Hindsight · review build · data not yet validated
50 / 50
claims audited
6%
error rate so far (2 corrected, 1 contested)
Read-only. The auditor agent records its decisions in the Hindsight dataset.
Individuals primarily use portable computers.
Grader A
Notebook PCs passed desktop PCs in worldwide unit shipments in Q3 2008 (38.6M vs 38.5M, iSuppli). Mobile phones also far outnumbered PCs. Portable devices were the primary new computers by 2009.
Test: At least 50% of new PC units: notebooks were about 50% of PC shipments in Q3 2008 and rising.
Grader B
Worldwide notebook shipments passed desktop shipments in Q3 2008, and smartphones added more portable computing on top. By 2009 most new personal computers were portable.
Test: New-unit test: notebooks >50% of PC shipments from Q3 2008 (38.6M vs 38.5M).
Portable computers will have become dramatically lighter and thinner than the notebook computers of ten years earlier.
Grader A
By 2008-2009 netbooks weighed about 1 kg (2.2 lb) and took about 20% of the portable market; ultra-thin notebooks such as the 2008 MacBook Air shipped. Typical 1999 notebooks weighed 6-7 lb. Smartphones added a much lighter class of portable computer.
Test: At least 20% of new portable units: netbooks peaked at about 20% of the portable computer market, plus thin-and-light notebooks.
Grader B
1989 notebooks weighed about 6-7 lb and were about 2 in thick. In 2008-2009 the MacBook Air was 3 lb and 0.75 in, and netbooks weighed about 2 lb.
Test: Device test: ultraportables and netbooks (netbooks near 20% of portable market) at roughly half the weight and thickness of 1989 notebooks.
Personal computers are available in a wide range of sizes and shapes, and are commonly embedded in clothing and jewelry such as wristwatches, rings, earrings and other body ornaments.
Grader A
In 2009 smartwatches were experimental specialty items; computers in clothing, rings and earrings did not exist as products. Wrist wearables became commercial only from 2012-2015, and computers in clothing and jewelry never became common.
Test: At least 20% of users wearing computers in clothing or jewelry: effectively 0% in 2009; clothing, ring and earring computers still well under 5% by 2014.
Grader B
In 2009 no computers were commonly embedded in clothing or jewelry. Smartwatches arrived 2013-2015 and stayed below mainstream by 2014; ring and earring computers never became common.
Test: Household test: wearable computers well below 5% of consumers by 2014; rings/earrings ~0%.
People typically have at least a dozen computers on and around their bodies, which are networked using “body LANs” (local area networks).
Grader A
In 2009 a typical person carried one or two computing devices (phone, maybe a music player or laptop). Bluetooth personal area networks linked a phone to a headset, but a dozen networked body-worn computers was not typical, and it was not typical by 2014 either.
Test: At least 20% of people with 12 or more networked on-body devices: effectively 0%.
Grader B
In 2009 a typical person carried one to three devices (phone, maybe a media player and Bluetooth headset). Nothing near a dozen networked on-body computers existed, then or by 2014.
Test: Household test: users with ~12 networked body devices ~0%.
Most users have servers in their homes and offices where they keep large stores of digital “objects,” including their software, databases, documents, music, and movies.
Grader A
Dedicated home and office servers holding users' media and software stayed a niche. Microsoft ended its Windows Home Server line in 2013. People kept files on their PCs and increasingly moved to online services (69% of US internet users already used some cloud service in 2008).
Test: At least 20% of households with a dedicated home server: far below 5% in 2009 and 2014.
Grader B
Home servers (Windows Home Server, NAS boxes) stayed a small niche. Users kept files on their PCs and moved toward cloud services, not personal servers.
Test: Household test: dedicated home servers well below 5% of households by 2014.
Digital objects such as books, music albums, movies, and software are rapidly distributed as data files through the wireless network, and typically do not have a physical object associated with them.
Grader A
Digital distribution was large but not typical across media in 2009. Digital channels were 27% of global recorded-music revenue in 2009; e-books were small (Kindle sold about 1.5-3 million units by end 2009; 17-21% of Americans read an e-book only in 2011-2012). Movies and software were still mostly sold on discs. Most delivery was wired broadband, not wireless.
Test: Share of media sold without a physical object: music about 27% of revenue in 2009 (above 5%, below majority); books well under 5% in 2009; e-book reading reached about 21% of Americans in 2012.
Grader B
Music went digital first: digital sales passed physical by 2011. E-books were about 3% of US trade sales in 2009 and grew only into the 2012 period; movies and software were still mostly on disc in 2009. Most downloads went over wired broadband with Wi-Fi at the edge.
Most users have servers where they keep digital “objects” such as virtual reality environments (although these are still at an early stage).
Grader A
Personal servers were rare (see 008) and virtual reality environments were not a common kind of personal digital object. The largest virtual world, Second Life, peaked at about 88,000 concurrent users in 2009 and then stalled.
Test: At least 20% of users keeping VR environments on their own servers: effectively 0%.
Grader B
Users did not keep virtual reality environments on personal servers in 2009 or by 2014.
Test: Household test: ~0%.
There are services to keep one’s digital objects in central repositories, but most people prefer to keep their private information under their own physical control.
Grader A
Central repositories existed in 2009 (webmail, online photo and file storage), and 69% of US internet users used at least one such service in 2008. But most personal files (documents, photos, music) stayed on users' own PCs, and 90% of users said they would be very concerned if a provider sold their data. The two-part description matches 2009.
