Kids Can't Wait — the Apple IIe into 9,250 schools, and the tax code as a distribution channel

proposals.md filed this under idea, at line 712, inside the ## idea block that runs from line 195 to line 721. The filing is right and the id is wrong. The id names a machine. The machine is the least interesting thing in the file.

The Apple IIe was announced in January 1983 at $1,395 and it was a cost-reduction of a four-year-old design: the chip count fell from about 120 to 31, lowercase arrived, 80 columns and 64 KB became standard. Nothing became possible that had not been possible on an Apple II Plus. It is a fine machine and it stayed on sale until 15 November 1993, nearly eleven years, the longest-lived computer Apple ever made — and that fact is a consequence of this entry rather than a part of it. What is worth citing here is the programme the machine was the payload for, and the programme is Kids Can't Wait.

The positive case for idea is the standard house one and it lands. build.py defines idea as "a framing later work is built out of rather than argued about," and the test set by shannon-chess-1950 and reused by dijkstra-1959, printing-press-1440 and ibm-pc-1981 is that there is a mechanism in it you could implement. The mechanism here has two parts and both are artifacts:

1. Get the subsidy written. Find an existing tax preference that applies to somebody else's customers, have your own tax department draft the statute that extends it to yours, and find a member to introduce it. 2. Give the current model away into that class, inside the window the statute opens, and book the difference between your cost and the retail price.

That is implementable, it was implemented in California in 1983 for a net cost of about a million dollars against twenty-one million of retail, and it is being implemented again in 2026 by four companies at once — with the subsidy now paid by the vendor rather than the taxpayer, which section 2 argues is the only part that has changed.

Why not moment. The house definition, fixed by eliza-1966 when it argued its own way out of the kind and reused since by clippy-1996, siri-2011, speak-and-spell-1978 and lego-mindstorms-1998, is a date on which something visibly happened in public. There are dates here — a House vote on 22 September 1982, a machine on January 1983, an announcement in May 1983 — and, exactly as in ibm-pc-1981, the dates are bookkeeping. A moment's evidential content is its date. Nothing became true on any of these three that was not true the week before; what makes the entry citable is a procedure, and the procedure has a five-year history before 1983 (section 1) and a forty-three-year sequel after it (section 2).

Why not interpretation. These are the events. The reading of them — that a generation's first contact with a computer was a policy outcome and not a market one — is the proposal's own claim, and section 3 grades it rather than adopting it.

Why not prediction. The file contains three graded claims, but they belong to Jobs, to the statute and to the proposal, not to the object. If the canon ever files Jobs's 1982 forecast on its own, it should file it with its own dates.

---

1. What it is

The provision Jobs found

Since well before 1982 the Internal Revenue Code had carried an enhanced deduction for corporate gifts of scientific property — §170(e)(4), under which a company donating research equipment to a university could deduct not merely its cost but its cost basis plus half the appreciation, capped at twice cost basis. For a manufacturer whose retail price is several times its build cost, that is a subsidy of real size. It applied to universities, and it required the equipment be used for research.

Jobs's idea was to strip the research requirement and extend the class to elementary and secondary schools. In his own account, thirteen years later: Apple wanted to put a computer in every American school, there were about 100,000 of them, and the company could not afford to simply do it. Under the provision it could: "donate a hundred thousand computers… one to each school in America and it would cost our company ten million dollars."

The bill, the vote, and the Senate

Apple's director of taxes, Michael Rashkin, helped draft it. Rep. Fortney "Pete" Stark (D-CA) introduced it on 23 February 1982 as the Computer Equipment Contribution Act of 1982, H.R. 5573; Sen. John Danforth (R-MO) carried the Senate version. The bill amended the code to allow the charitable deduction for computers given to qualified educational organisations including museums and libraries, and — a clause worth carrying to section 3 — required that the equipment be used directly by the students and that the contributions be nondiscriminatory as to geographic areas or economic status.

Jobs refused to hire lobbyists and did it himself: "I actually walked the halls of Congress for about two weeks, which was the most incredible thing," meeting, by his count, two-thirds of the House and over half the Senate. The House passed it 323–62 on 22 September 1982. Jobs called that the largest favourable majority of any tax bill in the history of the country. The Senate never took it up, and it died with the 97th Congress.

Jobs blamed Bob Dole, then chairman of Senate Finance, for the rest of his life. A caution about the retelling, recorded because it is the kind of thing this canon exists to catch: the 1995 oral history was not read directly here — the Smithsonian transcript returned HTTP 403 — and the summary that was read renders Jobs as blaming "Speaker Bob Dole" during "Carter's lame-duck session." Both are wrong on the record: the Speaker in 1982 was Tip O'Neill, Dole chaired Finance, and the president was Reagan. Whether the errors are Jobs's at thirteen years' distance or the summariser's could not be determined here, and the entry does not attribute them to him.

