Investigators used yellow printer tracking dots on the leaked NSA report to help identify Reality Winner. CBS News reports the document’s nearly invisible ‘DocuColor’ dot grid encodes the printer’s date, time, and serial number. Security researcher Rob Graham read the code on the published pages and determined they were printed on May 9 at 6:20 p.m. from a machine with serial number 29535218. According to the criminal complaint and CBS’s summary, an internal audit then narrowed who had printed the report, and only Winner had also emailed The Intercept from her work computer, leading agents to her door.
CurioWire Answers
Technology questions and answers
A focused collection of clear answers connected to CurioWire technology stories and nearby search topics.
Showing 47 technology questions.
The EURion constellation is a pattern of five small circles on many banknotes that allows scanners and image‑editing software to detect currency and prevent reproduction. Wikipedia notes researcher Markus Kuhn identified the motif in 2002 after a color copier refused to duplicate a 10‑euro note, and that in some devices the mere presence of five circles is enough to stop processing. The pattern appears in varied forms on currencies including the euro, U.S. dollar, and Japanese yen, and software modules associated with the Central Bank Counterfeit Deterrence Group also incorporate banknote detection.
EXIF metadata can reveal the camera make and model, capture date and time, and, when geotagging is enabled, the GPS coordinates of where a photo was taken. According to Wikipedia’s EXIF overview, these tags are embedded in the image file (commonly in the APP1 segment of JPEG/TIFF) and include technical details like exposure, aperture, and focal length. Manufacturers may add device‑specific information, including a serial number, in proprietary MakerNote fields. Because EXIF can expose location and device identifiers by default, the page highlights associated privacy risks for journalists, whistleblowers, and ordinary users.
Shine a blue LED light on the color print and examine it with a magnifying glass or microscope. EFF explains the dots are tiny, yellow, and repeated across each page, and that a blue light, a magnifying glass, or a microscope makes the pattern easier to see. Under blue illumination, the yellow marks become conspicuous, revealing a structured, repeating grid on the paper that signals the presence of a tracking pattern. This practical method lets consumers determine whether their color laser print includes machine‑identifying dots.
EFF warns that tracking dots threaten privacy and anonymous speech because no law limits their use strictly to anti‑counterfeiting. Its analysis says decoded dots can yield a printer’s serial number and manufacturer, which distributors can match to a purchaser; it cautions that agencies beyond the Secret Service—or even foreign governments and private entities—could exploit this to unmask pamphleteers or other speakers without notice or judicial oversight. EFF calls for legal protections and transparency around these marking technologies to preserve free expression and prevent misuse.
Netflix replaced its five‑star ratings with thumbs up/down in April 2017 to gather clearer, simpler feedback and make recommendations more personally relevant. Instead of public star averages, titles now display a personalized “percentage match” that reflects how likely Netflix thinks you’ll enjoy them based on your viewing habits and behavior. The company framed the shift as a way to get direct signals about taste—thumbs indicating whether Netflix’s prediction was right or wrong—so its algorithms can adjust. In practice, the experience emphasizes private, individualized matching over crowd scores, with the thumbs acting as explicit inputs while the percentage match communicates predicted fit for each member.
Explicit feedback is when users actively state preferences—for example, rating an item, ranking choices, or selecting one of two options—while implicit feedback is inferred from behavior, such as what you view, how long you watch (dwell time), what you click, or what you purchase. Recommender systems commonly leverage both: explicit signals are clearer but sparse, and implicit signals are abundant but noisier and require interpretation as preference indicators. Typical explicit examples include 1–5 ratings or creating a favorites list; implicit examples include watch history and viewing duration. Many systems blend these approaches within hybrid recommenders to improve accuracy and mitigate issues like data sparsity and cold start.
