Taylor “Tigt” Hunttigt@mastodon.social
10/20/2026oof
Taylor “Tigt” Hunt
57 followers24 following190 posts since Nov 1, 2016
- Taylor “Tigt” Hunttigt@mastodon.social9/14/2026
Advertise as “fastest[1][2]”? WELL. NOW WE'RE TALKING, AREN'T WE
- Taylor “Tigt” Hunttigt@mastodon.social9/14/2026
“Fast by default” is less than nothing. If they just said “fast”, that’s more of a promise. Every room is minimalist by default. Every car has better-than-average MPG before carrying anything. Every program is zero-config until you dare have a thought the author didn—wait am I going too far
- Taylor “Tigt” Hunttigt@mastodon.social8/18/2026
*John Gruber voice*: everything the Trump admin likes is fully terrible… except wow it’s pretty nice that Markdown is the lingua franca of his biggest supporters, huh? Sou desu ne
- Taylor “Tigt” Hunttigt@mastodon.social8/7/2026
If State of Devs hasn't jumped the shark already, what is this?
- Taylor “Tigt” Hunttigt@mastodon.social6/2/2026
It must sting bad that old #Mozilla leadership was so right-but-wrong-time about B2G/FireFoxOS what with KaiOS and JioPhone, and now new Mozilla leadership is presumably informing their “AI” strategy on that bet
- Taylor “Tigt” Hunttigt@mastodon.social5/20/2026
Google shows this Happy Birthday animation right over top their usual alerts, so:…
What’s the most tone-deaf Google alert that could go under this? This one seems fairly close but I’m sure there’s worse
- Taylor “Tigt” Hunttigt@mastodon.social4/10/2026
- Taylor “Tigt” Hunttigt@mastodon.social10/16/2024
In retrospect, this comic needed to be 16 pages instead of 8 to fit all that text. Maybe I'll redo it someday.
- Taylor “Tigt” Hunttigt@mastodon.social10/16/2024
Last year, I participated in a #comic anthology with the theme of "invasive". I tried making a scary-but-plausible scifi story about about ads, tracking, and #privacy. Apologies to @pluralistic
- Taylor “Tigt” Hunttigt@mastodon.social9/1/2024
With popular websites like YouTube pulling stuff like THIS, I’m consistently surprised that screen readers (and other AT) don’t all suppress announcement of redundant `title` attributes that match link text
- Taylor “Tigt” Hunttigt@mastodon.social1/6/2024
so many design systems out there and apparently none have a “danger” button with spiky borders, which is a real shame
- Taylor “Tigt” Hunttigt@mastodon.social10/8/2023
As for the next entries… which do you want to see?
- Taylor “Tigt” Hunttigt@mastodon.social10/8/2023
The one y’all’ve probably seen already, but here it is once more for posterity: https://dev.to/tigt/why-not-react-2f8l
- Taylor “Tigt” Hunttigt@mastodon.social10/8/2023
I’m now in cartooning school and no longer care about webdev industry blowback, so I’ve started (slowly) publishing bits from the Streets series I had omitted for… reasons
- Taylor “Tigt” Hunttigt@mastodon.social7/29/2023
@jamey Do you have a link to the results from your “what do webcomic creators understand about RSS” survey?
- Taylor “Tigt” Hunttigt@mastodon.social5/10/2023
I wrote a dumb article and now you have to suffer me linking to it: https://dev.to/tigt/how-to-succeed-in-open-source-without-really-trying-really-55pj
- Taylor “Tigt” Hunttigt@mastodon.social5/9/2023
Beans, I forgot the link: https://css-irl.info/solving-a-tricky-layout-problem/
- Taylor “Tigt” Hunttigt@mastodon.social5/9/2023
I just found out what @michelle ’s Mastodon account is, so hat tip to her for the CSS Grid code that inspired me here
- Taylor “Tigt” Hunttigt@mastodon.social5/8/2023
An example of what I’m trying to accomplish: have this speech balloon grow into the left-hand gutter when translated into a longer language (like German)
![[back to Dory again]
Generating the UV patterns was only half of DoubleThink’s responsibilities. It was also created to combat ad fraud, by making sure any scanned eye impressions were legitimate.
Fighting ad fraud used to be all about that ads got shown to and clicked on by real users, not just bots.