The rival bill, which is the most transferable object in the file

When the Apple bill was reintroduced in 1983 it acquired a competitor. On 5 April 1983, Jim Wright (D-TX) — House Majority Leader, and Speaker only from 1987 — introduced the Computer Contribution and Teacher Training Act of 1983, H.R. 2417, known to everyone as the Tandy bill. It did the same thing as Apple's, with conditions attached:

• educational software worth at least 10% of the machine's value must come with the donation;
• the dealer must supply the software and the teacher training "in the immediate vicinity of the participating teacher's school";
• the donated machine must have two disk drives storing 180 KB each.

Radio Shack had a nationwide dealer network. The Apple II stored 140 KB per drive. Neither clause names a company; both describe one, and between them they made a TRS-80 Model 4 eligible and an Apple II not. The specification was the lobbying. With four or more competing bills on the table, none of them cleared committee — not Wright's and not Apple's reintroduction. Nothing passed at the federal level, ever.

California, which passed a better version

Apple had already hedged. In September 1982 — Ken Uston's contemporaneous account gives 29 September — Governor Jerry Brown signed Assemblyman Charles Imbrecht's bill, AB 3194, giving a 25% tax credit against California corporate income tax on the fair market value of computer equipment donated to schools. It ran from 1 January 1983 to 30 June 1984.

Note what the state gave that the Congress had been asked for: Jobs asked Washington for a deduction and Sacramento handed him a credit, which is worth several times more per dollar because it comes off the tax owed rather than off the income.

The giveaway

Kids Can't Wait was announced in May 1983. Jobs put Stephen Scheier on it full-time with an ad hoc team of four. The offer was one package to each of roughly 9,250 eligible California elementary and secondary schools — schools with fewer than 100 students were excluded — consisting of:

> a 64K Apple IIe computer, a display monitor, one floppy disk drive, and a copy > of Apple Logo

plus software coupons from more than twenty-five educational publishers. Retail value $2,364 a package; over $21 million of product if every school took one. Apple's gross cost: $5.2 million. Projected California credit: $4 million. Net cost: about $1 million — a 95% reduction, on Uston's figures, on the price of introducing your product to a state's entire school population. (The arithmetic does not quite close: 25% of $21 million would be about $5.25 million rather than $4 million. Uston reports both numbers without reconciling them and this entry does not resolve it; treat the $4 million as his projection, not as a settled figure.) By 4 July 1983 UPI was reporting the machines as delivered. Counts vary across sources — 9,250 (Uston, 1983), roughly 9,000 (Watters; the UPI headline), almost 10,000 (Jobs, 1995) — and the spread is reported rather than averaged.

The window is the part to notice. The statute ran from 1 January 1983. The Apple IIe was announced in January 1983. The giveaway was announced in May and delivered by July, and the credit expired in June 1984. A company launched a machine, gave it to a state, and deducted it, inside the window of a statute its own tax director had helped draft. Every step of that was legal and none of it was accidental.

This was not the first run of the procedure. In 1978, through a competitive evaluation, the Minnesota Educational Computing Consortium chose the Apple II over the TRS-80 and bought 500 machines for that state's schools. MECC then built the era's dominant educational software catalogue — The Oregon Trail is the one everybody remembers — and from 1981 sold it outside Minnesota. The software ran on Apple IIs because Minnesota's schools had Apple IIs, and schools elsewhere bought Apple IIs because that is what the software ran on. Jobs took to Congress in 1982 a mechanism his company had already tested on one state five years earlier.

What actually arrived, measured

The proposal's claim is about access, so the entry has to say what access amounted to. Two independent measurements exist for exactly this period, and both were read directly for this file.

**Henry Jay Becker, How Schools Use Microcomputers: Summary of the 1983 National Survey (Center for Social Organization of Schools, Johns Hopkins University, March 1985) surveyed a scientific sample of 2,209 US elementary and secondary schools during the 1982–83 school year** — the year of the IIe and of Kids Can't Wait — with forms returned by principals and by each school's designated primary computer-using teacher, at 70% response from the 1,580 schools that had micros.

• A majority of micro-owning schools had fewer than five machines; fewer than 10% had as many as fifteen. Writing in March 1985, Becker says micros were by then "present in more than two-thirds of U.S. schools" — that sentence is dated 1985, not 1983, and this file does not use it as a 1983 figure.
• In a typical micro-owning school, about 70 students used a computer in a given week — about one student in seven.
• Median weekly time per elementary student user: 23 minutes. A third of them got 15 minutes or less — about three minutes a day. Only one elementary user in fifty got more than an hour a week. Secondary median: 45 minutes.
• And the finding that decides what "one computer per school" was worth: elementary schools with more micros did not give students more minutes; they gave more students the same minutes. Adding a machine widened exposure and did not deepen it.