The 2017 settlement required Vizio to pay $2.2 million and to prominently disclose and obtain affirmative express consent before collecting or sharing smart‑TV viewing data. The stipulated federal court order also mandated deletion of data gathered before March 1, 2016 and the implementation of a comprehensive privacy program with biennial assessments. According to the FTC’s complaint, Vizio had installed software that captured second‑by‑second information from 11 million TVs and appended demographic details, then sold the information to third parties for uses including targeted advertising. The order bars misrepresentations about privacy and codifies consent and programmatic safeguards around television viewing data.
Yes. YouTube states that likes help its system predict your interest in similar videos, while dislikes (and “Not interested” or “Don’t recommend channel”) signal what to avoid recommending in the future. These are among the primary signals YouTube lists—along with watch history, search history, subscriptions, and satisfaction surveys—that personalize Home and Up Next suggestions. Different parts of YouTube emphasize different signals: the video you’re currently watching is the main input for what plays next, whereas Home relies more on your watch history. Using these controls, viewers can directly influence which topics and channels appear less or more frequently in recommendations.
Thumbs on Pandora directly steer a station’s playlist. A Thumbs Up indicates you like a track and leads Pandora to play similar songs and artists on that station, while a Thumbs Down both skips the track immediately and tells the system to avoid that song (and similar material) on the station going forward. Pandora’s station model applies these signals to refine the mix, using your explicit inputs to broaden or narrow what you hear. The platform’s features note that repeated downvotes can exclude tracks or even an artist from a given station, aligning the station’s output with your stated preferences.
Yes—many modern LCD TVs still ship with overscan enabled by default, but you can usually disable it in the TV’s on‑screen settings. Overscan is when the television crops the outer edge of the active picture, a legacy behavior from the analog CRT era. Although flat‑panel displays generally don’t require overscan, some models retain it, which can hide interface elements near the borders. Turning overscan off tells the TV to show the full transmitted image rather than a slightly zoomed, edge‑trimmed version. If your set offers a menu option to modify this behavior, disabling it ensures that graphics and text at the edges remain visible.
Android’s TV guidelines say some TV devices may clip the outer edge of app layouts, so any elements that must always be visible should be placed within an overscan‑safe area. If you build with the AndroidX Leanback toolkit, you shouldn’t add your own extra margins, because Leanback’s templates already include overscan‑safe padding. This prevents critical text or controls from being cropped by specific set behaviors while avoiding redundant spacing that would push content too far inward. The advice is particularly relevant for playback screens, where transport controls, titles, and metadata need to remain readable from couch distance on varied living‑room displays.
Microsoft’s Xbox app guidance recommends keeping interactive or essential UI within a TV‑safe area roughly 48 pixels from the left and right edges and 27 pixels from the top and bottom after drawing content to full screen. The Windows Developer Blog explains that some televisions hide the extreme edges, so once you extend your app to the screen bounds, you should inset key controls into this safe region. Following these margins helps focus navigation behave reliably and keeps text and buttons consistently visible across different TVs and viewing arrangements, without leaving obvious borders when content is allowed to reach the screen edge.
In iOS, the safe area is exposed via UIView’s safeAreaLayoutGuide, which defines the portion of a view not covered by the status bar or other visible bars, and developers create constraints to keep content inside it. Apple’s documentation notes that for a view controller’s root view, the guide automatically accounts for system bars and any additional insets you specify via additionalSafeAreaInsets. Constraining to this guide adapts layouts across different device configurations without hard‑coding offsets, helping ensure tappable and readable elements don’t collide with system UI while maintaining consistent, edge‑aware spacing throughout an app.
LG’s webOS TV developer guide specifies a 20‑pixel margin around all four screen edges as the overscan‑safe area, and it recommends placing essential content within this region. This practice anticipates that certain TVs or modes may clip the extreme perimeter, so reserving this margin helps ensure buttons, menus, and text remain readable and fully visible in living‑room viewing. Because webOS defines this screen‑level safe area, app layouts should be designed to respect it consistently across scenes rather than adding ad‑hoc padding only for specific views or components.
