Software Copyright Checklist: Registration, Deposit, Evidence, and Comparison

By Casey Scott McKay · 2024-08-02

Seventeen phases covering software copyright from the registration program a company should run before any dispute exists to the settlement structures these cases actually reach. The prevention phases build the release-registration habit that determines whether a future case has any economics, with the deposit and redaction choices that decide what a plaintiff can later prove. The investigation phases audit chain of title and run the section 412 analysis, both of which can end a case in three weeks and both of which the other side will find if you do not. The evidence phases sequence access and independent creation ahead of code comparison. The expert phases give a filtration protocol, a report structure, and the shared-errors evidence that persuades. A protective-order phase covers the review terms that quietly decide what your expert can do.

IP and Technology > Information Technology | Checklist | Published 2 August 2024 - Updated 29 October 2025 | Casey Scott McKay - marksy.us

Summary. Seventeen phases covering software copyright from the registration program a company should run before any dispute exists to the settlement structures these cases actually reach. The prevention phases build the release-registration habit that determines whether a future case has any economics, with the deposit and redaction choices that decide what a plaintiff can later prove. The investigation phases audit chain of title and run the section 412 analysis, both of which can end a case in three weeks and both of which the other side will find if you do not. The evidence phases sequence access and independent creation ahead of code comparison. The expert phases give a filtration protocol, a report structure, and the shared-errors evidence that persuades. A protective-order phase covers the review terms that quietly decide what your expert can do.

Keywords: software registration program · identifying portion deposit · trade secret redaction · section 412 timing · chain of title contractors · present assignment · access logs · version control discovery · independent creation record · clean room protocol · abstraction filtration comparison · merger and scenes a faire · shared errors · expert report structure · source code protective order · fair use software · google v oracle scope · dtsa parallel claim · preemption check · design-around settlement

What this checklist is for

This is the working document for software copyright, before and during a dispute. It does not re-teach the framework. If you cannot say in one sentence what Google v. Oracle assumed rather than decided, read Copyright in Code first. The reasoning behind each box is in Proving or Defending a Software Copyright Claim. This document tells you what to do, in order.

Who should use it. In-house counsel at a software company with no registration program; litigators on either side of a code-copying dispute; counsel advising on a reimplementation or a clean room; and anyone conducting IP diligence on a software asset.

What you'll need before you start. The contributor list for every module at issue; contractor and employment agreements; acquisition documents and any recorded assignments; the registration file and deposit copies; the version control repository with full history; access and personnel records; and the open source inventory.

The worked matter. Thornfield Analytics sells a supply-chain forecasting platform. A senior engineer, Ravi Oyelaran, leaves for Meridian Flow. Fourteen months later Meridian ships a product whose forecasting module produces familiar outputs and whose API is close enough that a Thornfield customer's integration script runs against it with two changes. Thornfield has one 2019 registration with a fifty-page deposit; the current version is four major releases newer.

| Phase | What you accomplish | Typical elapsed time | |---|---|---| | 1 | Build the release registration program | ongoing | | 2 | Choose deposit and redaction options | 1 week | | 3 | Maintain chain of title continuously | ongoing | | 4 | Audit chain of title when a dispute arises | 2-5 weeks | | 5 | Run the § 412 analysis | 1-3 weeks | | 6 | Prove or disprove access | 2-5 months | | 7 | Build or attack independent creation | 2-5 months | | 8 | Run or reconstruct the clean-room protocol | 2-8 weeks | | 9 | Negotiate the source code protective order | 3-8 weeks | | 10 | Choose and prepare the expert | 3-6 weeks | | 11 | Run abstraction-filtration-comparison | 3-8 months | | 12 | Find the shared errors | within Phase 11 | | 13 | Brief fair use accurately | 4-8 weeks | | 14 | Coordinate the parallel claims | 3-6 weeks | | 15 | Move on filtration; price the remedies | 3-6 months | | 16 | Settle in the structures these reach | 2-6 months | | 17 | Budget and review | annually |

Phase 1 — Build the release registration program

[ ] Register every major release as it ships, not once at launch.
- Why. 17 U.S.C. § 412 makes statutory damages and attorney's fees unavailable for infringement commenced before registration, unless registration occurred within three months of first publication. This provision determines whether a future case is worth bringing, and the decision is made years earlier by whoever did or did not register.
[ ] Register within three months of first publication wherever possible, which preserves statutory damages for infringement occurring at any time.
[ ] Assign an owner — usually the release manager, with counsel's template — and attach it to the release checklist so it happens automatically.
[ ] Retain the exact deposit copy and the corresponding build for every registration, in a location that survives repository migrations.
[ ] Note that registration must be made, not merely applied for, before suit. 17 U.S.C. § 411(a); Fourth Estate Public Benefit Corp. v. Wall-Street.com, LLC, 586 U.S. 296, 301-11 (2019). Special handling exists and it costs.
[ ] Budget: a few thousand dollars a year. It is the highest-return item in software IP management.
- See. Registering a Copyright.