Test: Two-part test: central storage services used by at least 20% of users (69% used some cloud service, mostly webmail) and the majority of personal files still kept locally (online file storage was a minority activity in 2008-2009).
Grader B
Online file repositories existed (Dropbox had 3M users by Nov 2009, plus webmail and photo sites), but most people kept their files on their own PCs and drives in 2009.
Test: User test: cloud file-sync users ~1-2% of internet users in 2009; local storage the majority practice.
Cables are disappearing. Communication between components, such as pointing devices, microphones, displays, printers, and the occasional keyboard uses short-distance wireless technology.
Grader A
Short-range wireless (Bluetooth, RF mice and keyboards, Wi-Fi printers) was common in 2009, and Bluetooth headsets were mainstream for phones. But most desktop peripherals, displays and printers still connected by cable in 2009; cables were not disappearing.
Test: At least 50% of new peripheral units wireless: met for phone headsets connectivity in phones, not met for displays, printers or most mice and keyboards in 2009. A measurable share above 5% exists.
Grader B
By 2009 Bluetooth headsets and wireless mice/keyboards were common and Wi-Fi printers were appearing, but displays, most printers and most desk peripherals still used cables.
Test: Share test: wireless peripherals (headsets, mice) well above 5% of users; cable-free component links not the norm by 2011.
Computers routinely include wireless technology to plug into the ever-present worldwide network, providing reliable, instantly available, very high bandwidth communication.
Grader A
By 2009 notebooks (half of PC shipments) routinely shipped with Wi-Fi, and 56% of Americans had gone online wirelessly. Mobile broadband passed 600 million subscriptions worldwide at end 2009. The 'very high bandwidth, reliable, ever-present' part was overstated, but the device feature was standard.
Test: At least 50% of new portable computers shipping with wireless networking: met in 2009; 56% of Americans had used wireless internet access (Pew, April 2009).
Grader B
Wi-Fi was standard in nearly all 2009 notebooks and in smartphones; 56% of US adults had gone online wirelessly by April 2009. The bandwidth was not 'very high' by later standards, but the routine inclusion is met.
Test: Device test: Wi-Fi in >50% of new notebooks by 2009; user test 56% of US adults used wireless internet.
The majority of text is created using continuous speech recognition (CSR) dictation software, but keyboards are still used. CSR is very accurate, far more so than the human transcriptionists who were used up until a few years ago.
Grader A
In 2009 almost all text was typed. Dictation software was a niche tool. Speech recognition reached parity with professional human transcribers on the standard conversational benchmark only in 2016 (5.8% vs 5.9% word error). Voice assistants arrived with Siri in late 2011 and did not make speech the main way to create text.
Test: Majority of text created by dictation: well under 5% of text in 2009 and 2014; CSR accuracy below human transcribers until 2016.
Grader B
Speech dictation (Dragon) stayed a niche tool; keyboards created almost all text in 2009 and in 2014. Machine transcription reached human parity only on a 2017 benchmark.
Test: Share test: text created by dictation well below 5% by 2014.
Also ubiquitous are language user interfaces (LUIs), which combine continuous speech recognition (CSR) and natural language understanding. For routine matters, such as simple business transactions and information inquiries, LUIs are quite responsive and precise. They tend to be narrowly focused, however, on specific types of tasks. LUIs are frequently combined with animated personalities. ...
Grader A
Narrow speech interfaces for routine transactions existed: phone IVR systems with speech recognition (AT&T from 1992) were common in call centers, and Google's GOOG-411 ran from 2007. But they were not 'ubiquitous' in daily computing, were not known as precise, and were rarely combined with animated personalities. Smartphone voice assistants arrived from 2011 (Siri).
Test: Routine use by at least 20% of the relevant industry (large call centers): plausibly met for speech IVR; general-user daily use of language interfaces under 20% in 2009, reached only with Siri (2011) and later assistants.
Grader B
Speech-driven phone systems for narrow tasks (airline, bank, directory services like GOOG-411) were widespread by 2009, and Siri arrived in 2011. 'Ubiquitous' overstates 2009, and animated personalities on these interfaces were rare.
Computer displays have all the display qualities of paper — high resolution, high contrast, large viewing angle, and no flicker. Books, magazines, and newspapers are now routinely read on displays that are the size of, well, small books.
Grader A
E Ink readers (Kindle from November 2007, 6-inch 167 ppi) gave paper-like, flicker-free, wide-angle reading, but with 4-level grayscale and slow refresh; mainstream LCDs did not have paper qualities. Kindle sales were about 1.5-3 million by end 2009, so book-size reading devices were not routine. E-book reading reached 21% of Americans only in February 2012.
Test: At least 20% of users routinely reading books, magazines or newspapers on book-size displays: well under 20% in 2009; about 21% had read an e-book in the past year by 2012.
Grader B
E-ink readers (Kindle 2007) gave paper-like, flicker-free book-sized displays, but with lower contrast than paper. E-books were 3% of US trade sales in 2009 and passed hardcovers only in Q1 2012.
Computer displays built into eyeglasses are also used. These specialized glasses allow users to see the normal visual environment, while creating a virtual image that appears to hover in front of the viewer. The virtual images are created by a tiny laser built into the glasses that projects the images directly onto the user’s retinas.
Grader A
Retinal-scanning laser displays in eyeglasses stayed in research and defense prototypes (Microvision Nomad) in 2009. No consumer product existed. Later smart glasses (Google Glass, 2013) used a prism display, not retinal lasers, and stayed far below 5% of users.
Test: At least 20% of users with eyeglass displays: effectively 0% in 2009 and under 5% in 2014.