The Bell System adopted all-number calling to expand numbering capacity and replace exchange-name prefixes, starting in 1958. Numbers became fully numeric (area code, central office code, and line number), increasing the pool of assignable prefixes and averting the exhaustion forecast under the older 2L–5N scheme. Field tests showed fewer dialing errors with the new format. The shift drew public resistance—groups such as the Anti‑Digit Dialing League opposed it—and AT&T produced explanatory materials to ease adoption. Despite backlash in some big cities, the plan expanded during the 1960s as carriers standardized dialing to support growth and direct distance dialing.
ITU‑T Recommendation E.161 defines the standard mapping of letters to digits on telephone keypads, including PQRS on 7 and WXYZ on 9, and also specifies keypad arrangement and symbols. In force since 1988 (version 02/2001), it supports uses such as phonewords and multi‑tap text entry. Earlier practice varied: until the 1990s many layouts omitted Q and Z, placed them on the 1 key, or used PRS on 7 and WXY on 9. As texting and phonewords required the full alphabet, E.161 resolved these inconsistencies so devices display a consistent alphabet‑to‑number mapping today.
Phone keypads use 1–2–3 on the top row because Bell Labs’ human‑factors testing in the late 1950s found that layout worked best for dialing, and it became standard with DTMF push‑button phones. Calculators and computer numpads retained the adding‑machine tradition with 7–8–9 on top, optimized for rapid numeric entry. Bell Labs evaluated multiple arrangements and selected the current telephone pattern after extensive user studies; subsequent networks and services adopted it. The difference reflects separate design histories and tasks—the telephone’s layout optimized casual dialing accuracy and ease of use, while calculators optimized professional data‑entry speed.
“PEnnsylvania 6‑5000” was the Hotel Pennsylvania’s main number written in New York’s 2‑letters‑5‑numbers style and became famous via the Glenn Miller Orchestra’s 1940 hit of the same name. The letters PE mapped to 7 and 3, making the number 736‑5000 (later with Manhattan’s 212 area code). The hotel long claimed it was New York City’s oldest continuously used telephone number, and callers for years heard a greeting that included the song. The Pennsylvania exchange served the area around Penn Station. Although the hotel closed in 2020, the number’s exchange‑name origin and musical tie cemented its cultural resonance.
Vanity numbers are telephone numbers selected for memorability and branding, often expressed as phonewords like 1‑800‑FLOWERS, to make a business easier to reach and recall. They may be toll‑free or local and can also rely on repeated digits or patterns (for example, 1‑800‑800‑8000) as mnemonic devices. Companies request such sequences specifically for marketing so advertisements can feature a word or distinctive pattern instead of arbitrary digits. Beyond phonewords, memorable numeric sequences tied to brands, industries, or even broadcast frequencies are also used as vanity numbers to reinforce recognition across media.
South Korea’s Telecommunications Technology Association (TTA) set a 60–68 dB shutter‑sound standard for camera phones. The industry rule was created to curb privacy invasions, and it specifies that new camera‑phone models must emit an audible click in that range whenever a photo is taken. The standard followed a government camera‑phone use initiative and was developed with input from handset makers, mobile carriers, and academics before being finalized as a TTA standard. It applied to devices released from July 1, 2004, anchoring today’s region‑specific behavior on many Korean‑market smartphones.
Apple explicitly states that FaceTime isn’t available in the United Arab Emirates. On its support page covering wireless carrier features for the Middle East and India, Apple notes the UAE restriction and clarifies that in that country or region the wireless provider, not Apple, handles iPhone support, reflecting local regulations. This is a clear example of region‑specific feature availability in mainstream devices, where a core app can be restricted or disabled depending on where a device is sold or used.
Japan enacted a nationwide offense of photographing or filming on July 13, 2023, criminalizing the non‑consensual taking of sexually explicit images or videos. The statute also covers cases where someone is deceived into believing images will not be shown to others and includes provisions allowing prosecution of certain offenses committed abroad by Japanese nationals. The reform provides a clear, uniform legal basis to address voyeuristic acts across the country, distinct from any device‑level camera sound practices.