Phase 2 — Choose deposit and redaction options

[ ] Understand what may be deposited: identifying portions, commonly the first and last twenty-five pages of source code, with options for redaction of trade secret material. 37 C.F.R. § 202.20.
[ ] Choose the portions deliberately. The default first-and-last-25 selection frequently captures headers and boilerplate rather than the distinctive material a future claim would rest on.
[ ] Redact trade secret material, using the permitted options.
- Trap. A copyright deposit is a public record. An unredacted deposit publishes what it contains and can undermine a trade secret assertion over the same material. Check what was actually filed years ago, because the answer may already be public. See Trade Secrets and the DTSA.
[ ] Record, for each registration: version, date, what was deposited, what was redacted, and where the full build is archived.

Phase 3 — Maintain chain of title continuously

[ ] Written present assignment from every contractor, subcontractor, offshore firm, and intern.
- Why. Software is not among the nine categories eligible for commissioned work-for-hire treatment under 17 U.S.C. § 101, so an unassigned contractor owns their contribution.
[ ] Employment agreements with present assignment language — "hereby assigns," not "agrees to assign."
[ ] Founder contributions predating incorporation, assigned in writing.
[ ] For every acquisition, a copyright assignment with a schedule of works, not merely an "all intellectual property" clause, and consider recording under 17 U.S.C. § 205.
[ ] An open source inventory with license obligations tracked. See Running an Open Source Compliance Program; Copyleft and Consequences.
[ ] See Copyright Ownership and Chain of Title Checklist.

Phase 4 — Audit chain of title when a dispute arises

[ ] Identify every contributor to the accused modules specifically, not to the product generally.
[ ] Verify a written assignment for each contractor, and run the Community for Creative Non-Violence v. Reid, 490 U.S. 730, 751-53 (1989) analysis for anyone whose employment status is ambiguous.
[ ] Check acquisition documents for a schedule and a recorded assignment.
[ ] Check for joint authorship exposure — a joint author may license non-exclusively without the other's consent.
[ ] Check open source components and whether conditions were satisfied.
[ ] Do this in the first three weeks.
- Trap. A chain-of-title defect ends the case regardless of similarity, it is the cheapest defense available, and the defendant will find it in discovery if the plaintiff does not find it first.

Phase 5 — Run the § 412 analysis

[ ] Identify the registration date, the first publication date, and the date infringement commenced.
[ ] Determine which of three outcomes applies.
- Timely registration — statutory damages and fees available; the case has economics.
- Late registration, infringement after it — statutory damages and fees available for post-registration infringement; establish the commencement date carefully.
- Late registration, infringement before it — actual damages and profits only, which in software means a difficult apportionment case and usually a commercial resolution.
[ ] Identify which version was registered, and establish that the registered work contains the material allegedly copied.
- Trap. One 2019 registration and a program four releases newer is a proof problem. Retrieve the deposit copy and the corresponding build. If they are gone, register the current version now, which secures statutory damages for continuing infringement going forward.
[ ] Report the § 412 conclusion to the client before any strategy decision.

Thornfield, Phase 5. Registration in 2019 long precedes the alleged 2024 infringement, so statutory damages are available. That single procedural fact is worth more than most of the merits arguments and should shape the settlement posture from week one.

Phase 6 — Prove or disprove access

[ ] Collect: repository access logs; VPN and system logs; badge records; email and document-sharing histories; NDAs and evaluation agreements; delivery records under any license or partnership; personnel files; and exit forensics where the case follows a departure. See Trade Secret Protection and Departure Checklist.
[ ] Identify every person who had access to the plaintiff's code and contributed to the accused modules, by name and by commit.
[ ] Consider whether striking similarity supports an inference of access on its own — see Phase 12.
[ ] Defendant: establish that no one with access contributed to the accused modules; where such a person exists, establish precisely what they touched.
[ ] Do not rely on the inverse ratio rule. The Ninth Circuit abandoned it, Skidmore v. Led Zeppelin, 952 F.3d 1051, 1066-69 (9th Cir. 2020), and its status elsewhere is uneven.

