Story: "Hairpins" Part 4
Feb. 26th, 2014 12:03 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
ysabetwordsmithThis story belongs to the series Love Is For Children which includes "Love Is for Children," "Eggshells," "Dolls and Guys,""Saudades," "Turnabout Is Fair Play," "Touching Moments," "Splash," "Coming Around," "Birthday Girl," "No Winter Lasts Forever," "Hide and Seek," "Kernel Error," "Happy Hour," and "Green Eggs and Hulk."
Fandom: The Avengers
Characters: Phil Coulson, JARVIS, Clint Barton, Tony Stark, Steve Rogers, Natasha Romanova, Bruce Banner.
Medium: Fiction
Warnings: This story is mostly fluff, but it has some intense scenes in the middle. Highlight for details. These include dubious consent as Phil and JARVIS discuss what really happened when Agent Coulson hacked his way into Stark Tower, over which Phil has something between a flashback and a panic attack. They also discuss some of the bad things that have happened to Avengers in the past, including various flavors of abuse. If these are sensitive topics for you, please think carefully before deciding whether to read onward.
Summary: Uncle Phil needs to pick out pajamas for game night. He gets help from an unexpected direction.
Notes: Service. Shopping. Gifts. Artificial intelligence. Computers. Teamwork. Team as family. Friendship. Communication. Hope. Apologies. Forgiveness. Nonsexual ageplay. Nonsexual intimacy. Love. Tony Stark needs a hug. Bruce Banner needs a hug. #coulsonlives.
Begin with Part 1, Part 2, Part 3. Skip to Part 6, Part 7, Part 8, Part 9.
"Hairpins" Part 4
Wasn't there something about dancing bears ...? he thought, the image plucking at his memory. He knew that Natasha loved ballet, although she rarely expressed it. Perhaps with the right coaxing, she would let it out more.
"Dancing bears," Phil decided, and ran a search for that. He loved StarkSearch. In addition to the usual safe mode, it also had an Intelligent Search option that would learn a user's preferences and sort out the most promising results. It had started off a little better than Phil's former favorite, but the more he used it, the more accurate it seemed to get. Plus it didn't leave any breadcrumbs in cyberspace at large, and it offered him the option of saving or deleting his search on the tower servers.
At first the search turned up irrelevant things like circus bears, traveling caravans, and street performers. The same images appeared in paintings and sketches, some of them quite old. Then Phil started finding teddy bears in tutus, or t-shirts with dancing bears printed on them. "Getting warmer," he said.
Next Phil tried searching for 'dancing bears fabric' on the premise that if he could find suitable cloth, then he could commission someone to make it into pajamas. That led to big colorful quilt blocks and a wide variety of Grateful Dead swag. A few images of teddy bear picnics and Winnie the Pooh also appeared. Most of them had backgrounds of pink or blue, or worse, distracting patterns.
Phil set about sorting the images into things close to what he wanted, with the best ones at the top of the page and others lower down. Then he filtered out the irrelevant ones in the last row by dragging them into the trash. This was the true beauty of StarkSearch: it popped up new suggestions at the bottom and let Phil float them to wherever he felt they belonged. Occasionally a new one would appear higher up, usually with a question mark on the corner. He could tap that to confirm its position, or touch and pull to move it elsewhere.
Soon the first images of large or small bears on a white background began to appear. They were boring, but the concept was moving in the right direction. Phil's interest gave a hopeful little quiver upon seeing them. He bumped those toward the top of the screen. Then he saved the search.
Next, Phil opened some of the files he kept on Natasha. Clothing preferences, both professional and personal. Furniture. Things she had bought for the two-floor apartment that she shared with Clint. Favorite art. An assortment of books that she liked to read, mostly nonfiction.
It was a stretch, because Phil had very little information about her childhood, almost all of that ugly examples of systematic abuse. Still, Phil thought that he might manage to extrapolate from Natasha to Natka in terms of taste. He opened a comparison page and started dragging images onto that from Natasha's files. StarkSearch had a 'correlate' function that Phil also loved, which would seek for patterns across two or more sets of information. Once he had a page full of samples, he activated that and then went back to the search page.
* * *
Notes:
Tame dancing bears appear in circus performances and other venues. They are associated with Russia, although not restricted to that culture.