The second measurement is the Census Bureau's, from the October 1984 Current Population Survey sponsored by NCES and published by Robert Kominski in March 1988 as Computer Use in the United States: 1984, Current Population Reports P23-155.

• 6,980,000 households — 8.2% — had a computer. 22.9% of households at $50,000 or more; 1.7% of households below $10,000.
• Among children 3 to 17: 15.3% had a computer available at home, and of those 74.2% used it. By income, 3.4% of children in households below $10,000 had a home computer against 37% above $50,000.
• About 28% of the 45.6 million enrolled students used a computer at school.

Put the two together and the proposal's claim gets its strongest support. In October 1984, 8.2% of American households had a computer and 28% of enrolled students used one at school; by the time Becker wrote up his survey in March 1985, schools were past two-thirds. The two figures are eighteen months apart and the gap between them is not close. For most American children in this period the only computer within reach was in a school, and a large number of California schools had theirs because of a state tax credit. On that much, the proposal is right.

The equity finding, which does not go the way the story goes

Both surveys then say something the popular telling of Kids Can't Wait leaves out, and it is the reason this entry is worth writing.

Kominski sets up the hope explicitly:

> One mechanism for correcting differentials in access is the schools. Ideally, > schools--offering equal access to all--may work to compensate for inequities > that arise as a reflection of social background.

And then reports the measurement:

> Perhaps most notable is the relationship of school use with householders > educational attainment and family income. In both cases, school use by > children generally becomes more likely with increases in either householder's > educational attainment or family income. Both of these phenomena indicate a > possible indirect effect of family socioeconomic status on computer use > through the quality and equipping of schools.

The channel that was supposed to equalise tracked income too. School use by Black children was 15.9%, the lowest of any race group, against 28.9% for non-Hispanic children and 29.0% for boys; public-school students used a computer at school at 27.4% against private-school students at 32.0%.

Becker's Table 5 shows the same thing at higher resolution, and with a twist worth quoting exactly because it is not the twist people expect. Comparing elementary schools of similar gross economic standing:

| | highest-SES white | middle-SES white | lowest-SES white | pred. minority | |---|---|---|---|---| | median students per micro | 155:1 | 183:1 | 192:1 | 215:1 | | median hours/week the school's micros run | 10 | 13 | 10 | 20 | | median minutes/week per user | 24 | 22 | 35 | 18 | | school has a "Computerist" teacher | 46% | 33% | 35% | 14% | | intensive use for drill-and-practice | 13% | 18% | 9% | 33% | | intensive use for programming | 21% | 17% | 49% | 10% | | intensive use with above-average students | 24% | 30% | 51% | 26% | | intensive use with below-average students | 16% | 12% | 10% | 32% |

The first two rows are the ones most easily misread, so read them together. Predominantly minority elementary schools ran their machines for twice as many hours a week as the poor white schools — 20 against 10 — and each individual user still got half the time, 18 minutes against 35. Those schools were not using their computers less. They were spreading a scarcer resource across more children, which is exactly the trade Becker identified nationally when he found that elementary schools with more machines widen access rather than lengthen turns. The disadvantage is in the denominator, not in the effort.

Becker's own summary: in minority communities the schools that had micros were using them to bring up the achievement of their lower-performing students through drill; in white communities of similar income the schools gave the machines to their better-achieving students, taught them programming, and let them work independently. He is careful, and the care should be carried with the numbers: "it has yet to be shown which approach has more merit."

And across all schools, teachers reported the payoff landing where the machines landed. Asked what there had been "much more" of as a result of having a microcomputer, 24% said academic learning by above-average students; 7% said the same for below-average students and 6% for average ones.

Becker's verdict on the whole procurement, which is the single most transferable sentence in this file:

> The social climate demanding universal ("equitable") access had made it > difficult, if not impossible, to make rational judgments and comprehensive > plans.

The training, which is what the donation actually cost the recipient

InfoWorld reported in September 1983 that the hands-on teacher training attached to the donated machines varied dealer to dealer across California, from an hour's orientation talk to a whole day. One Los Angeles dealer, Take A Byte, near UCLA, gave one hour and twenty minutes to about thirty-five teachers. Tom Lester, who ran K-12 mathematics and computer teacher training in San Juan Unified, expected teachers so trained would simply "give out so many recipes" for Logo, and said the shortage of trained staff was so acute that some trainers "know one day's knowledge more than the people who are taking the training."

The gift was a computer. The cost was a curriculum, a room, a schedule and a trained adult, and none of those were in the box.