Phase 7 — Build or attack independent creation

[ ] Produce or obtain the complete version control history for the accused modules: commits, authors, timestamps, branches, merges, and messages.
[ ] Collect design documents, architecture decisions, and specifications, with dates.
[ ] Collect issue tracker and project management history showing the development arc.
[ ] Collect code review records.
[ ] Identify the dead ends — abandoned approaches, refactors, reverted commits.
- Why. A genuine development history is messy, and the mess is the evidence. A module that appears substantially complete in a few large commits, with no design documents and no dead ends, is very hard to explain.
[ ] Plaintiff: ask for the repository, not a production of files; ask for full history including deleted branches; ask about repository migrations, which frequently flatten history.
[ ] Sequence this before the expert work.
- Trap. Spending $250,000 on code comparison before reading the commit log is the most common sequencing error in this practice.

Phase 8 — Run or reconstruct the clean-room protocol

[ ] Prospectively: a specification team examines the original and never writes the new code; it produces a functional specification containing no expression from the original — no code, comments, variable names, or copied structural diagrams; counsel reviews and removes anything expressive; a development team with no exposure implements from the specification alone; and everything is logged, including every question passed between teams.
[ ] Retrospectively: reconstruct whatever record exists and be candid about gaps. A partial protocol honestly described beats a claimed protocol that deposition testimony contradicts.
[ ] Find the contaminated engineer before the other side does — the one person who saw the original and later joined the development team.
[ ] Note the boundaries: reverse engineering to reach unprotected elements is generally fair use where copying is a necessary step, Sega Enterprises Ltd. v. Accolade, Inc., 977 F.2d 1510, 1517-28 (9th Cir. 1992); Sony Computer Entertainment, Inc. v. Connectix Corp., 203 F.3d 596, 602-08 (9th Cir. 2000) — but a license prohibition, 17 U.S.C. § 1201 anti-circumvention, and trade secret exposure are separate questions.

Phase 9 — Negotiate the source code protective order

[ ] Where review occurs, and on what machine configuration.
[ ] Which comparison tools are permitted — ask the expert before agreeing.
[ ] Printing limits, in pages and aggregate, and handling of printed excerpts.
[ ] Who may access: outside counsel, experts, any in-house personnel.
[ ] Expert disclosure and approval in advance, with an objection window.
[ ] Designation tiers and a process for challenging over-designation.
[ ] How code excerpts appear in briefs and reports, and the sealing procedure.
[ ] Retention and destruction at the end of the case.
- Trap. A printing limit set without knowing the codebase size, or a tool list excluding the software your expert uses, is a concession you will not recover.

Phase 10 — Choose and prepare the expert

[ ] Retain the right profile, or both: a software engineer with domain experience who can speak to what the platform, framework, and industry dictate; and a computer science academic experienced in program comparison methodology.
[ ] Pull every published decision discussing the candidate and read for a signature methodology about to reappear in your case.
[ ] Request the four-year testimony list before signing. Fed. R. Civ. P. 26(a)(2)(B)(v).
[ ] Retain in a consulting capacity first and convert in writing. Fed. R. Civ. P. 26(b)(4)(D); draft reports protected under Rule 26(b)(4)(B).
[ ] Write the retention letter with: scope and capacity; expert control of methodology; a bare-transmittal rule for facts with a single numbered "Assumptions Provided by Counsel" exhibit; native-format preservation; and non-contingent compensation.
[ ] Structure the report to mirror Phase 11: qualifications and materials; abstraction levels and why; filtration element by element with each filtered element's source identified; what survives, quantified; the comparison with shared errors first; conclusions within the analysis's limits; and limitations.
- Why limitations matter. An expert who states them is credible on cross. One who overclaims is dismantled, and the dismantling takes the rest of the report with it.

Phase 11 — Run abstraction-filtration-comparison

[ ] Abstraction. Identify levels: object code; source code; modules and functions; data structures and interfaces; module hierarchy and control flow; architecture; ultimate function. State which levels the claim addresses.
[ ] Filtration. Remove, at each level:
- Ideas, procedures, processes, systems, methods of operation. 17 U.S.C. § 102(b).
- Elements dictated by efficiency — merger.
- Elements dictated by external factors — hardware, compatibility, industry demands, platform design guidelines, accepted programming practice. Computer Associates International, Inc. v. Altai, Inc., 982 F.2d 693, 709-10 (2d Cir. 1992).
- Public domain material: open source, standard algorithms, published specifications, prior art.
- Facts and data. Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340, 344-51 (1991); 17 U.S.C. § 103.
- Short phrases, names, titles. 37 C.F.R. § 202.1.
[ ] Comparison. Compare only what survives, and quantify it.
[ ] Document each filtered element's source in an exhibit — the platform guideline, the published specification, the open source project, the standard algorithm.
- Why. A filtration analysis supported by sources is the single most effective document either side produces.
[ ] Trap. A similarity analysis performed without filtration will be dismantled by showing most identified similarities are unprotectable. Filter first, always.