Machine learning may use any of several problem-solving methods. A good program can even learn language just by reading instructions and interacting with a game or other program. You can see JARVIS trying different things with the Avengers to see what works and what doesn't.
Breadcrumbs are bits of information left in cyberspace as people move through it. There are various ways to stay anonymous online, but they reduce the efficiency and effectiveness of many experiences. Restricting the information to a secure server is one way to get around that.
To see the progression of the search, view the quilt squares, Grateful Dead bears, teddy bear picnic, and Winnie the Pooh swatches. A search often starts with widely scattered results and then spirals closer to the goal.
Visual and interactive search engines can improve usability. JARVIS uses additional information to filter the raw results more finely as a search progresses. He also uses the placement of images on the page to indicate his level of confidence in the findings.
These are the large and small plain bear fabrics.
The reward system in the human brain is what delivers, among other things, that little thrill upon getting a right answer or reaching a goal. Given Phil's work in espionage and handling, it's reasonable to expect that his brain delivers a very satisfying jolt of pleasure upon finding information or helping one of his people. This deals with motivation and decision-making, in that people are more inclined to invest energy in activities that result in pleasure. Popular video games are designed to trigger the reward system. JARVIS combines his knowledge of human neurobiology and cyberspace theory to evolve programs that are not only effective but also enjoyable -- which of course activates his pleasure circuits in return.
A correlation function is a way of comparing different batches of information to see how they overlap and interact. This can be useful in such tasks as predicting which future data points will match a certain subset of criteria. There are many different heuristics, or ways of solving a problem based on personal experience. JARVIS is better at crunching numbers and handling raw data; Phil is better at making subtle emotional observations and estimates. Put them together and they move in the right direction quite briskly.
[To be continued in Part 5 ...]
Fandom: The Avengers
Characters: Phil Coulson, JARVIS, Clint Barton, Tony Stark, Steve Rogers, Natasha Romanova, Bruce Banner.
Medium: Fiction
Warnings: This story is mostly fluff, but it has some intense scenes in the middle. Highlight for details. These include dubious consent as Phil and JARVIS discuss what really happened when Agent Coulson hacked his way into Stark Tower, over which Phil has something between a flashback and a panic attack. They also discuss some of the bad things that have happened to Avengers in the past, including various flavors of abuse. If these are sensitive topics for you, please think carefully before deciding whether to read onward.
Summary: Uncle Phil needs to pick out pajamas for game night. He gets help from an unexpected direction.
Notes: Service. Shopping. Gifts. Artificial intelligence. Computers. Teamwork. Team as family. Friendship. Communication. Hope. Apologies. Forgiveness. Nonsexual ageplay. Nonsexual intimacy. Love. Tony Stark needs a hug. Bruce Banner needs a hug. #coulsonlives.
Begin with Part 1, Part 2, Part 3. Skip to Part 6, Part 7, Part 8, Part 9.
"Hairpins" Part 4
Wasn't there something about dancing bears ...? he thought, the image plucking at his memory. He knew that Natasha loved ballet, although she rarely expressed it. Perhaps with the right coaxing, she would let it out more.
"Dancing bears," Phil decided, and ran a search for that. He loved StarkSearch. In addition to the usual safe mode, it also had an Intelligent Search option that would learn a user's preferences and sort out the most promising results. It had started off a little better than Phil's former favorite, but the more he used it, the more accurate it seemed to get. Plus it didn't leave any breadcrumbs in cyberspace at large, and it offered him the option of saving or deleting his search on the tower servers.
At first the search turned up irrelevant things like circus bears, traveling caravans, and street performers. The same images appeared in paintings and sketches, some of them quite old. Then Phil started finding teddy bears in tutus, or t-shirts with dancing bears printed on them. "Getting warmer," he said.
Next Phil tried searching for 'dancing bears fabric' on the premise that if he could find suitable cloth, then he could commission someone to make it into pajamas. That led to big colorful quilt blocks and a wide variety of Grateful Dead swag. A few images of teddy bear picnics and Winnie the Pooh also appeared. Most of them had backgrounds of pink or blue, or worse, distracting patterns.
Phil set about sorting the images into things close to what he wanted, with the best ones at the top of the page and others lower down. Then he filtered out the irrelevant ones in the last row by dragging them into the trash. This was the true beauty of StarkSearch: it popped up new suggestions at the bottom and let Phil float them to wherever he felt they belonged. Occasionally a new one would appear higher up, usually with a question mark on the corner. He could tap that to confirm its position, or touch and pull to move it elsewhere.