Saudi Arabia banned the sale of camera phones nationwide in April 2004, then reallowed their sale in December 2004. The move reflected privacy and security concerns surrounding early mobile cameras and contrasts with other markets that permitted camera phones but relied on audible shutter cues or industry standards. The quick reversal underscores how early camera‑phone policies varied by country and evolved rapidly as the technology spread.
Twitter set tweets to 140 characters so a post could fit in a single 160‑character SMS, reserving 20 characters for the sender’s username. This constraint was present at launch in 2006 and quickly became part of Twitter’s identity, encouraging speed and brevity in how people wrote and read updates. The SMS-rooted cap linked early tweeting to mobile texting conventions, shaping a concise, headline-like style that persisted even as the service later adjusted what counted toward the limit and expanded the maximum length.
Emoji began in 1999 when Shigetaka Kurita designed a 176‑icon, 12×12‑pixel set for NTT DoCoMo’s i‑mode mobile platform in Japan. Intended to streamline mobile communication and differentiate the service, the pictographs gained traction domestically and were later standardized and propagated as major platforms incorporated emoji support. Subsequent work by Unicode and vendors like Google and Apple helped align and broaden the sets, turning a localized, carrier‑specific idea into a global visual language used across messaging apps, operating systems, and keyboards.
SMS falls to 70 characters when the message uses 16‑bit Unicode (UCS‑2/UTF‑16) encoding instead of the default GSM 7‑bit alphabet. An SMS payload fits 140 bytes; with 7‑bit encoding that allows up to 160 characters, but with 16‑bit encoding only 70 characters fit. Handsets switch encoding when a message includes characters outside the GSM 7‑bit set (for example, many non‑Latin scripts or certain symbols), which reduces the per‑message character capacity and may trigger segmentation for longer texts.
Telegrams were billed by the word, so writers compressed messages to minimize cost, creating a clipped “telegram style.” Telegraph companies enforced word rules (e.g., up to 15 characters per plain‑language word and 10 for coded words), which reinforced brevity and omission of nonessential words. This economic pressure produced a distinctive shorthand—abbreviations, compressed phrasing, and minimal punctuation—that influenced later concise forms of communication and remains a reference point for terse, information‑dense writing.
RCS upgrades SMS by operating over mobile data and adding modern features like read receipts, group chats, typing indicators, and higher‑quality media sharing. Critically, unlike SMS’s 160‑character cap, RCS does not innately impose a text character limit, enabling much longer messages within the same conversation flow. This makes RCS functionally closer to internet chat apps while remaining integrated with a phone’s default messaging experience, provided both parties and their networks support RCS.
Spotify’s Fewer Repeats shuffle generates several mathematically random orders, scores each one for “freshness” based on how recently you’ve listened to the tracks, then picks the sequence with the highest freshness score—so recently played songs are nudged later and you hear fewer repeats. Standard Shuffle, which assigns random values to tracks and orders them purely at random, remains available, but Fewer Repeats is the default for Premium. This approach preserves randomness while aligning with listener intuition by selecting among random sequences rather than deterministically reordering your playlist.
The Fisher–Yates shuffle is a list-shuffling algorithm that produces an unbiased permutation, meaning every possible ordering is equally likely. It works by iterating through the array and swapping each element with a randomly chosen element from the remaining positions, yielding uniform results when paired with a good random number source. The article contrasts this with naïve methods (like swapping with any index each time), which are provably biased, and notes common implementation pitfalls such as off‑by‑one errors and modulo bias that can reintroduce skew if not handled carefully.
The clustering illusion is the tendency to mistakenly view the inevitable streaks or clusters in small samples from random processes as non‑random. People underpredict how much variability appears in small samples and rely on heuristics like representativeness, which makes natural clumps look suspicious. Classic illustrations include analyses of WWII bombing maps and perceived patterns in stock prices. The bias helps explain everyday misunderstandings of chance: random data often contains apparent runs, but our cognition expects smoother dispersion than genuine randomness produces.