Phase 12 — Find the shared errors

[ ] Direct the expert to look, before anything else, for: the same bug; the same misspelled comment; the same redundant variable; the same dead code; the same idiosyncratic formatting or non-standard indentation; the same obsolete workaround for a problem that no longer exists.
- Why. Functional similarity has innocent explanations. A replicated typo does not. This is the evidence that persuades a court and it is frequently found in an afternoon.
[ ] Keep the two questions separate in the report: copying in fact — did the defendant copy — and unlawful appropriation — did it take enough protectable expression.
[ ] Where shared errors are strong, consider whether striking similarity supports the copying inference without independent proof of access.

Phase 13 — Brief fair use accurately

[ ] Factor one: does the use serve a different purpose? Interoperability, compatibility, security research, and analysis fare well; a competing product with the same purpose does not. Note Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith, 598 U.S. 508, 525-33 (2023) narrowed transformative use and post-dates Google.
[ ] Factor two: favors defendants more than in any other category, because functional works sit at the edge of copyright's concern.
[ ] Factor three: how much protectable expression relative to the purpose; intermediate copying during reverse engineering has been permitted.
[ ] Factor four: the market including licensing markets, and the public consequences of enforcement. Google LLC v. Oracle America, Inc., 593 U.S. 1, 33-36 (2021).
[ ] Use Google precisely. It assumed copyrightability and held fair use on a record about declaring code reimplemented so programmers could use acquired skills in a new environment. It did not hold that APIs are uncopyrightable, that reimplementation is always fair, or anything weakening protection for implementing code.
- Trap. Citing it broadly invites a correction that costs credibility on everything else in the brief. See Running a Fair Use Analysis; Fair Use Risk Assessment Checklist.

Phase 14 — Coordinate the parallel claims

[ ] Assess whether the trade secret claim should lead. 18 U.S.C. § 1836: no filtration, no registration prerequisite, injunctive relief, unjust enrichment, exemplary damages, and fees — the last two conditioned on the notice at 18 U.S.C. § 1833(b), which is in an agreement drafted years ago and should be checked in week one.
[ ] Draft the trade secret identification with the technical team. Courts commonly require reasonable particularity before discovery of the defendant's code, and too broad invites a motion while too narrow forecloses claims.
[ ] Add contract claims — license terms, confidentiality, invention assignment, reverse-engineering prohibitions. Not subject to filtration, registration, or expert comparison, and frequently the cleanest claim in the complaint.
[ ] Consider 17 U.S.C. § 1201 where a protection measure was defeated, noting the interoperability exemption at § 1201(f).
[ ] Check preemption. 17 U.S.C. § 301 preempts state claims equivalent to copyright. Trade secret generally survives on the extra element; a state claim resting on nothing but copying does not, and pleading one invites a motion.
[ ] Keep contemporaneous time entries that distinguish the claims, because a fee application will need to allocate.
[ ] See Where an Employee Can Go; Drafting and Enforcing Restrictive Covenants.

Phase 15 — Move on filtration; price the remedies

[ ] Defendant's strongest motion: after filtration, no protectable expression was copied — or fair use resolves it as a matter of law, as in Google. Support it with the Phase 11 exhibits.
[ ] Plaintiff's motion: narrower — ownership, registration validity, and access.
[ ] Price the remedies.
- Actual damages and profits, 17 U.S.C. § 504(b), with a severe apportionment fight in software.
- Statutory damages, § 504(c), per work — one registered program is one award.
- Fees, 17 U.S.C. § 505, guided by Kirtsaeng v. John Wiley & Sons, Inc., 579 U.S. 197, 202-09 (2016). In a $1.5 million case, often worth more than damages.
- Injunctive relief, 17 U.S.C. § 502, subject to eBay Inc. v. MercExchange, L.L.C., 547 U.S. 388 (2006). The remedy that matters where the objective is to stop a product.
- Limitations: three years, 17 U.S.C. § 507(b), no laches bar within the period, Petrella v. Metro-Goldwyn-Mayer, Inc., 572 U.S. 663 (2014).