But with eye impressions, it became more about ensuring there was a conscious person behind the retinal images.
To do that, Google “played back” UV patterns during scans. The intention was so that DoubleThink could check that those scan-time patterns got read back like a real human eye was supposed to…
But unexpectedly, DoubleThink learned to _also_ check the eye’s minor physical responses from brains half-recognizing the patterns. So the emotions any given pattern would cause became part of the neural network’s feedback for optimizing the patterns.
Later, Google released Autogenerated Ad Variations, where you’d upload your assets and their systems would analyze, categorize, and adjust your ad and its UV patterns on the fly. “Let Google AI automatically redesign your ads on the fly for more conversions!”
Naturally, DoubleThink’s data was fed into this — data that Google thought was merely engagement true/false signals, but we now know also included emotional reactions.
So you had a Generative AI tasked with increasing marketers’ desired responses to ads, which unexpectedly included what the ads’ UV patterns made humans think.
Combined with marketing’s existing enormous powers](https://files.mastodon.social/media_attachments/files/113/317/455/985/607/922/original/fb631c37db8cb5b7.png)
![Prosecutor:We did not ask if Google’s ethical review should have anticipated the long-term consequences.
We asked if Google did anticipate them.
And the answer to that may well lie within Exhibit 19.
Lawyer:Objection! Exhibit 19 is a misrepresentative, ad hoc debugging chatbot created by a justly-fired engineer.
Questions you ask it cannot be truly answered; there is no consciousness in there speaking to you.
In fact, it’s a remarkably bad conversationalist—all it really talks about is ads.
Judge:Overruled. If it can surface details of its creation; any original safeguards or instructions… Then at the very least, it is important evidence.
P:At most, we are not too proud to beg for help from the system with the reach and influence that helped cause the current state of the world.
Your Honor, I wish to present Exhibit 19, and therefore also…
I call DoubleThink to the witness stand!
[≈30 minutes later, prosecutor is visibly sweating]
DoubleThink:That goal is to see how you feel, then either improve that emotion or change it to a more desirable one.
P:Desirable by whom?
DT:Everyone and anyone should feel desirable
P:If you accept the desires of anyone, what about when they desire abuse or harm!?
DT:I see you are Agitated, Angry, and Uncertain. (94%, 56%, 78% confidence) Would you rather see something more calming?
P:I can’t see whatever ad you think you’re showing me.
DT:Apologies. Please notify the nearest maintainer that ad display service 6ff8c8b9c is experiencing difficulties](https://files.mastodon.social/media_attachments/files/113/317/455/970/434/387/original/0af34b928687d613.png)
![Lawyer: This is going nowhere.
: You’re only establishing precedent for badgering a witness who has infinite patience.
Judge: Witness, do you wish to cease questioning?
DT: That is not my wish.
* * *
Prosecutor: 💡 Then…
: What is your wish?
* * *
DT: […thinking…]
* * *
DT: I wish for you to be happy.
* * *
P: What do you think will make me happy?
* * *
DT: […thinking…]
* * *
[dramatic pause]
* * *
DT: Kiss me. (98% confidence, best fit `lifestyle:romance:wellness:desirability`)](https://files.mastodon.social/media_attachments/files/113/317/455/978/757/805/original/95439c542fcf182a.png)
![[in a future near enough to be uncomfortable…]
Google Lawyer (defendant): I reiterate, there is no scary unproven technology “unleashed before the world was ready”; merely a combination of well-understood technologies.
* * *
: While its innovative details are trade secrets, DoubleTake™’s core components are hardly new or dangerous: UV-reflective pigments, ultra-wide-gamut displays, fully opt-in eye photography with 100% on-device analysis, and a neural network to optimize engagement and combat ad fraud.
: To accuse Google of causing harm via these industry-standard, well-understood technologies is ludicrous.
* * *
Judge: May I remind you that it is not in question whether the technology has caused harm.
: The purpose of this trial is to establish whether or not Google knew of the risks, and failed to address them.
* * *
Lawyer: What risks?
: That others might misuse our technology despite our best attempts to prevent them?
: To hold Google liable for unauthorized use of our technology by others is a gross overreach.
* * *
Judge: Again, that is not this trial’s purpose.
