0

New NVIDIA paper says AI models do not always need long context

preview

New NVIDIA paper says AI models do not always need huge amounts of past conversation history (context) to work well. They show that regularly shortening the history (compaction) keeps accuracy almost the same on hard, long tasks.

New NVIDIA paper says AI models do not always need huge amounts of past conversation history (context) to work well. They show that regularly shortening the history (compaction) keeps accuracy almost the same on hard, long tasks.

0

The Montana Ranch Where Autistic People Are Wranglers and Guides

Reasons to be Cheerful The Montana Ranch Where Autistic People Are Wranglers and Guides reasonstobecheerful.world
preview

Erik’s Ranch is among a growing number of businesses striving to meet neurodivergent employees — and guests — where they are.

Erik’s Ranch is among a growing number of businesses striving to meet neurodivergent employees — and guests — where they are.

0

Junge Generation wendet sich von politischer Mitte ab

BR24 Junge Generation wendet sich von politischer Mitte ab www.br.de
preview

Die 18- bis 34-Jährigen trauen der Linken (29 Prozent) und der AfD (21) am ehesten zu, sich für ihre Belange einzusetzen. Weitgehend abgemeldet sind in dem Zusammenhang Union (5 Prozent), SPD (3) und Grüne (12). Im Durchschnitt aller Altersstufen halten die Deutschen AfD, Grüne (jeweils 18 Prozent) und Linke (15) für kompetent, sich für die Anliegen junger Menschen einzusetzen - auch hier mit großem Abstand zu Union (9) und SPD (8).

Die 18- bis 34-Jährigen trauen der Linken (29 Prozent) und der AfD (21) am ehesten zu, sich für ihre Belange einzusetzen. Weitgehend abgemeldet sind in dem Zusammenhang Union (5 Prozent), SPD (3) und Grüne (12). Im Durchschnitt aller Altersstufen halten die Deutschen AfD, Grüne (jeweils 18 Prozent) und Linke (15) für kompetent, sich für die Anliegen junger Menschen einzusetzen - auch hier mit großem Abstand zu Union (9) und SPD (8).

0

OccupySaaS — They laid us off. We're rebuilding their SaaS.

OccupySaaS — They laid us off. We're rebuilding their SaaS. occupysaas.com
preview

OccupySaaS is protest movement written in code.

Story is simple. Many engineers got laid off in tech. But the SaaS platforms they built keep growing and keep charging per seat. So the same industry rent our own work back to us. This project is answer to that. Laid off engineers use AI and public documentation to rebuild those platforms as free open source drop-in replacements. One seat at a time.

Method is clean room. Public docs only, open standards, public SDK behaviour. No leaked code, no decompile, no confidential info from former employer. AI help with build, humans write spec and review every change. Compatible is proven by running vendors own public SDKs against the server. If your existing code dont work unchanged, it is not done.

Two early releases: ZeroAuthy ( Auth0 compatible identity server) and SalesCompel( Salesforce compatible CRM)

They are early, not production ready yet. But the goal is the point.

OccupySaaS is protest movement written in code.

Story is simple. Many engineers got laid off in tech. But the SaaS platforms they built keep growing and keep charging per seat. So the same industry rent our own work back to us. This project is answer to that. Laid off engineers use AI and public documentation to rebuild those platforms as free open source drop-in replacements. One seat at a time.

Method is clean room. Public docs only, open standards, public SDK behaviour. No leaked code, no decompile, no confidential info from former employer. AI help with build, humans write spec and review every change. Compatible is proven by running vendors own public SDKs against the server. If your existing code dont work unchanged, it is not done.

Two early releases: ZeroAuthy ( Auth0 compatible identity server) and SalesCompel( Salesforce compatible CRM)

They are early, not production ready yet. But the goal is the point.

