hautepop
hautepop:

Six Degrees Of Health Goth
So I took my hand-sketched semantic network map from yesterday and did it again, properly.
Method:
1. Start at health goth, and look at the 3 related topics that Tumblr gives you. Write these down as sets of pairs:
health goth - street goth health goth - seapunk health goth - vaporwave
3. Ladder up from the 3 related topics, looking at each of their related topics. Do this process 6 times, building a list of pairs.
4. Import the dataset as a .csv file into Gephi as a list of directed edges.
5. Force Atlas 2 layout, weight node size by in-degree, and node colour by modularity.
6. Tweak node colour a bit manually so the different communities show up better.
Network properties:
Working assumption: Tumblr’s related tags are based on tags co-occuring in the same post. Consequently it’s based on user behaviour (not say hand of mod), and we can look at how tags are connected as a proxy for how Tumblr users think these ideas & concepts are related.
The relationships between tags loop round and keep interconnecting:health goth —> seapunk —> grunge —> vintage —> retro —> hipster —> girl —> women’s fashion —> chic —> korea —> women’s fashion
Where a group of tags keep linking into each other - e.g. one around women’s fashion, chic, street style, korea, china and girl - that forms a community (in network terms). The modularity algorithm can detect which groups of tags are most closely linked like this.
Interesting findings:
1. 79 tags are within 6 degrees of health goth. Given that each tag has 3 related tags, that shows there’s a lot of interconnection between them (we could potentially be dealing with I think 364 tags: 3^0+3^1+3^2+3^3+3^4+3^5).
2. Health goth itself has in-degree of zero, aka none of the tags it links out to ever link back to it. This suggests it’s a small subculture
3. Vaporwave has the highest in-degree, of 6. Now, this is partly because it’s close to our starting point so it has more opportunity to gain “ins” - but moreso because it’s the hub for a number of different design/art aesthetics: net art, web art, webpunk, seapunk, the word ‘aesthetic’ and of course health goth itself.
4. HBA links round to itself - ok strictly speaking hba links to hood by air, but I aggregated the two to get a nice loop :0
5. Korea and China are huge tags around women’s fashion and chic. This is massively interesting. I’m seeing it around streetwear and street goth too, though it doesn’t show up in this map. As a teenager I grew up on the FRUiTS magazine books of streetstyle from Tokyo… But it’s now looking like Seoul’s stolen Harajuku’s thunder.
6. Women’s fashion is correctly punctuated, but mens fashion, mens clothing and mens style are not.

hautepop:

Six Degrees Of Health Goth

So I took my hand-sketched semantic network map from yesterday and did it again, properly.

Method:

1. Start at health goth, and look at the 3 related topics that Tumblr gives you. Write these down as sets of pairs:

health goth - street goth
health goth - seapunk
health goth - vaporwave

3. Ladder up from the 3 related topics, looking at each of their related topics. Do this process 6 times, building a list of pairs.

4. Import the dataset as a .csv file into Gephi as a list of directed edges.

5. Force Atlas 2 layout, weight node size by in-degree, and node colour by modularity.

6. Tweak node colour a bit manually so the different communities show up better.


Network properties:

Working assumption: Tumblr’s related tags are based on tags co-occuring in the same post. Consequently it’s based on user behaviour (not say hand of mod), and we can look at how tags are connected as a proxy for how Tumblr users think these ideas & concepts are related.

The relationships between tags loop round and keep interconnecting:
health goth —> seapunk —> grunge —> vintage —> retro —> hipster —> girl —> women’s fashion —> chic —> korea —> women’s fashion

Where a group of tags keep linking into each other - e.g. one around women’s fashion, chic, street style, korea, china and girl - that forms a community (in network terms). The modularity algorithm can detect which groups of tags are most closely linked like this.


Interesting findings:

1. 79 tags are within 6 degrees of health goth. Given that each tag has 3 related tags, that shows there’s a lot of interconnection between them (we could potentially be dealing with I think 364 tags: 3^0+3^1+3^2+3^3+3^4+3^5).