Soon the first images of large or small bears on a white background began to appear. They were boring, but the concept was moving in the right direction. Phil's interest gave a hopeful little quiver upon seeing them. He bumped those toward the top of the screen. Then he saved the search.
Next, Phil opened some of the files he kept on Natasha. Clothing preferences, both professional and personal. Furniture. Things she had bought for the two-floor apartment that she shared with Clint. Favorite art. An assortment of books that she liked to read, mostly nonfiction.
It was a stretch, because Phil had very little information about her childhood, almost all of that ugly examples of systematic abuse. Still, Phil thought that he might manage to extrapolate from Natasha to Natka in terms of taste. He opened a comparison page and started dragging images onto that from Natasha's files. StarkSearch had a 'correlate' function that Phil also loved, which would seek for patterns across two or more sets of information. Once he had a page full of samples, he activated that and then went back to the search page.
* * *
Notes:
Tame dancing bears appear in circus performances and other venues. They are associated with Russia, although not restricted to that culture.
Machine learning may use any of several problem-solving methods. A good program can even learn language just by reading instructions and interacting with a game or other program. You can see JARVIS trying different things with the Avengers to see what works and what doesn't.
Breadcrumbs are bits of information left in cyberspace as people move through it. There are various ways to stay anonymous online, but they reduce the efficiency and effectiveness of many experiences. Restricting the information to a secure server is one way to get around that.
To see the progression of the search, view the quilt squares, Grateful Dead bears, teddy bear picnic, and Winnie the Pooh swatches. A search often starts with widely scattered results and then spirals closer to the goal.
Visual and interactive search engines can improve usability. JARVIS uses additional information to filter the raw results more finely as a search progresses. He also uses the placement of images on the page to indicate his level of confidence in the findings.
These are the large and small plain bear fabrics.
The reward system in the human brain is what delivers, among other things, that little thrill upon getting a right answer or reaching a goal. Given Phil's work in espionage and handling, it's reasonable to expect that his brain delivers a very satisfying jolt of pleasure upon finding information or helping one of his people. This deals with motivation and decision-making, in that people are more inclined to invest energy in activities that result in pleasure. Popular video games are designed to trigger the reward system. JARVIS combines his knowledge of human neurobiology and cyberspace theory to evolve programs that are not only effective but also enjoyable -- which of course activates his pleasure circuits in return.
A correlation function is a way of comparing different batches of information to see how they overlap and interact. This can be useful in such tasks as predicting which future data points will match a certain subset of criteria. There are many different heuristics, or ways of solving a problem based on personal experience. JARVIS is better at crunching numbers and handling raw data; Phil is better at making subtle emotional observations and estimates. Put them together and they move in the right direction quite briskly.
[To be continued in Part 5 ...]
(no subject)
Date: 2014-02-26 10:16 pm (UTC)I'm a little surprised at Phil's lack of google-fu, though. I can't figure out if he's really that bad at crafting a search string, or if it's just that it's always feels agonizingly slow to watch somebody else google.
Thoughts
Date: 2014-02-28 06:54 am (UTC)Yeah, it would be awesome.
>> I'm a grad student in statistics, and it makes me twitchy thinking about how much processing power that would take (never mind writing the program in the first place). <<
Things to consider:
* Tony started coding artificial intelligence at 17.
* JARVIS can render fully detailed 3D holograms in realtime with no apparent bandwidth drain.
* In this case, one of the sets was probably 2-3 dozen sample items from Natasha's stuff. If bandwidth was a concern, crossing that with a thousand or so search results would probably suffice, rather than recrawling the whole of cyberspace.
>> I'm a little surprised at Phil's lack of google-fu, though. I can't figure out if he's really that bad at crafting a search string, or if it's just that it's always feels agonizingly slow to watch somebody else google. <<
Phil is not bad at google-fu. He's searching for something outside of his experience range, which means he doesn't know the vocabulary to bring up the right things quickly -- plus he's searching for a general product type, rather than knowing exactly what he wants already. That slows down the rate.
If he'd gone looking for, say, the President's datebook or a particular model of gun, it wouldn't have taken him more than a minute or few. That's the kind of stuff that Phil handles regularly.