Apple’s iTunes 5 added Smart Shuffle, a setting that lets users change the “randomness” of shuffled songs. Apple described it as giving listeners greater control over random playback, alongside other 2005 updates like a new Search Bar and playlist folders. In practice, Smart Shuffle allowed adjusting how likely similar songs might appear in succession, reflecting Apple’s response to user perceptions of random playback while keeping the experience easy to tailor.
Play Something isn’t random shuffle; it starts a title Netflix thinks you’ll like based on your interests and prior viewing. At launch, Netflix said it may select something you’re already watching, a show or film on your list, an unfinished title to resume, or a new recommendation from its personalization algorithms. The feature evolved from earlier “Shuffle Play” tests and rolled out globally on TVs in April 2021, later reaching Android, as a way to move users from browsing to watching without playing a random selection.
Motorola’s iTap differed from T9 by offering one‑keypress word completions, phrase predictions, and a distinct interface that surfaced suggestions immediately. iTap could guess the remainder of a word once several letters were entered and presented alternatives in order of common usage. It stored unknown words for later selection and emphasized completions and common phrases to speed input. Developed by Motorola as a direct T9 competitor, it relied on a built‑in dictionary and contextual prediction while leaving UI choices that contrasted with many T9 implementations, making the experience feel more like autocompletion than simple disambiguation.
German engineer Friedhelm Hillebrand set the 160‑character limit for SMS after experiments showed that length was sufficient for short text messages. As chairman of the GSM non‑voice services committee in 1985, he evaluated how much space typical messages needed and concluded that 160 characters captured the essence of most communications. That figure became foundational to the GSM standard and later influenced early digital communication norms, including Twitter’s original 140‑character cap. The limit tied message design to concise expression and shaped how early mobile users wrote and shared information.
Swype is a virtual keyboard that lets users enter words by sliding a finger from the first to the last letter, lifting only between words, and it was created by Swype Inc. founder Cliff Kushler. The software combines error‑correction and a language model to predict the intended word and enabled typing speeds competitive with traditional tapping. Swype debuted on devices like Samsung’s Omnia II, was acquired by Nuance in 2011, and its mobile apps were discontinued in 2018 after years of influence on gesture‑based text input.
Modern keyboards like Gboard predict words with on‑device machine learning, using a neural spatial model and finite‑state transducers to interpret noisy taps and generate suggestions and corrections efficiently. Google describes a character‑level spatial model that maps touch points to intended keys and compact neural components tuned to run quickly on phones. These techniques power next‑word suggestions, autocorrect, and multilingual support while keeping latency low and functionality available offline, showing how predictive text evolved from simple dictionaries to advanced, device‑optimized language models.
Look for the Thunderbolt icon plus the number 4 on the cable’s connectors. Intel states that all Thunderbolt 4 connectors are marked with the Thunderbolt symbol and the numeral 4, making genuine TB4 cables easy to recognize. Intel also notes Thunderbolt cables are certified and that Thunderbolt 4 cables can provide top transfer speeds at lengths up to 2 meters, which helps distinguish them from many earlier high‑speed USB or Thunderbolt options. These markings and certification cues reduce guesswork about compatibility with TB4 docks, displays, and high‑speed storage connected through USB‑C ports on modern systems.
USB Power Delivery 3.1’s Extended Power Range allows up to 240 W by using 48 V at 5 A, and Type‑C cables generally must include E‑Marker chips that report their capabilities. Wikipedia notes that all Type‑C cables, except the minimal combination of USB 2.0 and only 3 A, must contain E‑Markers identifying current limits, signaling support, and whether the cable is passive or active. This identification enables safe negotiation for higher voltages and currents under EPR. In short, 240 W operation relies on compliant Type‑C cables with E‑Markers and appropriate 5 A support, while basic 3 A USB 2.0‑only cables do not require E‑Markers and won’t support EPR levels.