Grader B
Retinal-projection eyeglass displays stayed prototypes and defense projects; first commercial retinal-projection glasses came in 2018. Google Glass (2013, not laser-retinal) never reached 5%.
Test: User test: eyeglass displays ~0% of consumers by 2014.
Computers routinely include moving picture image cameras and are able to reliably identify their owners from their faces.
Grader A
Built-in webcams were common on 2009 notebooks, and Lenovo, Asus and Toshiba shipped face-recognition login. But the face login was not reliable: in February 2009 researchers defeated all three with a photograph of the owner. Reliable face identification of owners became mainstream only with Face ID in 2017.
Test: Two parts: built-in cameras on at least 50% of new notebooks (met or close in 2009); reliable face identification of owners (not met; below 5% in reliable form until 2017).
Grader B
Built-in webcams were common in 2009 notebooks. Reliable face recognition of owners was not routine: early laptop face-login was unreliable, and Kinect (2010) identified players only in one product; phone/PC face unlock became mainstream after 2014.
Test: Device test: webcams in >50% of new notebooks (hit); reliable owner face-ID well under 50% of new devices by 2014.
In terms of circuitry, three-dimensional chips are commonly used, and there is a transition taking place from the older single-layer chips.
Grader A
In 2009 logic and memory chips were planar; the only common 3D element was stacking of separate dies in packages. Intel shipped '3-D' tri-gate transistors in Ivy Bridge in April 2012, and Samsung started 3D (V-NAND) flash in 2013. So a transition to three-dimensional chips started in 2012-2014, not by 2009.
Test: At least 50% of new chips or routine use by at least 20% of chipmakers: not in 2009; tri-gate transistors became standard in Intel CPUs from 2012 and 3D NAND started in 2013.
Grader B
Stacked-die packages and multi-layer flash chips existed by 2004-2007 and were used in phones, but mainstream 3D chips (TSV DRAM 2011, HBM 2013, 3D NAND) came in the 2010s. In 2009 single-layer chips dominated.
Test: Industry test: 3D stacked chips >5% of memory/mobile packaging by 2009; monolithic 3D mainstream only mid-2010s.
Documents, circa 2009, routinely include embedded moving images and sounds.
Grader A
By 2009 web documents routinely embedded video and audio: 62% of US adult internet users watched video on sharing sites such as YouTube, whose players were embedded across web pages, and 19% did so on a typical day. Office documents also supported embedded media, although printed-style documents usually did not use it.
Test: At least 20% of users encountering documents with embedded moving images: 62% of adult internet users watched online video in 2009 (Pew).
Grader B
By 2009 web pages, the dominant form of electronic document, routinely embedded video and audio (YouTube embeds, news-site video).
Test: Share test: embedded video/audio routine on major web pages by 2009 (>20% of major sites); rare in word-processor files.
Students interact with their computers primarily by voice and by pointing with a device that looks like a pencil.
Grader A
In 2009 students used computers with keyboards and mice. Voice input was not used in classrooms, and pen input was limited to niche tablet PCs and interactive whiteboards. Touchscreens (finger, not pencil) spread from 2010, and voice never became the primary student input.
Test: At least 20% of students interacting primarily by voice and stylus: under 5% in 2009 and 2014.
Grader B
Students used keyboards and mice (later touch), not voice and stylus, in 2009 and 2014.
Test: Share test: voice/stylus primary interaction <5% of students by 2014.
Keyboards still exist, but most textual language is created by speaking.
Grader A
Same finding as 014: in 2009 almost all text was typed. Speech recognition reached human-transcriber parity on the standard benchmark only in 2016, and smartphone voice assistants started in late 2011.
Test: Majority of text created by speech: well under 5% in 2009 and 2014.
Grader B
Speech was not the main source of text in 2009 or by 2014; dictation stayed niche.
Test: Share test: dictated text <5% by 2014.
Recent controversial studies have shown that students can learn basic skills such as reading and math just as readily with interactive learning software as with human teachers, particularly when the ratio of students to human teachers is more than one to one. Although the studies have come under attack, most students and their parents have accepted this notion for years.
Grader A
The main US federal study of reading and math software (IES, February 2009) found no significant difference in achievement between classes using the products and classes not using them, in either year. There was no body of studies showing software matched human teachers, and there is no evidence that most students and parents accepted that idea.
Test: Acceptance by a majority of students and parents that software teaches basic skills as well as teachers: no evidence of this in 2009; the main study showed no effect.
Grader B
No body of studies showed software teaching basic skills as well as teachers, and parents did not broadly accept the idea. Federal evaluations of that era found weak or mixed effects of reading/math software.
Test: Acceptance test: no majority parental acceptance by 2014.
Many children learn to read on their own using their personal computers before entering grade school.
Grader A
No evidence shows that many children learned to read on their own from personal computers before school in 2009. Few young children had a personal computer, and the federal evaluation of early reading software (first grade) found no significant achievement gains.
Test: At least 20% of children learning to read independently via their own computer before grade school: no evidence of more than a small share in 2009 or 2014.
Grader B
There is no evidence that many children taught themselves to read on PCs before school; early literacy software was supplementary.
Test: Share test: self-taught reading via PC well below 5% of children.
Price deflation concerned economists in the early ’00 years, but they quickly realized it was a good thing. The high tech community pointed out that significant deflation had existed in the computer hardware and software industries for many years earlier without detriment.
Grader A
Deflation did concern economists in the early 2000s, but they did not conclude it was a good thing. The Federal Reserve treated deflation as a danger to prevent and cut rates to 1% to avoid it. The consensus view stayed negative.