What it bought Apple, and how it ended

By 1984 Apple held roughly half the US primary-and-secondary school computer market. By 1987 more than a million Apple IIs were in American schools. The IIe outlived the Macintosh's first decade on the strength of that base, selling until 15 November 1993.

Then it ended, and quickly. In the third quarter of 2014 Chromebook shipments into US K-12 passed iPads for the first time — 715,000 against 702,000. Apple's share of US K-12 mobile device shipments fell from 34% in 2014 to 25% in 2015 to 19% in 2016, while Chrome OS took about 58%. Thirty years of incumbency in a channel the company had effectively invented, undone in about three by a cheaper machine with a better fleet-management story — bought, again, by districts, on procurement terms, for reasons that had very little to do with which machine was better to learn on.

---

2. Why a reading would cite it

Four cases. The first is the entry's job.

(a) Concentration — free access for a class of users is a distribution strategy, and this is its base rate. The 2026 instance is not an analogy, it is the same move with the subsidy moved from the taxpayer to the vendor's balance sheet:

• April 2025 — Google begins bundling Gemini and NotebookLM into existing education cloud contracts; Utah's state board of education has since partnered with Google to deploy Gemini across K-12.
• November 2025 — OpenAI launches ChatGPT for Teachers, free for verified US K-12 staff, initially through June 2027, with the plan page now advertising the offer through June 2028.
• 14 July 2026 — Anthropic launches Claude for Teachers, free for verified US educators, with academic standards from all fifty states built in and a pilot in Detroit Public Schools Community District.
• Microsoft is running teacher training through its Elevate initiative.
• Education Week's 2025 survey found 61% of teachers using AI in some capacity, against 32% the year before.

eMarketer, describing the same set of moves, says the quiet part out loud — the companies are "subsidizing the present—free tools, discounted tiers, teaching guides—to lock in future loyalty," because they want the tool a student uses in the curriculum to follow that student to their first job.

That is Kids Can't Wait, and this entry supplies the questions a reading should ask about it, in order. What is the term of the free offer — California's credit ran eighteen months, and OpenAI's teacher offer has a stated end date that has already moved once; a district that rebuilds its lesson planning around a free tool is a district whose renewal terms are set by someone else in 2028. Who wrote the eligibility rule — see (b). And what is the denominator — see (c).

(b) Policy and regulation — the eligibility criterion is the lobbying. The Tandy bill is the cleanest worked example this canon has of an ostensibly neutral technical specification written to name one product and exclude another: two 180 KB drives, dealer training near the school. Apple's own bill was drafted by Apple's director of taxes. Both are the same move, and neither bill mentions a company. Cite this whenever an AI rule is written in apparently neutral technical terms — a training-compute threshold, an evaluation or red-team requirement, a data-residency rule, an audit standard, a procurement specification — and ask two questions: whose currently shipping product does the spec happen to describe, and who was in the room when it was drafted. LENSES.md asks the policy lens for what governments did. What the US government did in 1982–83 was fail to pass four bills, whereupon one state did the thing instead, which is also a finding.

(c) The public square — arrival is not access, and this entry has the arithmetic. By 1985 the answer to "how many American schools have a computer" was more than two-thirds. The answer to "how much computer does an American child get" was that one student in seven touched one in a given week and the median elementary user got twenty-three minutes — three minutes a day — and that adding machines widened the queue rather than lengthening the turn. Cite against any adoption headline of the form X% of schools / firms / clinics now use AI. The percentage is the easy number and it is almost never the one that matters; ask for the denominator, the minutes, and who inside the institution is actually getting the turn.

(d) Concentration and the public square — the channel does not equalise on its own, and this is the first time anyone checked. Kominski set up schools as the correcting mechanism and then found school use rising with family income and householder education. Becker found 215 students per machine in predominantly minority elementary schools against 155 in the highest-SES white ones, and found the machines used for drill in one and for programming in the other. Cite against the standing claim — made about laptops, about broadband, about coding curricula and now about model access — that giving a capability to schools closes a gap. **It did not in 1984, and the mechanism Kominski named for why is differences between schools rather than within them**, which nothing since has repealed.

The honest limit, stated here rather than buried. This is one company, one state, one eighteen-month tax window, and a product whose marginal cost was dominated by components. Free model access has a marginal cost near zero, which makes the giveaway far cheaper and changes the shape of the dependency it creates: nobody has to ship anything, and the thing given away can be withdrawn or repriced by editing a page. This entry is a base rate for how seeded channels behave, not a forecast, and a reading that runs it as a forecast is doing the thing section 4 warns about.

The disclosure: the model writing this file is made by one of the four companies named in case (a), and Claude for Teachers appears in that list. The convenient errors available here run in both directions — softening the case because it lands on the author's maker, or overstating it to demonstrate independence. The defence is that all four offers are stated the same way, in the same form, with their dates and their terms, and that the questions in case (a) are asked of the whole set and not of any one of them.

