. I had no idea I was so picky.

I'm sure my wife will be relieved.
 
Don't you, y'know ... NOT live in the United States?

--Patrick
I don't, no, but I think it serves as a handy illustration of the fact that there are so few women out there for me, which means she really is the special one, you know?

(Ironically, based on the criteria I gave to the site, my wife would not have qualified...)
 

GasBandit

Staff member
TIL that, according to this site, there are only 139,713 women in the US who fulfill my preferred criteria when it comes to dating, or approximately 0.043% of the total population of the US. That seems surprisingly low. I had no idea I was so picky.

I'm sure my wife will be relieved.
That's not so hard to believe, really... looking at it, just saying "women over 18" narrows it down to 38%, so you start from there. Remove "already in a relationship" and it goes down to 17%. Remove "over 65 years old" and it goes down to 13%. Remove "over 55?" down to 10%. Then you start whittling down to your "preferred" and voila.
 
A

Anonymous

Anonymous

Oversubscription is a term in the ISP industry that provides a metric about actual usage vs what people are sold.

So an ISP that has 3,000 customers, each promised 1Gbps of bandwidth, doesn't actually need a 3,000 Gbps connection to the internet, because their customers are using the internet at a far smaller rate of consumption on average.

This isn't new, and probably doesn't surprise many. However I just found out the average industry rates for oversubscription - once your customer base is large enough, the average customer consumes an average of 2mbps. So those 3,000 customers, each "buying" 1Gbps internet? They actually only budget an ISP uplink of 6Gbps. That ISP actually has a 10Gbps link, but their average usage is lower - 2.7Gbps. The number doesn't change much at different rates - a customer who's buying a 25Mbps still only uses an average of 2Mbps or less. While I suspect the heaviest users cause some consternation and throw things off, the limit, at these speeds, is often the systems/servers the customer is connected to. So if a download is slow the customer usually assumes it's because the server they connecting to doesn't have the bandwidth. Torrenting is meant to overcome this limitation, but even then users usually use it to download a few large things, and it's very bursty - once the large download is done their consumption goes back down.

I knew this was a thing, I just didn't realize it was so low.

The fun thing? Once your ISP gets large enough you can start talking to speed test websites and content providers, and they may come to an agreement with you to put servers in your datacenter specifically to serve your customers. So that speedtest which shows you're getting your full 1Gbps is only showing your connection between you and your ISP. This is useful, but you can't count on it showing your actual connection to the internet, merely the link between you and your ISP. This is one reason netflix made its own speedtest, which shows the speed between you and your assigned netflix datacenter. Eventually you'll be hosting akamai, netflix, amazon, google, and other CDN services in your datacenter specifically to serve your customers so they don't consume your actual uplink banwidth.

Lastly, cable services are almost all on-demand now. There's a small number of channels (estimated 10 or so) which are watched enough to actually be set up with their own assigned bandwidth on the lines, but the remaining 300+ channels in your package are only provided on the line once your cable box tells the cable company you've changed channels, and it takes a little time for them to set it up, so if your cable box takes its time switching to a channel, it's because no one is watching that on the network, and the datacenter has to set it up just for you. The majority of cable channels is on-demand.

I don't think there's much reason to be anonymous, and I'm not under NDA, but I don't want to find out a reason later, soooo...
 
Oversubscription is a term in the ISP industry that provides a metric about actual usage vs what people are sold.

So an ISP that has 3,000 customers, each promised 1Gbps of bandwidth, doesn't actually need a 3,000 Gbps connection to the internet, because their customers are using the internet at a far smaller rate of consumption on average.

This isn't new, and probably doesn't surprise many. However I just found out the average industry rates for oversubscription - once your customer base is large enough, the average customer consumes an average of 2mbps. So those 3,000 customers, each "buying" 1Gbps internet? They actually only budget an ISP uplink of 6Gbps. That ISP actually has a 10Gbps link, but their average usage is lower - 2.7Gbps. The number doesn't change much at different rates - a customer who's buying a 25Mbps still only uses an average of 2Mbps or less. While I suspect the heaviest users cause some consternation and throw things off, the limit, at these speeds, is often the systems/servers the customer is connected to. So if a download is slow the customer usually assumes it's because the server they connecting to doesn't have the bandwidth. Torrenting is meant to overcome this limitation, but even then users usually use it to download a few large things, and it's very bursty - once the large download is done their consumption goes back down.

I knew this was a thing, I just didn't realize it was so low.