Test: Consensus test: the dominant policy view (Federal Reserve) called sustained deflation destructive in 2002, the opposite of the claim.
Grader B
Economists did worry about deflation in the early 2000s: Fed Governor Bernanke gave a speech on it in November 2002. But the worry did not end in a view that deflation was good. Bernanke called sustained deflation 'highly destructive' and said the Fed should resist it. US CPI inflation stayed positive in every year from 1999 to 2008 (1.6% to 3.8%). The only negative year was 2009 (-0.36%), and it was treated as a crisis symptom. The first clause is true. The core claim (deflation accepted as good) is false.
Test: Not an adoption claim. Test: did price deflation occur, and did the economic consensus accept it as benign? US CPI: positive every year 2000-2008; -0.36% in 2009. Fed consensus: deflation is a danger.
The United States continues to be the economic leader due to its primacy in popular culture and its entrepreneurial environment.
Grader A
The US stayed the largest economy in the world in 2009-2010, at more than twice the size of number two. The stated causes (popular culture, entrepreneurship) cannot be tested, but the core claim holds.
Test: Rank test: US nominal GDP rank 1 in 2010 (about $15.0 trillion vs China $6.1 trillion).
Grader B
The US stayed the world's largest economy through 2009 and after. In 2009 US GDP was $14.48 trillion, versus $5.34 trillion for Japan and $5.19 trillion for China. The 'due to' causal clause cannot be tested. The leadership claim holds.
Test: Not an adoption claim. Test: largest national economy by GDP in 2009. US: $14.48T. Next: Japan $5.34T, China $5.19T (World Bank, current US$).
China has also emerged as a powerful economic player.
Grader A
China became a major economic power by 2009-2010. Its nominal GDP passed Japan's in 2010 and it became the second-largest economy.
Test: Rank test: China nominal GDP rank 2 in 2010 ($6.1 trillion, ahead of Japan).
Grader B
China's GDP rose from $1.10 trillion in 1999 to $5.19 trillion in 2009 and $6.19 trillion in 2010. In 2010 it passed Japan to become the world's second-largest economy. China was clearly a powerful economic player by 2009.
Test: Not an adoption claim. Test: China's rank by GDP. 2009: $5.19T, third (just behind Japan's $5.34T). 2010: $6.19T, second.
At least half of all transactions are conducted on-line.
Grader A
Online commerce was far below half of transactions. US e-commerce was 3.7% of retail sales in 2009, and still in single digits in 2014.
Test: Numeric: claimed at least 50% of transactions online; actual US retail e-commerce share 3.7% in 2009 (Census). Far outside the 25% band.
Grader B
Census E-Stats covers manufacturing, wholesale, retail and selected services. In 2010, e-commerce was 46.4% of manufacturing shipments, 24.6% of merchant wholesale sales, 4.4% of retail sales (4.0% in 2009) and 2.3% of selected service revenues. Weighted across these sectors, e-commerce was about 16% of value. Consumer-facing transactions were far below half. The 50% threshold was not reached by 2009 or 2011.
Test: Numeric test: predicted at least 50% of transactions online. Actual (US, 2009-2010): about 16% of covered economic activity by value; retail 4.0% (2009). 16% is far outside the 25% band of 50%.
Intelligent assistants which combine continuous speech recognition, natural language understanding, problem solving, and animated personalities routinely assist with finding information, answering questions and conducting transactions. Intelligent assistants have become a primary interface for interacting with information-based services, with a wide range of choices available. A recent poll ...
Grader A
No intelligent assistants with speech, language understanding and animated personalities were a primary interface in 2009. Siri shipped only in October 2011 and received mixed reviews. Voice assistants stayed a secondary interface through 2014; search boxes and graphical UIs remained primary.
Test: Share of information-service interactions done through conversational assistants: no assistant existed on phones before late 2010; well under 5% of interactions by 2014.
Grader B
No mainstream speech-driven intelligent assistant existed in 2009. Apple launched Siri as a beta on the iPhone 4S in October 2011, and Google Now followed in 2012. These assistants had no animated personalities. They did not become 'a primary interface' for information services by 2014: the web browser and apps stayed primary.
Test: Mainstream test: 20% of users use a speech-and-language assistant as a primary way to reach information services. 2009: effectively 0%. First mass-market assistant: Siri beta, October 2011. No evidence of primary-interface status by 2014.
Most purchases of books, musical “albums,” videos, games and other forms of software do not involve any physical object, so new business models for distributing these forms of information have emerged.
Grader A
Digital-only purchase was common for single music tracks and apps in 2009, but not for books (e-books about 3% of sales in 2009), videos or most games. E-book share grew to 8.5% by mid-2010 and e-books passed hardcovers only in Q1 2012. Across the listed categories, most purchases went non-physical only around 2012-2014, if at all for books.
Test: Share of purchases with no physical object: e-books 3% of US book sales in 2009, 8.5% mid-2010; iTunes sold 10 billion songs by Feb 2010. Measurable (>5%) but not a majority across categories in 2009.
Grader B
In 2009 most purchases in these categories still involved physical objects. E-books were about 3% of US book sales in 2009 and 8.5% by mid-2010. Music passed the threshold first: digital was 50.3% of US music purchases in 2011. E-books passed hardcover sales in Q1 2012. New distribution models (iTunes, App Store, Netflix streaming) did emerge. 'Most purchases' across books, albums, videos, games and software was not true by 2011. It was reached only category by category, from 2011 onward.