0

Uefa faces political flashpoint over Sevastopol’s ascent in Russian football

the Guardian Uefa faces political flashpoint over Sevastopol’s ascent in Russian football www.theguardian.com
preview

Promotion to third tier for Crimean club would be clear violation of governing body’s rules amid conflict in Ukraine

Promotion to third tier for Crimean club would be clear violation of governing body’s rules amid conflict in Ukraine

0

Fifa wants to restrict women’s Olympic football to under-23s in controversial move

the Guardian Fifa wants talks over restricting women’s Olympic football to under-23s www.theguardian.com
preview

Fifa wants the International Olympic Committee to restrict women’s Olympic football to under 23s owing to concerns over fixture congestion and workload

Fifa wants the International Olympic Committee to restrict women’s Olympic football to under 23s owing to concerns over fixture congestion and workload

0

Wooden Spoon, Golden Boot and playoff spots: what to watch for on MLS Decision Day

the Guardian Wooden Spoon, Golden Boot and playoff spots: what to watch for on MLS Decision Day www.theguardian.com
preview

The postseason is nearly upon us but there’s plenty to still be decided as the end of the regular season approaches

The postseason is nearly upon us but there’s plenty to still be decided as the end of the regular season approaches

40

Ablahshikar is a maximalist, sickos only salvage-survival CRPG about ripping organs out of warships and pawning off strange goods that's now out in early access

Rock Paper Shotgun Ablahshikar is a maximalist, sickos only salvage-survival CRPG about ripping organs out of warships and pawning off strange goods that's now out in early access www.rockpapershotgun.com
preview

I played the demo during NextFest and I will most certainly be buying the Early Access, if nothing else because it is much cheaper than the full release will be. Whether I actually play the EA or wait until 1.0 remains to be seen. And whether I am capable to decipher enough of the game to actually be able to beat said 1.0 release? Who knows.

What I do know is that the art is striking, the music banging and the writing is dense, distinctive and either poetic or lunatically gibberish (jury still out). The arcadey real-time bullet hell type ship combat is also quite fun.

I played the demo during NextFest and I will most certainly be buying the Early Access, if nothing else because it is much cheaper than the full release will be. Whether I actually play the EA or wait until 1.0 remains to be seen. And whether I am capable to decipher enough of the game to actually be able to beat said 1.0 release? Who knows.

What I do know is that the art is striking, the music banging and the writing is dense, distinctive and either poetic or lunatically gibberish (jury still out). The arcadey real-time bullet hell type ship combat is also quite fun.

0

PWHL announces rule changes, officiating team

The IX Hockey | Powered by The Ice Garden PWHL announces rule changes, officiating team | The IX Hockey | Powered by The Ice Garden www.theixsports.com
preview

With one month to go before the season begins, the PWHL has announced several rulebook updates, including mandatory neck protection.

With one month to go before the season begins, the PWHL has announced several rulebook updates, including mandatory neck protection.

0

Anti-Trans Ballot Measures Are Barely Passing, And They May Not Survive What Comes Next

Anti-Trans Ballot Measures Are Barely Passing, And They May Not Survive What Comes Next www.erininthemorning.com
preview

There are already plans to defang or defeat anti-trans ballot measures in blue states.

There are already plans to defang or defeat anti-trans ballot measures in blue states.

0

Could we poison AI if we all disagreed with it?

preview

I had a thought the other day about AI. What if everyone (or enough people) everywhere just started disagreeing with every and all AI responses and answers. What would happen if millions of people just started disagreeing with everything AI responded with. Even if it gave out right answers, or answers we believed were correct and we all just outright disagreed with everything all the time.

Would it eventually confuse LLMs? or alter their reality into something else and render them completely useless or corrupted?

Would it take hundreds of actual humans responding negatively all the time or would it take thousands or millions or billions of negative responses before it could (if it ever) had any effect.

I had a thought the other day about AI. What if everyone (or enough people) everywhere just started disagreeing with every and all AI responses and answers. What would happen if millions of people just started disagreeing with everything AI responded with. Even if it gave out right answers, or answers we believed were correct and we all just outright disagreed with everything all the time.

Would it eventually confuse LLMs? or alter their reality into something else and render them completely useless or corrupted?

Would it take hundreds of actual humans responding negatively all the time or would it take thousands or millions or billions of negative responses before it could (if it ever) had any effect.