2. Health goth itself has in-degree of zero, aka none of the tags it links out to ever link back to it. This suggests it’s a small subculture

3. Vaporwave has the highest in-degree, of 6. Now, this is partly because it’s close to our starting point so it has more opportunity to gain “ins” - but moreso because it’s the hub for a number of different design/art aesthetics: net art, web art, webpunk, seapunk, the word ‘aesthetic’ and of course health goth itself.

4. HBA links round to itself - ok strictly speaking hba links to hood by air, but I aggregated the two to get a nice loop :0

5. Korea and China are huge tags around women’s fashion and chic. This is massively interesting. I’m seeing it around streetwear and street goth too, though it doesn’t show up in this map. As a teenager I grew up on the FRUiTS magazine books of streetstyle from Tokyo… But it’s now looking like Seoul’s stolen Harajuku’s thunder.

6. Women’s fashion is correctly punctuated, but mens fashion, mens clothing and mens style are not.

hautepop

hautepop:

Anatomy of A Tumblr Trend part 2: the semantic network map

So Tumblr’s got this cool feature where, if you search for a tag, it gives you 3 related tags that are (presumably) most commonly used in combination with your search term.

So I searched for the 2 Tumblr trends I’m currently interested in - street goth and health goth - and followed these links to see what they connected to. This creates a kind of semantic network, a way to diagram how people on Tumblr link different style subcultures, based on the top three most-associated terms neighbouring each term.

The result: my crude map above, scribbled on a piece of paper.

The interesting thing: street goth and health goth aren’t connected.

They link only at the second degree, through “fashion” - which is supergeneric. People aren’t connecting the two terms any more closely.

Instead they split off with their own sets of associations. Health goth gets associated with normcore (K-HOLE’s term that’s been bizarrely adopted by mainstream fashion) and seapunk, the Tumblr trend that broke out into brief celebrity with Rihanna et al last year. This then links up to a much more visually-oriented (i.e. non-fashion) set of concepts - vaporwave, pale - ending up at net art and web art.

Meanwhile street goth is all about Hood By Air and Pyrex (labels), then “blvck”, streetwear, then trill, dope & other street slang.

So it looks like we’ve got here is two separate communities.

What separates them? Funnily enough for two monochrome aesthetics, it’s kind of black and white.

While both tags are a long way from monoethnic, street goth connects into hiphop streetwear culture which draws most directly from black culture. Whereas terms around health goth - normcore, hipster, grunge - are much whiter.

That’s really interesting.

*

(What the hell is streetgoth? Read my first Anatomy of a Tumblr Trend post here. For health goth, the source, a somewhat pretentious essay.)

thisistheverge
thisistheverge:

Valve’s VR headset revealed with Oculus-like features Valve’s prototype virtual reality headset was originally revealed to select developers earlier this year behind closed doors, but the company is now starting to demonstrate an updated version in public. At a VR meetup in Boston yesterday, attendees were given an even closer look at Valve’s new headset. It appears that the latest prototype has changed significantly, with a separate camera used to track spots on the head-mounted display. It’s a similar system to Oculus’ latest Crystal Cove concept, and marks a change from Valve’s previous approach of using QR codes and a camera attached to the head-mounted display to improve tracking.

thisistheverge:

Valve’s VR headset revealed with Oculus-like features
Valve’s prototype virtual reality headset was originally revealed to select developers earlier this year behind closed doors, but the company is now starting to demonstrate an updated version in public. At a VR meetup in Boston yesterday, attendees were given an even closer look at Valve’s new headset. It appears that the latest prototype has changed significantly, with a separate camera used to track spots on the head-mounted display. It’s a similar system to Oculus’ latest Crystal Cove concept, and marks a change from Valve’s previous approach of using QR codes and a camera attached to the head-mounted display to improve tracking.