Phase 16 — Settle in the structures these reach

[ ] A license, retroactive and prospective, royalty or lump sum.
[ ] A design-around commitment on an agreed schedule with a verification mechanism.
[ ] A supervised clean-room reimplementation, protocol agreed in the settlement, with a neutral verifying compliance.
[ ] A purchase of the module, the product line, or the company.
[ ] A cross-license, where both parties have exposure — common in software.
[ ] A personnel undertaking segregating the departing engineer from the module for a period, with certification.
[ ] A stipulated injunction for breach, so a second round is fast.
[ ] Mutual releases and a surviving confidentiality clause.
[ ] Price the alternative honestly: $700k-$3M per side through trial; $400k-$1.2M through summary judgment; $250k-$800k in expert work alone.
- The rule. Full litigation is justified where there is literal copying, clean title, timely registration, and a directly competing product. Where any of the four is missing, start the settlement conversation early.

Phase 12A — The adjacent categories clients call software

Three things arrive framed as software claims and behave differently. Run them through their own analysis rather than through Phase 11.

[ ] Data and databases. Facts are not copyrightable; a compilation is protected only in original selection, coordination, or arrangement. Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340, 344-51 (1991); 17 U.S.C. § 103.
- Trap. An exhaustive listing in an obvious order has no protectable selection. The United States has no sui generis database right, so a dataset is protected by contract, access control, and trade secret. See Who Owns the Data?; Running or Defending a Data Scraping Program.
[ ] User interfaces and screen displays. Protectable as audiovisual or pictorial works to the extent they contain original expression, and heavily filtered for elements dictated by function, convention, and platform design guidelines. Modern interface claims are usually stronger as trade dress or design patent claims. See Layering Protection for a Product Design; Protecting Trade Dress.
[ ] Generated code. Human authorship remains a requirement, and material lacking it has been refused registration. For a shipping product the live issues are provenance, license contamination from training data, and whether the company can represent ownership to a customer or an acquirer — addressed by policy and contract, not litigation. See Who Owns What the Machine Made?; Deploying Generative AI Without Losing Your IP; Generative AI IP Compliance Checklist.
[ ] A note for diligence. All three arrive in an acquisition as assets the seller represents it owns. Where the asset is a dataset, an interface, or generated code, the representation needs qualification, and a buyer that accepts an unqualified copyright representation over any of them has accepted a risk the law does not support. See IP Due Diligence Toolkit; Trademark Due Diligence in Mergers and Acquisitions.

Phase 16A — The pre-suit demand, and answering one

The first letter sets the shape of everything that follows, and in software it carries specific risks in both directions.

Sending one.

[ ] Complete Phases 4 and 5 first. A demand letter sent before the chain-of-title audit and the § 412 analysis is a letter you may have to retreat from.
[ ] Identify the work by registration number and version, not by product name.
[ ] Describe the basis in general terms without disclosing your filtration analysis, which is work product you will produce on your own schedule.
[ ] Ask for something specific and achievable: a preservation commitment, a meeting, or an explanation of the development history.
[ ] Send a preservation demand covering the repository with full history, design documents, issue tracker, code review records, and personnel assignments. This is the single most valuable paragraph in the letter, because it converts later gaps into spoliation.
[ ] Consider the declaratory-judgment consequence. A letter asserting infringement with a deadline invites a declaratory action in the recipient's home district, and the forum you lose may matter more than the leverage you gain. See Suing First: Declaratory Judgment Actions in Trademark Disputes; Declaratory Judgment Checklist.

Answering one.

[ ] Issue a litigation hold immediately, covering everything in the preservation list above whether or not it was demanded.
[ ] Do not clean up the repository. Deleting branches, squashing history, or rewriting commits after receiving a demand is the most damaging thing a defendant can do, and it is done by well-meaning engineers who think they are tidying.
[ ] Identify every person with any exposure to the plaintiff's code and instruct them in writing not to discuss the matter or alter anything.
[ ] Run Phases 4 and 5 against the plaintiff's position — its chain of title and its registration are as vulnerable as yours.
[ ] Assess the § 412 answer early. If statutory damages and fees are unavailable to the plaintiff, the economics of the dispute change entirely and your settlement posture should reflect it.
[ ] Respond substantively rather than with a holding letter, and consider whether the facts support a declaratory action in your own forum.