The fun thing? Once your ISP gets large enough you can start talking to speed test websites and content providers, and they may come to an agreement with you to put servers in your datacenter specifically to serve your customers. So that speedtest which shows you're getting your full 1Gbps is only showing your connection between you and your ISP. This is useful, but you can't count on it showing your actual connection to the internet, merely the link between you and your ISP. This is one reason netflix made its own speedtest, which shows the speed between you and your assigned netflix datacenter. Eventually you'll be hosting akamai, netflix, amazon, google, and other CDN services in your datacenter specifically to serve your customers so they don't consume your actual uplink banwidth.

Lastly, cable services are almost all on-demand now. There's a small number of channels (estimated 10 or so) which are watched enough to actually be set up with their own assigned bandwidth on the lines, but the remaining 300+ channels in your package are only provided on the line once your cable box tells the cable company you've changed channels, and it takes a little time for them to set it up, so if your cable box takes its time switching to a channel, it's because no one is watching that on the network, and the datacenter has to set it up just for you. The majority of cable channels is on-demand.

I don't think there's much reason to be anonymous, and I'm not under NDA, but I don't want to find out a reason later, soooo...
It's an excellent summary, but public knowledge. I don't know about the SPECIFIC numbers being public, but I don't think you need to worry about NDA.

That being said, being paranoid can also mean safe, so I don't blame you.


For more information on top of this, I would mention the old thing about how bandwidth was often consumed in "bursts" dictated network design in the past. A lot of older ISP infrastructure was built around people surfing the web or downloading files (think late 90s). They were "burst" transfers, and thus you only needed a very low amount of bandwidth to serve "everybody" with "high speed" 10Mbps. Once a page loaded, you were using 0 bandwidth, so you didn't need as much to serve EVERYBODY. Think of a game on steam, after the initial download, you're not using the bandwidth much.

But then YouTube and Netflix happened. Video on the web before that was often that you downloaded the file then played it. Still on the old paradigm. Hence so much talk about DRM then was because the FILE was played locally, and thus people could play it as they wished, and thus DRM to control the file. But now with streaming, you're not "get on, use, and get off the pipe" but you're CONSTANTLY using 2-10Mbps (or more for HD, or 4K, or whatever). This "breaks" that old model, and so more ACTUAL bandwidth per customer is needed, especially during peak times.

This is why getting bandwidth between about 4-6pm is often horrific. They have residential coming online, but still a significant number of business customers still needing bandwidth. They have max demand, but can't "shift" their big pipes to where it's needed. They're saturated.

An interesting problem either way. I thought it was worth it to amend your post though.
 
Total women in US

17,443,112



% of population

5.398213309378559%
There are only 5 left in stock for me![DOUBLEPOST=1498495294,1498494921][/DOUBLEPOST]Okay, I actually got this down to 1 single woman.

A healthy young Asian doctor with a yearly income over 100K who doesn't have nor want children. (5' 6" with a B or C cup).
 
Last edited:
There are only 5 left in stock for me![DOUBLEPOST=1498495294,1498494921][/DOUBLEPOST]Okay, I actually got this down to 1 single woman.

A healthy young Asian doctor with a yearly income over 100K who doesn't have nor want children. (5' 6" with a B or C cup).
Don't fall for it. It's just Bhamv in a wig.
 
And if you're looking, I found you an uneducated fat old white guy with a high income and no children. Erect penis of indeterminate length.
Hmmm, he must be one if the 22K men it said would meet my criteria. ;)

I didn't think I was being picky, either. I left a lot of options checked. I guess it's a good thing I got married after all.
 
TIL-the current Kid Flash's father is...a Reverse-Flash...and unlike Professor Zoom who legit STOLE the name had NO connection to Eobard Thawne...Rebirth exists for a reason.
 
Ahahahahaha! According to that site, I don't exist! I'm an f-ing unicorn baby!

Also...
Total Men in US: 29,209
% of population 0.009039471522870325%
 
Last edited:

GasBandit

Staff member
Damn, and I thought *I* was picky.

Also, this test doesn't ask any questions about hair color and I feel like it should.


Total women in US
1,150,573
% of population
0.3560740475479963%
 
It feels gratifying to me that I can truthfully tell any woman I fancy that she's 1 in a million. (Well, 1.15 million, but we'll round that off)
Really? I've discovered that telling my wife that there are probably only a dozen or so people out there in the world that I might desire more than her has rather the opposite effect.

--Patrick
 
Top