There are many types and gradations of transactions to gain access to these products. You can “buy” a book, musical album, video, etc. which gives you unlimited permanent access.
Grader A
Buying a permanent digital copy of a song, album or video was routine in 2009. iTunes had sold 8 billion songs by July 2009 and was the largest US music vendor.
Test: Routine use in the relevant industry: iTunes, the largest US music retailer from April 2008, sold owned downloads (10 billion songs by Feb 2010).
Grader B
From 2003 the iTunes Store sold music and later video and books for permanent ownership. It was the largest US music vendor by April 2008. 'Buy' access to digital media was routine before 2009.
Test: Mainstream test: 20% of media buyers purchase permanent digital copies. Digital was 50.3% of US music purchases by 2011; iTunes had sold tens of billions of songs by 2014 (35 billion by May 2014).
Alternatively, you can rent access to read, view, or listen once, or a few times. Or you can rent access by the minute.
Grader A
Digital rental was a routine offer by 2009. iTunes started movie rentals in January 2008, and cable video-on-demand and online stores also sold time-limited rentals. Rental 'by the minute' did not become a standard model, but the main claim (rent to view once or a few times) held.
Test: Routine use by at least 20% of the relevant industry: the largest US digital media store (iTunes) and major VOD services offered 24-48 hour rentals from 2008.
Grader B
Digital rental models were in routine use by 2009. Amazon Unbox (September 2006) offered video rentals and purchases. Netflix launched streaming in January 2007; its early plans metered streaming hours, which is a form of time-based access. iTunes offered movie rentals from January 2008. Per-minute rental in the strict sense stayed rare. The rent-once and rent-for-a-period models were mainstream by 2011.
Test: Mainstream test: 20% of target users rent digital media for limited access. Netflix streaming (from 2007) and iTunes, Amazon and cable video-on-demand rentals were in wide use by 2009-2011. Strict per-minute pricing: rare.
Access may be limited to one person or to a group of persons (for example, a family or a company). Alternatively, access may be limited to a particular computer, or to any computer accessed by a particular person or by a set of persons.
Grader A
DRM that bound purchases to a person's account and a set of authorized computers was standard in 2009. Apple FairPlay limited playback to five authorized computers per account. The dominant stores all used such schemes.
Test: Routine use by at least 20% of the industry: FairPlay (iTunes, the largest US music vendor) limited access to five authorized computers per account.
Grader B
This describes DRM access controls, and they were routine by 2009. Apple's FairPlay (2003-2009) limited purchased music to five authorized computers tied to a user's account, plus CD-burn limits. Access was tied both to a person (the account) and to specific machines. Group and family access schemes also existed, although Apple's own Family Sharing came later (2014).
Test: Mainstream test: 20% of digital media sales carry person- or device-bound access rules. The iTunes Store, the largest US music vendor by April 2008, sold all its music under FairPlay limits (five authorized computers) until January 2009.
There is a strong trend towards the geographic separation of work groups. People are successfully working together despite living and working in different places.
Grader A
Distributed work was already common. In 2009, 24% of employed Americans did some or all of their work at home on days they worked. Distributed teams across sites and countries were routine in large firms and in open-source software.
Test: Share of US employed persons who worked at home on days worked: 24% in 2009 (BLS ATUS), above 20%.
Grader B
By 2010, 24% of employed Americans did some or all of their work at home on days they worked. Distributed teams working through email, conference calls and web conferencing were common in knowledge industries by 2009. 'Strong trend' and 'successfully working together' are loose, but the direction and scale match.
Test: Mainstream test: 20% of workers work routinely away from co-workers. 2010 figure: 24% of employed persons did some or all work at home on days worked (BLS ATUS).
The average household has more than a hundred computers, most of which are embedded in appliances and built-in communication systems.
Grader A
Most household computers were indeed embedded processors in appliances. But the best-known estimate put the average US household at about 60 embedded microprocessors in 2000. No reliable source shows an average above 100 in 2009. The figure is probably too high, but the direction is right.
Test: Numeric: claimed >100 computers per household; available estimate about 60 embedded processors per US household (2000). The 2009 figure is not measured; likely below 100 and outside the 10% band.
Grader B
Kurzweil counts computers 'embedded in appliances', so the embedded-processor reading dominates. Computerworld estimated about 60 embedded microprocessors in the average US household in August 2000. Microcontroller use grew over the decade, and a 2009 car alone held dozens of controllers. No retrieved source gives a 2009 figure. The count plausibly reached about 75-125 but cannot be confirmed as over 100. If 'computers' means general-purpose or networked devices, the count was under 10 and the verdict is miss.
Test: Numeric test: predicted over 100 computers per household. Nearest measured figure: about 60 embedded microprocessors per average US household (2000). The 2009 value is unmeasured; extrapolation puts it near the 25% band, not clearly within 10%.
Household robots have emerged, but are not yet fully accepted.
Grader A
Household robots emerged but stayed niche, which matches the claim. iRobot's Roomba launched in 2002 and sold a million units by 2004; robot vacuums were a real but minority product in 2009.
Test: Existence plus non-mainstream: Roomba on sale from 2002, 1 million units by 2004; household penetration well below 20% in 2009.
Grader B
iRobot launched the Roomba in September 2002 and sold 1 million units by 2004. Home robots (vacuum, pool and gutter robots) were on the market by 2009 but were owned by a small share of households. That matches 'emerged, but not yet fully accepted'.
Test: Test: household robots on the market but below mainstream (20% of households). Roomba: 1 million units by 2004; about 30 million iRobot home robots by 2020. 2009 US household penetration was well below 20%.