---

3. What it got right, and what it got wrong

idea does not require this section. It is here because there are four dated claims on the table and the entry is worth more graded than repeated.

Claim 1 — Jobs's, 1982

Extend §170(e)(4) to schools and Apple can put a computer in every school in America — a hundred thousand machines — for ten million dollars. Due on passage.

Wrong about the vehicle, right about the arithmetic. No federal bill ever passed. The House gave him 323–62 and it died in the Senate, and when he came back in 1983 the field was crowded with rival bills written to rival specifications and nothing survived committee. But where a legislature did give him the mechanism, the numbers came out roughly as he said: $21 million of retail delivered into 9,250 schools for about $1 million net, or a little over $100 a school. A mechanism that depends on a statute is exactly as reliable as the statute, and this one was defeated by a competitor with a better-placed member, not by anything about its merits — which is a general finding and not a sour one.

Claim 2 — the proposal's own, canon/proposals.md, 17 August 2026

> the Apple IIe, and computing arriving in schools. Cite when: the question is > who gets early access to a capability and what that does to who builds the > next one. A generation's first contact with a computer was a policy outcome, > not a market one.

Graded in four pieces.

• "who gets early access to a capability" — right, and the strongest clause in it. That is the question, and 1983 answered it concretely: whoever a vendor's tax department could make eligible for a subsidy, in whichever state passed the subsidy, minus the schools with under a hundred students.
• "A generation's first contact with a computer was a policy outcome, not a market one" — right in letter, and misleading in spirit, and the correction is this entry's contribution. The support is stronger than the proposal knew: 8.2% of American households had a computer in October 1984 and 1.7% of households under $10,000 did, against 28% of enrolled students using one at school in the same month and more than two-thirds of schools owning one by early 1985, so for most children the school was the only computer in reach. But it was not a policy in the ordinary sense. It was a company's bill. The state did not decide which children met which computer; a vendor decided, wrote the eligibility rule, chose the state, chose the machine, chose the software in the box, and the public paid a quarter of the fair market value in forgone revenue. "Policy outcome" invites the reading that a public body deliberated. What actually happened is better described as a private distribution decision executed through a public subsidy, which is a different thing and a more useful precedent.
• "the Apple IIe" — the wrong object. The machine is a cost-reduced Apple II Plus. The entry is Kids Can't Wait, and the id has been kept because ids are stable and the prose can carry the correction.
• "what that does to who builds the next one" — unproven, and this entry declines to supply the anecdote. Becker gives the strongest honest support available: "the major instructional outcome of microcomputers by 1983 was to increase the number of high school students who had some skill in computer programming and an understanding of computer concepts." That is a real finding and it should be cited. It arrives welded to its own qualification, in Becker's next clause: the equipment "came to be used primarily by the more independently-working 'above-average' students, primarily at the higher grade levels." It made programmers, and it made them disproportionately out of children who were already ahead. Beyond that, what exists is self-report — a great many people who write software for a living say they met a computer in a school — and lego-mindstorms-1998 catalogues the identical evidentiary trap at length for FIRST: a decade of follow-up on self-selected volunteers is encouraging and is not a causal design. No study establishing that early school access produced practitioners was found for this entry, and the entry says so rather than manufacturing one.

Claim 3 — the equity claim, made by the statute and by the era

H.R. 5573 required the donations be "nondiscriminatory as to geographic areas or economic status." In April 1983 the National Commission on Excellence in Education published A Nation at Risk and put half a year of computer science among its Five New Basics for high school graduation, alongside English, mathematics, science and social studies. Claim made 1982–83; first measured October 1984.

Wrong on the first measurement, in both surveys, on every cut. School use rose with family income and with householder education. Black children's school use was the lowest of any race group. Predominantly minority elementary schools had the worst student-per-machine ratio and the shortest turns and the least likely chance of a teacher who knew computers, and used what they had for drill while similarly poor white schools taught programming. The non-discrimination clause was in the bill that failed. The credit that passed had no such clause, and the 9,250 packages went to every eligible California school on a rule about enrolment size, which is equal treatment and is not the same as equal effect.

Claim 4 — Apple's implied commercial claim

Seed the schools and hold the market. Due continuously.

Right for thirty years and then wrong all at once. Half the K-12 market by 1984, a million machines by 1987, the donated model still selling in 1993. Then Chromebooks passed iPads in Q3 2014 and Apple was at 19% by 2016 against Chrome OS at 58%. The transferable shape: a seeded channel is durable against better products and is not durable against cheaper ones with lower administrative cost. What displaced Apple in American schools was not a better machine for children. It was a machine that a district IT department could deploy, lock, wipe and reissue at a fraction of the labour. Any 2026 reading watching an AI vendor buy a channel with free access should be watching that variable, not the benchmarks.