0

Wrathful Protector

A close-up shot of a winged angelic figure facing forward with head tilted slightly downward while its glowing red eyes gaze toward the viewer. The figure features pale, marble-textured skin covered in fine craquelure cracks, symmetrical curlicue waves of hair on either side of a center part, visible winged shoulders behind the neck, and a thin, glowing gold ring halo hovering above their head. The setting is a solid, featureless black background. Dramatic top-down light illuminates the halo, the crown of the head, and the collarbones, casting deep, high-contrast shadows over the eyes, nose, and lower face.
preview

(dkpc69) (2026)

Image description: A close-up shot of a winged angelic figure facing forward with head tilted slightly downward while its glowing red eyes gaze toward the viewer. The figure features pale, marble-textured skin covered in fine craquelure cracks, symmetrical curlicue waves of hair on either side of a center part, visible winged shoulders behind the neck, and a thin, glowing gold ring halo hovering above their head. The setting is a solid, featureless black background. Dramatic top-down light illuminates the halo, the crown of the head, and the collarbones, casting deep, high-contrast shadows over the eyes, nose, and lower face.

Full Generation Parameters:

DCCULTF@NTSY. Framed in a close-up, slightly angled from above, the composition centers on a solitary figure with pale, textured skin and delicate, softly curled hair blending into deep shadow. The face is partially obscured by darkness, with only the lower half illuminated by faint, muted light. Two glowing red eyes, set deep within the shadowed face, draw immediate attention and contrast sharply against the otherwise subdued palette. Above the head, a thin, golden circular halo with a weathered, metallic surface floats against the black background, its reflective rim subtly catching the dim light. The subject’s expression is somber, with lowered gaze and downturned lips, while the bare shoulders and upper chest reveal finely cracked paint and gentle tonal transitions. The surrounding environment dissolves into darkness, offering no discernible details beyond the illuminated form and the faint golden ring. Steps: 8, CFG scale: 1, Sampler: er_sde, Seed: 4290676277370904, engine: ComfyUI, Model: Krea2\krea2_turbo_fp8_scaled

(dkpc69) (2026)

Image description: A close-up shot of a winged angelic figure facing forward with head tilted slightly downward while its glowing red eyes gaze toward the viewer. The figure features pale, marble-textured skin covered in fine craquelure cracks, symmetrical curlicue waves of hair on either side of a center part, visible winged shoulders behind the neck, and a thin, glowing gold ring halo hovering above their head. The setting is a solid, featureless black background. Dramatic top-down light illuminates the halo, the crown of the head, and the collarbones, casting deep, high-contrast shadows over the eyes, nose, and lower face.

Full Generation Parameters:

DCCULTF@NTSY. Framed in a close-up, slightly angled from above, the composition centers on a solitary figure with pale, textured skin and delicate, softly curled hair blending into deep shadow. The face is partially obscured by darkness, with only the lower half illuminated by faint, muted light. Two glowing red eyes, set deep within the shadowed face, draw immediate attention and contrast sharply against the otherwise subdued palette. Above the head, a thin, golden circular halo with a weathered, metallic surface floats against the black background, its reflective rim subtly catching the dim light. The subject’s expression is somber, with lowered gaze and downturned lips, while the bare shoulders and upper chest reveal finely cracked paint and gentle tonal transitions. The surrounding environment dissolves into darkness, offering no discernible details beyond the illuminated form and the faint golden ring. Steps: 8, CFG scale: 1, Sampler: er_sde, Seed: 4290676277370904, engine: ComfyUI, Model: Krea2\krea2_turbo_fp8_scaled

0

Zitsupensya Nashiko - Pon No Michi

A young woman running while extending her right arm outward with an open hand, and holding a yellow-cased smartphone in her left hand against a plain white background. She has black bob hair with a curved ahoge on top, blue eyes, an open mouth, and a dark blue school bag slung over her right shoulder. Her outfit consists of a tan sweater vest over a collared white dress shirt with rolled-up sleeves, a dark blue pleated skirt, white socks, and dark brown loafers.
preview

(tsutomu_s) (2026)

Image description: A young woman running while extending her right arm outward with an open hand, and holding a yellow-cased smartphone in her left hand against a plain white background. She has black bob hair with a curved ahoge on top, blue eyes, an open mouth, and a dark blue school bag slung over her right shoulder. Her outfit consists of a tan sweater vest over a collared white dress shirt with rolled-up sleeves, a dark blue pleated skirt, white socks, and dark brown loafers.

