RoadRunner Recycling, Inc. v. Recycle Track Systems, Inc. et al

Filing 210

ORDER RE 132 SUMMARY JUDGMENT. Signed by Judge Alsup. (whalc2, COURT STAFF) (Filed on 11/23/2024)

1
2
3
UNITED STATES DISTRICT COURT

4
5

NORTHERN DISTRICT OF CALIFORNIA

6
7
8

ROADRUNNER RECYCLING, INC.,

9

Plaintiff,

10

United States District Court
Northern District of California

11
12
13

No. C 23-04804 WHA

v.
RECYCLE TRACK SYSTEMS, INC. and
RECYCLESMART SOLUTIONS, INC.,

ORDER RE SUMMARY JUDGMENT

Defendants.

14
15

INTRODUCTION

16

In this contract and trade secret case, defendants move for partial summary judgment.

17

Defendants do not challenge the contract claim brought by a former supplier. Instead, they

18

contend plaintiff’s asserted trade secrets respecting waste monitoring were not specified,

19

secret, or valuable — and, as to one, not owned by plaintiff. The motion is GRANTED.

20

STATEMENT

21

Parties contest seven remaining trade secrets, which bear repeating before reviewing the

22

factual and procedural histories: (1) An overall waste monitoring system, (2) a smart camera

23

apparatus for mounting in a dumpster, (3) the optical assembly within that apparatus, (4) a

24

collection of images taken by such cameras of dumpsters and labeled with attributes like

25

dumpster fullness, used for training computer models to identify when new images show the

26

same attributes (training data), (5) a computer model for identifying images that are subpar or

27

unworkable, (6) a computer model for identifying images showing the waste is contaminated,

28

and (7) a computer model for identifying images showing a dumpster is full.

1.

1
2

The following is undisputed or as a reasonable jury could find for plaintiff:

3

In 2013, Compology, Inc. — plaintiff RoadRunner Recycling, Inc.’s predecessor —

4

began building prototypes for an internet-connected “smart camera” apparatus to be mounted

5

in dumpsters (see Dkt. No. 145-24 at 26, 29).
That same year, Compology’s co-founder, Benjamin Chehebar, began applying for

6

United States District Court
Northern District of California

FACTUAL HISTORY.

7

patents for an overall waste monitoring system (see, e.g., U.S. Patent App. Pub. No.

8

2014/0278630 (pub’d Sept. 18, 2014)). One patent publication described gathering data from

9

dumpsters using wirelessly connected cameras or sensors, processing that data to identify the

10

dumpster’s fullness or attributes, and instructing what to do next, like empty it (ibid.).

11

In 2016, Chehebar later showcased one result: The R11 smart camera apparatus. In a

12

YouTube video published online, Chehebar shared with a live audience and the video’s viewer

13

a “teardown” of the R11 camera apparatus, showing its insides and talking about many of its

14

components (Dkt. No. 146-30 ¶¶ 13–15, 25). And, in 2018, Chehebar submitted filings for

15

Compology’s internet-connected cameras to the FCC for publication (see id. ¶¶ 16–17, 25).

16

Chehebar requested that some schematics and diagrams not be published (see Dkt. 34-1 at ECF

17

25). But he did not object to the online publication of photographs showing parts of the

18

camera apparatus’s exterior, interior, circuitry, and other features (see id. at ECF 2 & Exhs.).

19

In 2018, Compology continued filing patent applications related to waste monitoring.

20

One claimed to teach taking images of the inside of a dumpster, labeling them with dumpster

21

or waste attributes (like fullness), and using those labeled images to train a computer model to

22

identify when new images had the same attributes (e.g., U.S. Patent App. Pub. No. 20/0013024

23

(pub’d Jan. 9, 2020); cf. Dkt. No. 133-9 Exh. 1).* Such methods became generally known by

24

publication or otherwise before defendants used them (see infra).