USB4 Version 2.0 can achieve 80 Gbit/s over existing passive “USB 40Gbps” cables using a new PAM3 encoding scheme, so new passive cables aren’t strictly required. However, active cables are not forward‑compatible in the same way; the specification adds a new speed grade for active cables to support the higher rates. USB4 Version 2.0 also upgrades DisplayPort tunneling to DisplayPort 2.1, enabling higher‑bandwidth video carriage within the same link. Practically, many passive 40Gbps USB‑C cables can deliver the new 80Gbps mode, while active cables may need to be the newer grade to realize full performance.
DisplayPort Alt Mode 2.0 brings full DisplayPort 2.0 capabilities to USB‑C and provides seamless interoperability with USB4. VESA explains it enables beyond‑8K resolutions and up to 80 Gbps of DisplayPort video data over the USB‑C connector, or 40 Gbps with simultaneous SuperSpeed USB data. The spec integrates with USB4 for discovery, configuration, and power management, and DisplayPort can also be tunneled over USB4 and Thunderbolt. Together, these updates make USB‑C a single‑cable path for very high‑resolution displays, data, and power, spanning use cases from docking and gaming to AR/VR and pro HDR workflows.
HDMI Alternate Mode for USB‑C is listed as “Not being updated.” Wikipedia’s USB‑C Alternate Modes table shows HDMI Alt Mode (announced in 2016 and based on HDMI 1.4b) is no longer being updated, while DisplayPort Alt Mode remains current. The same section notes that DisplayPort signals can be carried either directly via DisplayPort Alt Mode or encapsulated in Thunderbolt, and that DisplayPort 2.0 can run alongside USB4 over USB‑C. These status notes clarify why USB‑C display solutions typically center on DisplayPort technologies rather than native HDMI Alt Mode.
UEFI standardizes removable-media boot by defining a firmware boot manager and a default loader path, so devices like USB sticks can boot automatically via \EFI\BOOT\BOOT .EFI (for example, \EFI\BOOT\BOOTX64.EFI on x86‑64). Unlike legacy BIOS, which relied on boot sectors and device‑specific quirks, UEFI discovers OS loaders by path and executes them directly. The firmware reads the EFI System Partition, supports FAT12/16/32 for this partition, and can detect loaders for multiple architectures. This approach removes many historical inconsistencies and makes booting from USB or other removable media predictable across systems.
PXE is a client‑server boot framework that starts a system by fetching a software image over the network on PXE‑enabled hardware. A PXE‑capable NIC and firmware use standard protocols like DHCP to obtain configuration and TFTP to download a network bootstrap program, which then loads the operating system components. Originating in Intel’s Wired for Management, PXE became part of UEFI and is widely used in data centers for OS installation and deployment. It enables diskless booting, consistent imaging workflows, and centralized control when removable media are impractical or unavailable.
Hybrid ISO images embed both an El Torito boot record (for optical discs) and a master boot record (for hard‑disk–like media), allowing the same ISO to boot from a CD/DVD or when written directly to a USB flash drive. ISOLINUX, part of the Syslinux project, added support for creating such “hybrid ISO” images so a single distribution image can serve both use cases. This improves installation reliability and convenience by eliminating the need for separate optical and USB images while preserving compatibility with legacy BIOS and modern systems.
Phasing out the Compatibility Support Module (CSM) removes legacy BIOS‑style booting, leaving systems to boot natively via UEFI’s boot manager and standardized loader paths. UEFI firmware can automatically find loaders on removable media using default paths such as \EFI\BOOT\BOOTX64.EFI, and vendors have announced CSM retirement timelines, reducing availability of legacy boot modes. As a result, install and recovery media need to provide UEFI‑compatible loaders and file systems rather than relying on BIOS boot sectors or compatibility modules.