Intelligent roads are in use, primarily for long-distance travel. Once your car’s computer guidance system locks onto the control sensors on one of these highways, you can sit back and relax. Local roads, though, are still predominantly conventional.
Grader A
No highway with guidance sensors that cars lock onto was in service in 2009 or by 2014. The US Automated Highway System demo ran once in 1997 on I-15 and the program ended. Investment moved to sensors in the vehicle, not in the road.
Test: Share of long-distance highways with vehicle-guidance infrastructure in use: zero by 2014.
Grader B
No highway had guidance sensors that let cars drive themselves, in 2009 or by 2014. The US National Automated Highway System Consortium held a technically successful demo on I-15 near San Diego in 1997. Investment then moved to self-contained autonomous vehicles instead of instrumented roads. SARTRE (from 2009) was a research test, not a deployment.
Test: Mainstream test: 20% of long-distance highway mileage or travel on automated highways. Actual: 0% in 2009 and 2014; the only activity was demos and research projects.
A company west of the Mississippi and north of the Mason-Dixon Line has surpassed a trillion dollars in market capitalization.
Grader A
No company reached $1 trillion market capitalization by 2014. Microsoft (Redmond, the intended company) was worth $269 billion at end of 2009 and first passed $1 trillion in 2019. Apple and Amazon passed $1 trillion only in 2018.
Test: Numeric: claimed >$1,000bn; Microsoft actual $268.55bn at end 2009 (73% short).
Grader B
No company reached a $1 trillion market capitalization by 2014. The first were Apple (August 2, 2018), Amazon (September 4, 2018) and Microsoft (April 25, 2019). The Seattle-area firms (Amazon, Microsoft) match the geographic hint. The milestone came about a decade late.
Test: Numeric/threshold test: any company at or above $1T market cap by 2009 (grace to 2014). Actual: none until August 2018.
Litigation, of which there has been a great deal, has placed some constraints on the widespread distribution of personal data.
Grader A
Privacy litigation did constrain data sharing. The Lane v. Facebook class action over Beacon led Facebook to shut the service in September 2009 and pay $9.5 million. Other suits and settlements had similar effects. Regulation also played a large part, but the claim of 'some constraints' from litigation holds.
Test: Existence and effect: class action directly ended a major data-sharing product (Facebook Beacon, shut 21 Sep 2009).
Grader B
Privacy litigation did place some limits on distribution of personal data by 2009. The Lane v. Facebook class action over Facebook Beacon, which shared users' purchases, ended in a $9.5 million settlement in 2009, and Facebook shut Beacon down on September 21, 2009. The claim is modest ('some constraints'). Most constraints came from statute, but litigation clearly contributed.
Test: Not an adoption claim. Test: privacy litigation measurably constrains data sharing by 2009. Example: Beacon shut down (Sept 2009) with a $9.5M class-action settlement.
Government agencies, however, continue to have the right to gain access to people’s files…
Grader A
Government access to personal records was retained and expanded. The USA PATRIOT Act (2001) let the FBI obtain records under Section 215 and via National Security Letters without probable cause.
Test: Legal-status test: statutory government access to records in force in 2009 (Patriot Act Section 215, NSLs).
Grader B
US government agencies kept and widened their access to personal communications and records. The NSA warrantless surveillance program was revealed in December 2005. The FISA Amendments Act of 2008 relaxed warrant requirements and gave this access a statutory base.
Test: Not an adoption claim. Test: agencies retain legal access to personal data in 2009. The FISA Amendments Act of 2008 confirmed and widened that access.
There is continuing concern with an underclass that the skill ladder has left far behind. The size of the underclass appears to be stable, however.
Grader A
The underclass was not stable. The US poverty rate rose from 11.3% in 2000 to 14.3% in 2009 and 15.9% in 2011. Concern about a left-behind group did continue, but the stability claim failed.
Test: Numeric stability: poverty rate 11.9% (1999) vs 14.3% (2009), a 20% relative increase; SNAP participation reached an all-time high of 47.6 million in 2013.
Grader B
The US poor population did not stay stable. The official poverty rate rose from 11.9% (32.8 million people) in 1999 to 14.3% (43.6 million) in 2009 and 15.1% in 2010. It was still 14.8% in 2014. That is an increase of about a third in the number of people in poverty. The 'continuing concern' part is true; the stability claim is false.
Test: Numeric test: size of the underclass stable from 1999 to 2009. US official poverty rate: 11.9% (1999) to 14.3% (2009), +20% relative. Number in poverty: 32.8M to 43.6M, +33%. Both exceed the 25% partial band.
Most visual art is the result of collaboration between human artists and their intelligent art software.
Grader A
Most visual art in 2009 was not made in collaboration with intelligent software. Tools like Photoshop are instruments, not creative partners. Autonomous art programs such as Harold Cohen's AARON remained unique examples. Generative AI art arrived only around 2022.
Test: Share of visual art made with AI co-creator software: well under 5% by 2014.
Grader B
In 2009 most visual art was not co-created with intelligent software. Digital tools were used, but they were editing tools (Photoshop, Illustrator), not collaborators that generate art. Generative AI image tools reached the mainstream only in 2022 (DALL-E 2, Midjourney, Stable Diffusion), long after 2014. The later backlash against AI art on DeviantArt dates from 2020 onward.
Test: Test: over 50% of visual art made in collaboration with intelligent generative software. 2009-2014: near 0%; generative tools became mainstream in 2022.