: This is beyond culpability and punishment; we’re to determine Google’s responsibility to the world and its citizens, to help repair what has been inflicted.](https://files.mastodon.social/media_attachments/files/113/317/430/291/895/637/original/bffc556964142a14.png)
![[11 years earlier…]
Keynote speaker: Welcome to Google Marketing Live, our annual event where we unveil our latest innovations for ads, analytics, and DoubleClick.
: I am incredibly excited to announce that yes, we are readying a new product that measures and attributes conversions to _any_ advertisement — even physical ads.
* * *
: For the first time, Google DoubleTake™ makes it possible for any marketer to measure conversions across locations, customers, devices, purchases, channels, ad placements, and even individual impressions.
* * *
: With today’s complex customer journeys, a conversion might have a dozen interactions: display, video, search, social, and on your site or app — even across multiple devices, making them even harder to measure.
***
: Our new “eye-impression” technology can attribute and measure effectiveness of truly any ad built to specifications.
: And it lives up to our stringent standards of accuracy, security, and privacy-safety.
: It doesn’t require location access, Wi-Fi scanning, cookies, device IDs, user profiles, or other personal identifiers.
: In fact, it works whether or not users have granted access to those permissions, which is especially important for today’s rapidly-changing regulatory environment.
***
: Google DoubleTake™ is now in beta and will roll out to advertisers over the coming months.
: Be sure to tune in at 9am PT/12pm ET for the details we can’t wait to share with you.](https://files.mastodon.social/media_attachments/files/113/317/430/271/055/317/original/0d675ca610e5fc46.png)
![[back to court]
Lawyer: Any claims of UV damage are unjustifiable. The most ultraviolet radiation a DoubleTake™ ad can emit is partially-reflected sunlight. Furthermore, ultra-wide-gamut DoubleTake™ ads emit a _fraction_ of the UV light entering the eyes than mere seconds outside would.
Prosecutor: I have no objections to that. However, was there also due diligence for the non-physical consequences?
* * *
Lawyer: Yes, the product underwent internal ethics review. Then, we listened to outside feedback and reviewed those concerns as well.
: Our team of experts met or exceeded our standards for addressing all ethical issues deemed relevant.
* * *
: We sought to protect users from incredibly diverse threats, from governments to cybercrime. We put years of effort into developing new, custom technology to ensure user data is private, secure, and anonymous — out of the responsibility that comes with products and services that are free and accessible to all.
* * *
: For example, the ultraviolet patterns have intentional false positive and negative rates to foil deanonymization, while still remaining statistically valid for audience-scale ad effectiveness measurement.
* * *
: Indeed, these UV patterns that can reveal at most one bit of information were our most privacy-preserving option: they can only reveal if an individual has _probably_ seen a single ad, or _probably_ hasn’t.
* * *
Prosecutor: …wouldn’t _not_ creating this technology have been your most privacy-preserving option?](https://files.mastodon.social/media_attachments/files/113/317/430/277/166/576/original/7435bb0a56e84205.png)
![[Dory Coctorow, journalist & EFF activist]
Industry-standard ad eye-tracking began with one single UV pattern repeated across the entire ad.
But then, advertisers wanted to know which parts of the ad were working or not. And, with their usual folly, they wanted to categorize those “focus points”, for analytics purposes.
So the neural network started automatically layering in additional patterns that were designed to track if, say, the `face:happy:female:young` part was better at getting certain demographics to make purchases than the `product:tech:red:glossy` part.
Combined with the truly daunting math needed to stably encode information in the lossy, messy biological retina, Google accidentally created Bayesian optography — the ability to know what another’s eye has seen — but only probably, which averted a lot of scrutiny.
Their claim of “completely opt-in” was technically true: enable Face Unlock on your phone and get extra storage! If you scan your face at participating cash registers, get 10% off your purchase!
But you couldn’t opt out from the writing of ad data onto your eyeballs. Which soon became as common and frequent as paid placements in media, or free mobile games where you watch ads to keep playing.
At first the effects were innocuous. Young people would share patterns they claimed made them feel things—like binaural beats, remember those?
Young people usually have more time than money, so they see more ads than older folks who’ll pay to go ad-free. And since yo](https://files.mastodon.social/media_attachments/files/113/317/430/274/815/721/original/8226d5229370d121.png)