Full Generation Parameters:

anime screencap, official art, studio animation style, score 7, masterpiece, best quality, very aesthetic, absurdres, highres, High Definition, extremely detailed eyes and face, intricate details, soft skin texture, anatomically correct fingers, sharp-focus, crisp edges, detailed lines, clean lineart, A dynamic full-body capture of the character mid-hop or lightly floating with one leg raised and arms spread wide, smiling brightly at the viewer against a pure white background., <lora:Pon-No-Michi-Zitsupensya-Nashiko-Anima:1>, 1girl, solo, Pon-No-Michi-Zitsupensya-Nashiko, black hair, blunt bangs, single ahoge, low ponytail at nape tied with large yellow hair bow, large expressive aqua-blue eyes, large breasts, prominent bust, bare legs, wearing white long-sleeved pointed-collar shirt with open neck, fitted light beige V-neck knit sweater vest tight across chest, dark navy pleated miniskirt, white crew socks, dark loafers, carrying dark navy rectangular school shoulder bag with strap, holding black smartphone with yellow bumper case in hand Negative prompt: score 1, score 2, score3, 6 fingers, 6 toes, ai-generated, bad eyes, bad pupils, bad iris, bad hands, bad fingers, watermark, patreon logo, blurry, jpeg artifacts, sepia, photorealistic, realistic, photograph, photography, film grain, 3d, 3d render, lowres, bokeh, depth of field, sharp focus, lens flare, film look, cinematic lighting, scanned, raw photo, physical, grainy, real life, realistic skin, realistic texture, mosaic censored, (mosaic:1.5), (censoring:1.5), (bar censor:1.5), (white censor:1.5), (blurred:1.3), (pixelated:1.4), (censor blur:1.4), (black bar:1.4), (shredded:1.2), (noise:1.2), (grid:1.2), (soft focus:1.3), (blurry:1.3), lowres, blurry edges, smudged, haze, washed out colors, low contrast, depth of field, fairy, pokemon (creature), (lipstick, lip:1.5) Steps: 21, CFG scale: 4, Sampler: ER SDE, Seed: 3617153882, RNG: NV, Model: oneObsessionBranch_matureAnimaV1, Shift: 3, width: 768, height: 1024, Version: neo-2.29, Emphasis: Original, Module 1: qwen_image_vae, Module 2: miaomiaoHarem_aniAnimeColoring10_txt, Module 3: qwen_3_06b_base, Model hash: 1c91955cea, VAE Decoder: TAESD, Schedule type: Simple

(tsutomu_s) (2026)

Image description: A young woman running while extending her right arm outward with an open hand, and holding a yellow-cased smartphone in her left hand against a plain white background. She has black bob hair with a curved ahoge on top, blue eyes, an open mouth, and a dark blue school bag slung over her right shoulder. Her outfit consists of a tan sweater vest over a collared white dress shirt with rolled-up sleeves, a dark blue pleated skirt, white socks, and dark brown loafers.

Full Generation Parameters:

anime screencap, official art, studio animation style, score 7, masterpiece, best quality, very aesthetic, absurdres, highres, High Definition, extremely detailed eyes and face, intricate details, soft skin texture, anatomically correct fingers, sharp-focus, crisp edges, detailed lines, clean lineart, A dynamic full-body capture of the character mid-hop or lightly floating with one leg raised and arms spread wide, smiling brightly at the viewer against a pure white background., <lora:Pon-No-Michi-Zitsupensya-Nashiko-Anima:1>, 1girl, solo, Pon-No-Michi-Zitsupensya-Nashiko, black hair, blunt bangs, single ahoge, low ponytail at nape tied with large yellow hair bow, large expressive aqua-blue eyes, large breasts, prominent bust, bare legs, wearing white long-sleeved pointed-collar shirt with open neck, fitted light beige V-neck knit sweater vest tight across chest, dark navy pleated miniskirt, white crew socks, dark loafers, carrying dark navy rectangular school shoulder bag with strap, holding black smartphone with yellow bumper case in hand Negative prompt: score 1, score 2, score3, 6 fingers, 6 toes, ai-generated, bad eyes, bad pupils, bad iris, bad hands, bad fingers, watermark, patreon logo, blurry, jpeg artifacts, sepia, photorealistic, realistic, photograph, photography, film grain, 3d, 3d render, lowres, bokeh, depth of field, sharp focus, lens flare, film look, cinematic lighting, scanned, raw photo, physical, grainy, real life, realistic skin, realistic texture, mosaic censored, (mosaic:1.5), (censoring:1.5), (bar censor:1.5), (white censor:1.5), (blurred:1.3), (pixelated:1.4), (censor blur:1.4), (black bar:1.4), (shredded:1.2), (noise:1.2), (grid:1.2), (soft focus:1.3), (blurry:1.3), lowres, blurry edges, smudged, haze, washed out colors, low contrast, depth of field, fairy, pokemon (creature), (lipstick, lip:1.5) Steps: 21, CFG scale: 4, Sampler: ER SDE, Seed: 3617153882, RNG: NV, Model: oneObsessionBranch_matureAnimaV1, Shift: 3, width: 768, height: 1024, Version: neo-2.29, Emphasis: Original, Module 1: qwen_image_vae, Module 2: miaomiaoHarem_aniAnimeColoring10_txt, Module 3: qwen_3_06b_base, Model hash: 1c91955cea, VAE Decoder: TAESD, Schedule type: Simple