25
26
27
28

*

Other patents acknowledged in the record and pertaining to various parts of the overall waste
monitoring system are: U.S. Patent App. No. 62/778,775 (filed Dec. 19, 2018); U.S. Patent No.
10,943,356 (pub’d June 18, 2020); U.S. Patent No. 11,122,388 (pub’d Dec. 24, 2020); U.S. Patent
App. Pub. No. 2021/0158308 (pub’d May 27, 2021); U.S. Patent App. Pub. No. 2021/0158097
(pub’d May 27, 2021); U.S. Patent No. 11,172,325 (pub’d Nov. 9, 2021); and U.S. Patent App.
Pub. No. 2023/0196307 (pub’d June 22, 2023).
2

1
2

No. 145-33 at 54). It sold waste-monitoring services that deployed its cameras into its

3

customers’ dumpsters, analyzed the images with its computer models, and provided an online

4

portal and APIs for customers to glean insights about the dumpsters — and even images with

5

metadata (Dkt. No. 46-4 at 3–4). A contract limited customers’ use of the system.

6

United States District Court
Northern District of California

By this point, Compology had financial backing and a working business model (cf. Dkt.

Also by this point, all parties agree much of the technology in the case was no longer

7

secret: As to the physical apparatus, all agree that between the YouTube video, the FCC

8

filings, and other disclosures, all parts of Compology’s camera apparatuses were made

9

public — except for twelve parts (Dkt. No. 146-30 ¶¶ 13–15). Defendants argue that even

10

those twelve were disclosed or readily known (Reply 6–7). As to the digital tools, parties no

11

longer dispute that the methods used for developing the computer models based on images had

12

become known through the patent-related publications or otherwise (Opp. 21; Reply 7–9).

13

Then came the purported misappropriations: As to the physical apparatus, in 2019, one

14

of Compology’s contracting customers, defendant RecycleSmart Solutions, Inc., disassembled

15

Compology’s R12 camera, or so a jury could find. An email from RecycleSmart to a hardware

16

designer requested a part number and attached four photographs like those published by the

17

FCC (compare Dkt. No. 147-7 (email), with Dkt. No. 133-10 (FCC); Dkt. No. 158-3 at 366–

18

67, 439–40). RecycleSmart began developing its own camera, called the Pello. As to the

19

digital tools, between 2020 and 2021, RecycleSmart downloaded large volumes of labeled

20

images with Compology’s knowledge; the images were of RecycleSmart’s dumpsters and

21

labeled by Compology’s software (see Dkt. No. 146-28 (“Hoarty Op.”) ¶ 97; Dkt. No. 133-2 at

22

39). Parties dispute contract language establishing ownership and use rights in the labeled

23

images (Opp. 24; Reply 11–14). A jury could find RecycleSmart used them to train its model.

24

In just a few years, RecycleSmart accomplished what had taken Compology ten (cf. Dkt.

25

No. 145-33 at 54). In 2022, plaintiff RoadRunner Recycling, Inc. acquired Compology. And,

26

in 2023, defendant RecycleTrack Systems, Inc. acquired defendant RecycleSmart. These

27

corporate changes are not at issue.

28
3

1

RELEVANT PROCEDURAL HISTORY.

2

In August 2023, RoadRunner filed contract and trade secret claims in the County of San

3

Francisco (Dkt. No. 1-1). The case was removed, the complaint amended (Dkt. No. 15), and

4

the case reassigned to the undersigned judge (Dkt. No. 19).

5

The Court soon granted RecycleSmart’s motion to dismiss the trade secret claims (Dkt.

6

No. 43). RoadRunner had not identified “trade secrets.” For starters, it had defined its “trade

7

secrets” as “confidential information that includes trade secrets” (id. at 4). Even as the district

8

court worked to construe bounds RoadRunner had not, it found no secrets:

9
10
11
United States District Court
Northern District of California

2.

12
13

?

The alleged smart camera apparatus was publicly disclosed by
sources like those above (ibid.);

?

The alleged “AI system” was not even pleaded to have any
“specific, non-public aspects” (ibid.); and,

?

The “machine-learning model and preexisting image
database” failed because these were defined for the first time
as “cleanup work in the opposition brief” (id. at 5).

14

In forecasting any motion for leave to amend, the order underscored that the third category

15

likely “conflate[d] access to interfaces that allowed RecycleSmart to ingest data [from] the

16

Compology system,” such as ones that allowed RecycleSmart to retrieve labeled images, “with

17

access to the Compology system’s backend source code,” such as its computer model (id. at 6).