---

4. Commonly misused as

build.py reserves this section for limit. It is here because this entry exists to be used as an analogy and the analogy is habitually run the wrong way.

1. "Apple gave computers to California schools." Apple executed a tax position. On the contemporaneous accounting the company's net outlay was about $1 million against $21 million of retail, and the state of California made up much of the difference in forgone revenue. That does not make it a bad programme — the machines were real, they arrived, and a company that had lost the federal vote still delivered on its own state — but the word "gave" imports a moral register that the arithmetic does not support, and it is the register Apple bought. Cite this against any 2026 framing of free access to schools as philanthropy without first asking who is paying and what they are buying.

2. "The Apple II won the classroom because it was the best educational computer." It won a procurement — a MECC evaluation in 1978 — and then the software followed the installed base and the base followed the software. Creative Computing did name the IIe the best educational computer above $1,000, and that judgment is real; it is also a magazine's, arriving after the standard was set. The standard was set by a bid, not by a benchmark. The 2026 form of this error is reading a district's model choice as evidence about model quality when it is usually evidence about contracts, admin tooling and whoever got there free first.

3. "Free access democratises a capability." The first time anyone measured it, in October 1984, the school channel tracked family income. Anyone citing "free for every verified teacher" as an equity measure owes three numbers that 1983 shows are the ones that bind: which schools have the staff and the scheduled time to use it, how long the free term runs, and what happens at renewal to a district that has rebuilt its curriculum on it.

4. "A technology subsidy is neutral because it names no company." Four federal bills, each drafted around a different vendor's product line — 180 KB drives, dealer proximity, software bundled at 10% of value — and not one of them names a firm. The one that passed was a state credit whose window opened the month its principal beneficiary launched a new machine. Neutral language is not neutral policy, and the test is not whether a rule names an incumbent but whether it describes one.

5. "Kids Can't Wait proves early exposure creates a generation of builders." Not established, and the best-documented mechanism runs the other way inside each school. See claim 2 above; the honest version is much smaller and still worth saying — a great many practitioners date their start to a school computer, which is a self-report about a self-selected group.

---

5. Ancestry

descends_from is [sputnik-1957], and it is one edge rather than several because only one candidate survives the test that descent must explain shape.

sputnik-1957 already carries the two things this entry is made of. The first is the institution: NDEA Title VII, in the entry's own words, was "the first federal money for educational technology research" — the precedent for the United States answering a competitiveness scare by re-equipping its classrooms with machines. The second is the mechanism, and it is the reason the arrow is worth drawing. sputnik-1957's load-bearing correction is that the NDEA was not written after Sputnik: the bill and its constituency predated the launch, a Senate committee clerk suggested selling an education bill as a defence measure, and the crisis supplied "a name and a majority" to a policy people had been holding for years. That is H.R. 5573 exactly, with the constituency reduced to a single company: a pre-existing tax preference for university research equipment, extended to a new class, drafted by the beneficiary's own tax director, and sold on a national-competitiveness argument that A Nation at Risk would restate in April 1983. Descent that explains shape is the test, and this passes it.

What is deliberately not claimed:

• Logo, and Papert. Apple Logo was in the donated box, and the pedagogy in that box is Seymour Papert's. lego-mindstorms-1998 has already settled how the canon handles this: the genuine ancestors are Logo (1967) and Mindstorms the book (1980), neither is in canon/, neither is in proposals.md, and they will not be invented into a header here either. If the chain ever writes the book, this entry should acquire the edge then. perceptrons-1969 is Minsky and Papert and is not an ancestor of anything in this file — the same ruling lego-mindstorms-1998 made, for the same reason.
• ibm-pc-1981. A sibling, and the contrast is the useful part. IBM published a complete description and granted nothing, and lost the platform to people who reimplemented it. Apple published nothing, gave away the objects themselves, and captured a channel for thirty years. Publishing an interface disperses a capability; giving away an artifact concentrates a customer base. Those are opposite mechanisms and a reading that reaches for "openness" to describe both has lost the distinction that makes either useful.
• lego-mindstorms-1998. Fifteen years downstream and a sibling in theme — proposals.md pairs the two, and the pairing is right for reading and wrong for the graph. Read together they make one point neither makes alone: the 1983 machine was given to institutions and mostly produced drill and twenty-three minutes; the 1998 machine was sold to households and was reverse-engineered by a stranger in eight weeks. Who receives a capability determines what is done with it more reliably than what the capability is.

Ids named in this file that do not exist in canon/ and are not being invented into the header: macintosh-1984 (proposed at line 717 of proposals.md, unwritten), and MECC, A Nation at Risk, the Tandy bill, One Laptop per Child and the Chromebook, none of which are proposed anywhere and all of which appear here only as prose.