0

BluePixel: Toward Native 4K Agentic Image Editing

BluePixel: Toward Native 4K Agentic Image Editing tech.vivo.com
preview

Abstract

Instruction-driven image editing has reached native 4K, yet how such editors are trained remains undisclosed, and pretraining one from scratch is beyond most teams. The task itself compounds the difficulty: a single end-to-end editor cannot make precise parametric adjustments, match specialized experts, or interpret the vague instructions users actually write. We argue that 4K editing is a system problem rather than a single-model problem, and present BluePixel, which couples a native-4K editing model with an agent harness; the complete training recipe is disclosed in this paper. The model makes 4K training practical: it turns a 1K-pretrained editor into a native-4K one through progressive VAE adaptation that keeps 4K training as affordable as 1K, absorbs a long tail of vertical editing skills through continual learning, and refines the aesthetic-fidelity trade-off with reinforcement learning. Distillation via source-anchored target projection (SATP) then compresses 50 denoising steps to 4, completing a 4K edit in 8.7 s on a single Kunlunxin P800 XPU while preserving task success rate and background consistency. The harness, BluePixel Agent, makes editing controllable: it decomposes user instructions into specialist and general operations, verifies every result with the same reward model that guided training, and retries within bounded attempts. On VINS-4KEval++ (3,305 native-4K cases), BluePixel ranks second among seven models, behind only GPT Image 2 in non-edited-region consistency.

Paper: arxiv.org/abs/2610.19882

Code: github.com/vivo/BluePixel

Model: huggingface.co/openvivo/BluePixel-Turbo

Project Page: tech.vivo.com/BluePixel/

Abstract

Instruction-driven image editing has reached native 4K, yet how such editors are trained remains undisclosed, and pretraining one from scratch is beyond most teams. The task itself compounds the difficulty: a single end-to-end editor cannot make precise parametric adjustments, match specialized experts, or interpret the vague instructions users actually write. We argue that 4K editing is a system problem rather than a single-model problem, and present BluePixel, which couples a native-4K editing model with an agent harness; the complete training recipe is disclosed in this paper. The model makes 4K training practical: it turns a 1K-pretrained editor into a native-4K one through progressive VAE adaptation that keeps 4K training as affordable as 1K, absorbs a long tail of vertical editing skills through continual learning, and refines the aesthetic-fidelity trade-off with reinforcement learning. Distillation via source-anchored target projection (SATP) then compresses 50 denoising steps to 4, completing a 4K edit in 8.7 s on a single Kunlunxin P800 XPU while preserving task success rate and background consistency. The harness, BluePixel Agent, makes editing controllable: it decomposes user instructions into specialist and general operations, verifies every result with the same reward model that guided training, and retries within bounded attempts. On VINS-4KEval++ (3,305 native-4K cases), BluePixel ranks second among seven models, behind only GPT Image 2 in non-edited-region consistency.

Paper: arxiv.org/abs/2610.19882

Code: github.com/vivo/BluePixel

Model: huggingface.co/openvivo/BluePixel-Turbo

Project Page: tech.vivo.com/BluePixel/

49

GTA VI is out on 19 November, but fake sites already offer "leaked" copies. One is flagged as a Trojan by 34 of 68 antivirus engines, Netcraft found.