Writers use voice-activated word processing…
Grader A
Writers did not generally use voice dictation in 2009. Dictation software (Dragon NaturallySpeaking) existed and served niche groups such as medical and legal professionals and people with disabilities. Keyboards stayed the normal tool for writing through 2014; phone voice typing was used for short messages, not writing.
Test: Share of writers using voice word processing as a main method: no source shows 5% or more by 2014. No adoption figures exist for Dragon in 2009.
Grader B
Writers in general did not use voice-activated word processing in 2009 or by 2014. Dragon NaturallySpeaking (June 1997) was the main desktop dictation product. It served a niche: medical and legal dictation, and users with accessibility needs. Most writing stayed keyboard-based. Smartphone voice typing grew after 2011 for short messages, not for writers' word processing. No retrieved source gives a usage share. No evidence suggests 5% of writers dictated routinely.
Test: Test: 20% of writers routinely compose by voice. 2009: niche (Dragon users, mainly medical and legal dictation). No measured share found; keyboard input remained standard through 2014.
Style improvement and automatic editing software is widely used to improve the quality of writing.
Grader A
Spelling and grammar checking was universal in word processors, and Word had style options. But dedicated style-improvement and automatic-editing software was not widely used in 2009; Grammarly was founded in 2009 and such tools spread in the 2010s.
Test: Dedicated style or editing software users: under 20% of writers in 2009; built-in checks near 100% of Word users but these are basic checks, not style improvement.
Grader B
Style and grammar checking was built into Microsoft Word and so was available to most writers in 2009. It was basic, and experts judged its advice unreliable. Software that improves style in a way writers widely rely on (Grammarly and similar tools) spread in the mid-2010s. Availability was near universal, but real use for style improvement was limited.
Test: Test: 20% of writers routinely use style-improvement software. Word's style and grammar options were present for nearly all users, but quality was contested. Dedicated style tools reached the mainstream after 2014.
Nonetheless, the core process of creating written language is less affected by intelligent software technologies than the visual and musical arts. However, “cybernetic” authors are emerging.
Grader A
Both parts held. Core writing stayed a human activity with less software influence than music or visual work. Automated authors did emerge: Thomson Reuters automated financial stories from 2006, and Narrative Science, StatSheet and Automated Insights produced machine-written articles around 2010.
Test: Emergence test: machine-written news in commercial use from 2006 (Thomson Reuters) and around 2010 (StatSheet, Narrative Science).
Grader B
Both parts held. Creative writing was far less changed by software than music and visual production were. 'Cybernetic' authors appeared in a narrow form: Narrative Science (founded 2010) and Automated Insights produced automated sports and finance stories. AP began automated earnings stories in 2014, and the LA Times Quakebot published in 2014. Automated authorship was emerging by 2010-2011.
Test: Test: automated authors exist and are emerging (not mainstream) by 2011. Narrative Science was founded in 2010; AP and LA Times automated stories began in 2014.
Users also experience fantasy environments with no counterpart in the physical world.
Grader A
Fantasy environments with no physical counterpart were mass entertainment by 2009. World of Warcraft had over 10 million subscribers in 2008 and 12 million in 2010, and many other online and offline fantasy games existed.
Test: WoW subscribers: 10-12 million (2008-2010), a single title; fantasy game play routine among gamers.
Grader B
Fantasy virtual worlds were a mass entertainment category by 2009. World of Warcraft alone had more than 10 million subscribers in January 2008 and 12 million in October 2010. Many other fantasy MMOs and single-player fantasy games existed.
Test: Mainstream test: 20% of gamers use fantasy virtual environments. WoW: 10M+ subscribers (2008) and 12M (2010). 53% of US adults play video games, and fantasy settings are a major genre.
The security of computation and communication is the primary focus of the U.S. Department of Defense. There is general recognition that the side that can maintain the integrity of its computational resources will dominate the battlefield.
Grader A
Cyber security became a major DoD concern: the Secretary of Defense created US Cyber Command in June 2009. But it was not the primary DoD focus; ground wars in Iraq and Afghanistan dominated budgets and attention, and Cyber Command became a full combatant command only in 2017-2018.
Test: Priority test: dedicated cyber command created June 2009 (sub-unified), one of many priorities; not the primary focus.
Grader B
Cyber security rose sharply at DoD around 2009, but it was not 'the primary focus'. Secretary Gates ordered US Cyber Command on June 23, 2009; it reached full operational capability in October 2010. The 2011 DoD cyber strategy treated cyberspace as an operational domain alongside land, sea and air. In 2009 the primary DoD focus was the ground wars in Iraq and Afghanistan, with about 186,000 troops in country. Cyber became a top-tier priority, one domain among several.
Test: Test: computation and communication security is the primary DoD focus. 2009-2011: new Cyber Command and cyberspace declared an operational domain. Budget and forces remained dominated by conventional ground operations (about 186,300 troops in Iraq and Afghanistan in FY2009).
Humans are generally far removed from the scene of battle.
Grader A
Humans were not far removed from battle in 2009. The US had about 146,000 troops in Iraq at the end of 2008 plus a growing force in Afghanistan, and ground combat and casualties continued. Drone operators far from the front were a small minority of combatants.
Test: Share of combat personnel remote from the battlefield: small; over 100,000 US troops deployed in Iraq alone in 2009.
Grader B
In 2009 the US had about 186,300 troops in country in Iraq and Afghanistan, plus more than 100,000 in regional support. Most combat was infantry counterinsurgency at close range. Remote drone operations grew, but humans were not generally far from the battle.