---

6. Sources

Marked by what was actually read, in the house convention.

• **Henry Jay Becker, How Schools Use Microcomputers: Summary of the 1983 National Survey, Center for Social Organization of Schools, The Johns Hopkins University, March 1985 (ERIC ED257448). Read directly, as rendered PDF pages, from files.eric.ed.gov/fulltext/ED257448.pdf. The primary source for the survey design (2,209 schools, 1982–83 school year, 70% teacher response from 1,580 micro-owning schools, 18-page teacher questionnaire); most micro-owning schools having fewer than five machines; the "more than two-thirds of U.S. schools" line, which is Becker writing in March 1985 about the present rather than reporting a 1983 survey result**, and which is used with that caveat attached throughout this file; ~70 student users a week and "about one student in every seven"; Table 4 (elementary median 23 minutes a week, 31% at 15 minutes or less, one in fifty above an hour; secondary median 45 minutes); the finding that elementary schools with more micros extended access rather than time; Table 1 and the drill-versus-programming split by level; Table 5 in full, reproduced above; Table 6 (24% "much more" learning by above-average students against 7% below-average and 6% average); the "yet to be shown which approach has more merit" caution; the "major instructional outcome… by 1983" passage and the "more independently-working 'above-average' students" clause immediately following it; and the interpretive remarks, including the quoted sentence on the social climate demanding universal access. Pages 32–38 were read for the interpretive section; pages 22–31 were not.
• **Robert Kominski, Computer Use in the United States: 1984, US Bureau of the Census, Current Population Reports Series P-23 No. 155, March 1988, from the October 1984 CPS sponsored by the National Center for Education Statistics. Read directly**, as rendered PDF pages, from www2.census.gov/library/publications/1988/demographics/p23-155.pdf (highlights and the sections on access and use by children). Source for 8.2% / 6,980,000 households; 22.9% at $50,000+ and 1.7% below $10,000; 30.2% of children 3–17 using a computer at home or school; 15.3% home availability and 74.2% use where available; 3.4% versus 37% home computers by household income; "about 28 percent of the 45.6 million students enrolled"; school use by Black children at 15.9%, non-Hispanic at 28.9%, boys at 29.0%; Table A's public 27.4% against private 32.0%; and both quoted passages — the schools-as- correcting-mechanism setup and the finding that school use rises with family income and householder education.
• **Ken Uston, "9,250 Apples for the teacher," Creative Computing, Vol. 9 No. 10, October 1983, p. 178. Read via fetch**, from atarimagazines.com/creative/v9n10/178_9250_Apples_for_the_teac.php. The contemporaneous account, and preferred over the retellings wherever they overlap. Source for the May 1983 announcement; ~9,250 eligible schools and the 100-student floor; the package contents quoted verbatim; $2,364 retail and over $21 million total; the $5.2 million gross cost, $4 million projected credit and ~$1 million net; Stephen Scheier and the four-person team; the Imbrecht bill signed 29 September 1982, its 25% credit on fair market value, and its 1 January 1983 – 30 June 1984 term; and the note that the federal bill died in the Senate during the late-1982 lame-duck session.
• Audrey Watters, "How Steve Jobs Brought the Apple II to the Classroom" (hackeducation.com, 25 February 2015). Read via fetch. Source for Michael Rashkin's role in drafting H.R. 5573; the bill's "used directly by the students" and "nondiscriminatory as to geographic areas or economic status" requirements; Jobs's "walked the halls of Congress" quotation and the 323–62 House vote; Dole and the Senate Finance Committee; the 1983 reintroduction stalling against Wright's bill; AB 3194 signed by Jerry Brown in September 1982; the package contents and the $2,364 figure, corroborating Uston; and the InfoWorld September 1983 teacher-training material — the hour-to-a-day range, Take A Byte's 1h20m for ~35 teachers, and Tom Lester's "give out so many recipes" and "one day's knowledge more" quotations, which are InfoWorld's reporting as relayed by Watters and were not read in the original.
• Steve Jobs, oral history interview with Daniel Morrow, Computerworld Honors Program (with the Smithsonian), April 1995. Not read in the original — americanhistory.si.edu/comphist/sj1.html returned HTTP 403. The material used here comes from a fetch of Computerworld's published excerpt and from search-result summaries: the ~100,000 schools and "ten million dollars" arithmetic, the extension of the university research-equipment provision to K-8, Stark in the House and Danforth in the Senate, the refusal to hire lobbyists, "the largest favorable majority of any tax bill in the history," and the ~10,000 California machines. The "Speaker Bob Dole" and "Carter's lame-duck session" errors are in the summary read here, not verified against the transcript, and section 1 declines to attribute them to Jobs.