Phase 17 — Budget and review

| Task | Elapsed | Cost | |---|---|---| | Release registration program | ongoing | $3k-$12k/yr | | Chain of title audit | 2-5 weeks | $20k-$70k | | Registration and § 412 analysis | 1-3 weeks | $8k-$25k | | Access and independent creation discovery | 2-5 months | $80k-$250k | | Protective order negotiation | 3-8 weeks | $20k-$60k | | Expert filtration and comparison, both sides | 3-8 months | $250k-$800k | | Summary judgment | 3-6 months | $120k-$400k | | Through trial | 18-36 months | $700k-$3M per side | | Negotiated license or design-around | 2-6 months | $60k-$250k |

[ ] Present the table with the cheap and dispositive rows first and the expensive contingent rows second. It produces better decisions and it is the honest description of how these cases go.
[ ] Diarize an annual chain-of-title and open source review, and confirm the registration program is running.

Key Authorities at a Glance

| Authority | What it provides | Phase | |---|---|---| | 17 U.S.C. § 102(b) | Ideas and methods of operation unprotected | 11 | | 17 U.S.C. § 101 | Program definition; work-for-hire categories | 3, 4 | | 17 U.S.C. § 103 | Compilations | 11 | | 17 U.S.C. § 107 | Fair use | 13 | | 17 U.S.C. § 205 | Recordation of transfers | 3 | | 17 U.S.C. § 301 | Preemption | 14 | | 17 U.S.C. § 411(a) | Registration prerequisite | 1 | | 17 U.S.C. § 412 | Timely registration for statutory damages and fees | 1, 5 | | 17 U.S.C. § 502 | Injunctions | 15 | | 17 U.S.C. § 504 | Damages and statutory damages | 15 | | 17 U.S.C. § 505 | Fees | 15 | | 17 U.S.C. § 507(b) | Limitations | 15 | | 17 U.S.C. § 1201 | Anti-circumvention; interoperability exemption | 8, 14 | | 37 C.F.R. § 202.20 | Deposit and redaction | 2 | | 37 C.F.R. § 202.1 | Short phrases and names | 11 | | Fed. R. Civ. P. 26 | Expert disclosure and consulting protection | 10 | | Computer Associates Int'l, Inc. v. Altai, Inc., 982 F.2d 693 (2d Cir. 1992) | Abstraction-filtration-comparison | 11 | | Google LLC v. Oracle America, Inc., 593 U.S. 1 (2021) | Fair use on declaring code | 13 | | Andy Warhol Found. v. Goldsmith, 598 U.S. 508 (2023) | Transformative use narrowed | 13 | | Sega Enters. Ltd. v. Accolade, Inc., 977 F.2d 1510 (9th Cir. 1992) | Intermediate copying | 8 | | Sony Computer Entm't, Inc. v. Connectix Corp., 203 F.3d 596 (9th Cir. 2000) | Same, emulation | 8 | | Feist Publ'ns, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340 (1991) | Originality; facts unprotected | 11 | | Community for Creative Non-Violence v. Reid, 490 U.S. 730 (1989) | Employee status | 4 | | Fourth Estate Pub. Benefit Corp. v. Wall-Street.com, LLC, 586 U.S. 296 (2019) | Registration must be made | 1 | | Skidmore v. Led Zeppelin, 952 F.3d 1051 (9th Cir. 2020) | Inverse ratio rule abandoned | 6 | | Kirtsaeng v. John Wiley & Sons, Inc., 579 U.S. 197 (2016) | Fee standard | 15 | | eBay Inc. v. MercExchange, L.L.C., 547 U.S. 388 (2006) | Injunction standard | 15 | | Petrella v. Metro-Goldwyn-Mayer, Inc., 572 U.S. 663 (2014) | No laches bar within the period | 15 | | 18 U.S.C. § 1836 | DTSA | 14 | | 18 U.S.C. § 1833(b) | Notice conditioning DTSA fees | 14 |

The five things people get wrong

Registering once, at launch. § 412 decides whether a case has economics, and a single stale registration frequently means it does not.

Comparing before filtering. A similarity report produced without filtration is dismantled by showing most of the similarities are unprotectable — and the dismantling discredits the parts that were good.

Engaging the expert before reading the commit log. Access and independent creation resolve most of these disputes on ordinary records, at a fraction of the cost.

Citing Google v. Oracle broadly. It assumed copyrightability and decided fair use on declaring code, on a specific record, before Warhol tightened the transformative inquiry.

Leading with copyright against a departing engineer. The DTSA claim usually should lead, and copyright should ride alongside.