Grand Theft Auto VI' Release Phishing Lures | Netcraft Research www.netcraft.com
preview

cross-posted from: https://piefed.world/c/tech/p/1514223/gta-vi-is-out-on-19-november-but-fake-sites-already-offer-leaked-copies-one-is-flagged-a

cross-posted from: https://piefed.world/c/tech/p/1514223/gta-vi-is-out-on-19-november-but-fake-sites-already-offer-leaked-copies-one-is-flagged-a

110

Pillars of Eternity and Pillars of Eternity II: Deadfire get major updates with controller support in Beta

GamingOnLinux Pillars of Eternity and Pillars of Eternity II: Deadfire get major updates with controller support in Beta www.gamingonlinux.com
preview

Pillars of Eternity II is one of the best RPGs ever made, and I would have loved controller support when I played it (twice in a row, because it was so good!) over a year ago.

Pillars of Eternity II is one of the best RPGs ever made, and I would have loved controller support when I played it (twice in a row, because it was so good!) over a year ago.

0

The Panoramic Survey Telescope and Rapid Response System, known as Pan-STARRS, provides an unprecedented high-resolution map of the celestial sphere. Utilizing a 1.8-meter telescope and the world's

preview

The Panoramic Survey Telescope and Rapid Response System, known as Pan-STARRS, provides an unprecedented high-resolution map of the celestial sphere. Utilizing a 1.8-meter telescope and the world’s largest digital camera, the project monitors the sky for transient objects, cataloging billions of individual stars and galaxies to track cosmic variability. This ongoing survey is fundamental for identifying near-Earth asteroids and mapping the large-scale structure of our universe.

#space #astronomy #astrophysics

@Space@lemmy.science.social @science@beehaw.org @science@lemmy.world @space@beehaw.org @space@lemmy.world @science@newsmast.community @space@newsmast.community @space@lemmy.ml #space #science #nasa #astronomy

The Panoramic Survey Telescope and Rapid Response System, known as Pan-STARRS, provides an unprecedented high-resolution map of the celestial sphere. Utilizing a 1.8-meter telescope and the world’s largest digital camera, the project monitors the sky for transient objects, cataloging billions of individual stars and galaxies to track cosmic variability. This ongoing survey is fundamental for identifying near-Earth asteroids and mapping the large-scale structure of our universe.

#space #astronomy #astrophysics

@Space@lemmy.science.social @science@beehaw.org @science@lemmy.world @space@beehaw.org @space@lemmy.world @science@newsmast.community @space@newsmast.community @space@lemmy.ml #space #science #nasa #astronomy

0

Il giovedì del libro!

preview

Il giovedì del libro!
@libri@feddit.it @libri @cultura #UnoLibri
.
Questa settimana ho un #SuggerimentoDiLettura che mi guarda dallo scaffale da molto tempo: I MORTI VIVENTI di George Romero.
Esatto, si tratta proprio di quel George Romero che alla fine degli anni '60 definì il concetto di zombie come lo conosciamo: un essere senza volontà e anima, che agisce in preda a un unico contagioso impulso, il cannibalismo, un tabù che determina il collasso apocalittico della società. (Per me) interessantissimo.

Il giovedì del libro!
@libri@feddit.it @libri @cultura #UnoLibri
.
Questa settimana ho un #SuggerimentoDiLettura che mi guarda dallo scaffale da molto tempo: I MORTI VIVENTI di George Romero.
Esatto, si tratta proprio di quel George Romero che alla fine degli anni '60 definì il concetto di zombie come lo conosciamo: un essere senza volontà e anima, che agisce in preda a un unico contagioso impulso, il cannibalismo, un tabù che determina il collasso apocalittico della società. (Per me) interessantissimo.

0

How Is Pokémon Go Still A Thing? [13:21]

play · youtube.com
YouTube How Is Pokémon Go Still A Thing? www.youtube.com
preview

Pokémon Go is still going, and it has been doing some very weird things while I wasn’t looking.

The game has spent years getting players to scan real-world locations, which later became useful for training robots, smart glasses and other AI systems. Very normal Pokémon stuff.

So I’m looking at what Niantic did with all that data, how Pokémon Go ended up connected to delivery robots and military technology, and why millions of people are still happily playing it.

Pokémon Go is still going, and it has been doing some very weird things while I wasn’t looking.

The game has spent years getting players to scan real-world locations, which later became useful for training robots, smart glasses and other AI systems. Very normal Pokémon stuff.

So I’m looking at what Niantic did with all that data, how Pokémon Go ended up connected to delivery robots and military technology, and why millions of people are still happily playing it.