Test: Test: most combat personnel operate remotely. FY2009: about 186,300 US boots on the ground in Iraq and Afghanistan; remote operators were a small fraction.
Many of these flying weapons are the size of small birds, or smaller.
Grader A
Most US drones in 2009 were Raven scouts: hand-launched and unarmed, larger than a small bird. Armed flying devices the size of small birds did not exist in service. Nano scout drones reached only small numbers of troops after 2012 (grader background knowledge, not retrieved).
Test: Share of flying weapons of small-bird size or smaller: effectively zero in 2009; very few by 2014.
Grader B
No bird-sized or smaller flying weapons were in wide service by 2009 or 2014. Micro air vehicles were fielded for reconnaissance only: the T-Hawk (from 2007) searched for roadside bombs, and the 16-gram Black Hornet Nano reached British troops in Afghanistan in 2012. Neither was a weapon. 'Many' bird-sized flying weapons did not exist.
Test: Test: a large share of flying weapons are bird-sized or smaller. 2009-2014: 0% of armed systems; small UAVs were unarmed reconnaissance tools.
Privacy issues concerning access to these records (as with many other data bases of personal information) have emerged as a major issue.
Grader A
Privacy of electronic medical records was a major issue in 2009. The HITECH Act of 2009 extended HIPAA privacy and security rules to business associates, created breach notification for incidents affecting 500+ people and tiered penalties.
Test: Policy-salience test: federal legislation on health-record privacy enacted in 2009 (HITECH privacy provisions).
Grader B
Health-record privacy was a major policy issue by 2009. The HITECH Act (2009) extended HIPAA privacy and security rules to business associates. It required breach notification to HHS, the media and affected people for breaches of 500 or more records. It set tiered penalties up to $1.5 million a year per provision. This followed the HIPAA Privacy Rule, and the national debate over electronic records tied spending on EHRs to privacy protections.
Test: Not an adoption claim. Test: medical-record privacy is a major legislative issue by 2009. HITECH (2009) added breach notification and stronger HIPAA enforcement.
Doctors routinely train in virtual reality environments, which include a haptic interface. These systems simulate the visual, auditory and tactile experience of medical procedures, including surgery.
Grader A
Medical simulation was embraced by medical schools and residencies, and VR surgical simulators with haptics existed. But in 2009 their use was not routine or standardized; simulation remained un-standardized and much training used mannequins and box trainers, not haptic VR.
Test: Routine haptic-VR training among surgical trainees: measurable minority (simulators in use at teaching hospitals), not standardized; below 20% of training time.
Grader B
Simulation training was established in medical education by 2009. Laparoscopic skills programs such as FLS used box trainers, and high-fidelity models had existed since the 1990s. VR surgical simulators with haptic feedback existed as commercial products and were used in some residency programs. They were not routine for doctors in general. Most simulation used physical trainers and mannequins, not haptic VR. No retrieved source gives an adoption share for VR haptic trainers.
Test: Test: 20% of doctors or training programs routinely use VR simulators with haptics. 2009: simulation training common, but mostly box trainers and mannequins. VR haptic simulator adoption: present in some programs, share unmeasured and below routine.
Simulated patients are available for continuing medical education, for medical students, and for people who just want to play doctor.
Grader A
Simulated patients were available for medical education by 2009. High-fidelity patient simulator mannequins (for example METI's Human Patient Simulator) were used in US medical schools across all four years and in residency. The 'play doctor' consumer part was served by games, but the main claim is availability.
Test: Availability in medical schools: AAMC survey reports simulation taught across all four years at US medical schools.
Grader B
Simulated patients were available for medical and nursing education and continuing education by 2009. Laerdal's SimMan line of high-fidelity patient simulators (SimMan 3G in 2009) was widely used, and nursing research found that simulation can replace up to 50% of clinical hours. Consumer 'play doctor' games (e.g. Trauma Center, 2005) also existed. The claim is about availability, and it holds.
Test: Test: simulated patients available for medical students, CME and the public by 2009. High-fidelity mannequins (SimMan line) were common in medical and nursing schools; consumer medical sim games existed.
There is serious speculation on the potential sentience (i.e., consciousness) of computer-based intelligence.
Grader A
Serious academic work on machine consciousness existed by 2009. The International Journal of Machine Consciousness started in 2009, and researchers such as Aleksander, Haikonen and Franklin (LIDA) published computational models of consciousness.
Test: Existence test: dedicated peer-reviewed journal (2009) and books and models (2003) on machine consciousness.
Grader B
Serious academic work on machine consciousness existed by 2009. The International Journal of Machine Consciousness published its first volume in 2009. Researchers such as Aleksander, Haikonen, Shanahan and Franklin (LIDA) proposed architectures, and Chalmers analysed whether artificial systems could be conscious.
Test: Not an adoption claim. Test: serious (peer-reviewed, institutional) speculation on machine consciousness by 2009. A dedicated journal started in 2009.
Test: Share test: digital music ~40% of US sales in 2009 (>5%), majority 2011; e-books 3% 2009, >20% ~2012; across all media mainstream only 2012-2014.
Test: Industry test: speech IVR used routinely by large call-center operators (>5%, likely >20%); animated-personality LUIs ~0%.
Test: Share test: e-books 3% 2009, 8.5% mid-2010, >20% ~2012 (mainstream in 2012 window).
Test: Test: over 50% of purchases in each category non-physical. 2009 figures: e-books about 3%. Music: under 50% in 2009; 50.3% of purchases in 2011 (Nielsen/Billboard). Books and video: not a majority by 2011.