• H.R. 5573, Computer Equipment Contribution Act of 1982 — introduced by Rep. Fortney "Pete" Stark on 23 February 1982, passed the House 323–62 on 22 September 1982, never taken up by the Senate. Search-result summaries of the congress.gov and GovTrack pages only; both returned HTTP 403 on fetch, as did the Joint Committee on Taxation's Description of Miscellaneous Tax Bills (JCX-28-82), which would be the best primary source for the bill's provisions and revenue estimate and should be tried again by a later session.
• H.R. 2417, Computer Contribution and Teacher Training Act of 1983 — the "Tandy bill," introduced 5 April 1983, and its 10%-software, dealer-proximity and two-180 KB-drive conditions against the Apple II's 140 KB. Search-result summaries of trs-80.org/the-apple-bill-and-the-tandy-bill/ only; two attempts to fetch the page returned an empty body. Confidence here is a notch lower than elsewhere in the file, and for a specific reason worth recording: the same summary calls Jim Wright "Speaker of the House" and a Republican, and in April 1983 he was neither — he was House Majority Leader and a Texas Democrat, and became Speaker in 1987. The specification detail is the most citable thing in section 1 and it rests on the weakest source in this list; a later session should read the bill text.
• §170(e)(4) and the enhanced deduction for scientific property — cost basis plus half the appreciation, capped at twice cost basis. Search-result summaries only, from GiftLaw Pro and Urban Institute secondary material; the statute was not read.
• UPI, "Apple gives 9,000 free computers to schools," 4 July 1983. Not read — upi.com returned HTTP 403. Used only for the fact that deliveries were being reported by early July and for the headline's "9,000," both from the search-result listing.
• Wikipedia, "Apple IIe." Read via fetch. Source for the January 1983 introduction at $1,395, the 15 November 1993 discontinuation and "longest-lived computer in Apple's history," the chip-count reduction from about 120 to 31, the 64 KB and 80-column standard configuration, and Creative Computing's "best educational computer above $1,000."
• Apple's education share, MECC, and the Chromebook displacement. Search-result summaries only. Roughly half the K-12 market in 1984 and over a million Apple IIs in American schools by 1987, from AppleInsider's "Apple and education: Four decades of highs and lows"; MECC's 1978 selection of the Apple II over the TRS-80 and its 500-machine order, and the 1981 national release of its software catalogue, from the Wikipedia MECC article and museum sources; Chromebooks at 715,000 against 702,000 iPads in Q3 2014 and Apple's 34% / 25% / 19% against Chrome OS at 58%, from Futuresource and IDC figures as reported by Digital Trends, EdSurge and EdWeek. None of the underlying market-research reports were read, and AppleInsider disputed the 2014 "overtook" framing at the time; the direction of travel is what should be relied on.
• ***A Nation at Risk: The Imperative for Educational Reform*, National Commission on Excellence in Education, April 1983. Partially read — the copy at e4e.org/sites/default/files/a_nation_at_risk_1983.pdf was rendered and the Findings and Appendix A/B pages were read directly, which is the source for the commission's charter (approved 5 August 1981 by Secretary T.H. Bell) and the eighteen-month schedule. The Five New Basics recommendation, and the half-year of computer science in it, was not located in the pages read** and is reported here from search-result summaries; a reading quoting it should say April 1983 and cite Recommendation A.
• The 2025–26 education offers. Chalkbeat, "Anthropic launches Claude for Teachers as AI companies battle for classrooms," 14 July 2026 — read via fetch — for the launch date, the free-for-verified-US-K-12-educators terms, the fifty-state standards, the Detroit Public Schools Community District pilot, the Utah–Google partnership, and Education Week's 61%-versus-32% teacher figures. eMarketer, "Google, OpenAI, and Anthropic bet on classrooms for future AI sales" — read via fetch — for Google's April 2025 bundling of Gemini and NotebookLM into education cloud contracts, OpenAI's November 2025 ChatGPT for Teachers with free verified US K-12 access and the June 2027 term now advertised through June 2028, and the "subsidizing the present… to lock in future loyalty" characterisation. Microsoft's Elevate initiative is from the search-result summary only.

What could not be established, listed so nobody re-does the work: the text of H.R. 5573 and of H.R. 2417 as introduced, and the JCT's description of either (three separate 403s); the Smithsonian oral-history transcript; the reconciliation of Uston's $4 million projected credit against 25% of a $21 million fair market value; whether Pete Stark was in fact Jobs's own representative, which Watters asserts and which the district boundaries make worth checking; how many of the 9,250 eligible California schools actually claimed a package; and whether any contemporaneous evaluation of Kids Can't Wait's educational effect was ever conducted — none was found, and the two surveys used here measure the era rather than the programme.