18

RoadRunner moved for leave to amend and filed an identification of its purported secrets

19

(Dkt. Nos. 45–47). Among the fourteen combination or individual trade secret claims were:

20

(1) “Overall Waste Metering System,” a combination now broader than the “AI
system” earlier rejected (Dkt. No. 46-4 (“Identification”) 4–5);

21
22

(2) “Smart Camera Apparatus,” a combination rearticulating what was rejected
(id. at 5–7);

23

(3) “Optical Assembly and Design,” arising within the smart camera (id. at 7–9);

24

(4) “Training Data,” a claim to collections of labeled images used for training
computer models to recognize dumpster and waste attributes (id. at 11–15);

25
26
27

(5)–(7) claims to computer models trained by such data and used for “Subpar
Image Detection,” “Container Fullness Analysis,” and “Contamination Analysis”
— new articulations of the rejected machine-learning claims (id. at 15–18).

28
4

1

“Out of an abundance of caution,” the motion was “GRANTED, subject to paring down these

2

amended allegations during discovery and subsequent motion practice with the benefit of a

3

better record” (Dkt. No. 71 (“Prior Order”) at 2).

United States District Court
Northern District of California

4

Relatedly, a sealing order in this case warned RoadRunner to narrow even its assertions

5

of confidential information (Dkt. No. 93). RoadRunner sought sprawling redactions of

6

marketing puffery, quotations from legal treatises, and more (see id. at 3–5). Even so, the

7

sealing order provided RoadRunner another opportunity to request redactions to avoid deciding

8

the substantive issues in a sealing motion (id. at 6). Notably, parties’ exchanges were protected

9

by a stipulated order that barred in-house counsel from accessing an opposing party’s materials

10

without its approval or a court order (see Dkt. No. 68 §§ 7.3–7.4; Dkt. No. 70).

11

RoadRunner never submitted an amended trade secret identification. Instead, a

12

RoadRunner expert, Leo Hoarty, submitted reports opining on a subset of the identified trade

13

secrets — the seven above — which parties debate in this motion. RoadRunner’s trade secret

14

identification and Expert Hoarty’s identifications have areas of overlap and conflict. As does

15

RoadRunner’s opposition briefing, below.

16
17

This is the first substantive ruling in this case since the motion for leave to amend was
granted “subject to paring down” the trade secret claims (Prior Order 2).

18

3.

19

With “the benefit of a better record,” defendants now move for summary judgment on the

20

trade secret claims (Dkt. Nos. 132–33). RoadRunner opposes (Dkt. No. 145–47). Defendants

21

reply (Dkt. Nos. 156, 158–59; see also Dkt. No. 157 (request for notice)). These trade secret

22

claims are asserted under the California Uniform Trade Secrets Act (“CUTSA”) and the

23

federal Defense of Trade Secrets Act (“DTSA”), and substantially the same standards apply to

24

both. InteliClear, LLC v. ETC Glob. Holdings, Inc., 978 F.3d 653, 657–58 (9th Cir. 2020); see

25

also Cal. Civ. Code § 3426.1(b); 18 U.S.C. § 1836(b)(1).

THE INSTANT MOTION.

26

Defendants do not challenge RoadRunner’s breach-of-contract claim, though they contest

27

the contract terms purporting to define the ownership and use of the labeled images (Reply 13).

28

This order must construe any material facts that would determine these contract issues, like any
5

1

other facts, in favor of the non-moving party. See, e.g., Villiarimo v. Aloha Island Air, Inc.,

2

281 F.3d 1054, 1061 (9th Cir. 2002). Parties submitted supplementary briefing on contract

3

terms that relate to a trade secret (Dkt. No. 182 (“Supp. Br.”); Dkt. No. 205 (“Supp. Opp.”)).

United States District Court
Northern District of California

4

This order follows full briefing and argument.

5

ANALYSIS

6

At trial, plaintiff RoadRunner ultimately must show (1) it possessed a trade secret that

7

(2) defendant misappropriated, (3) causing plaintiff damage. InteliClear, 978 F.3d at 657–58.

8

For this motion, defendants focus on dispatching the first prong. To meet even that prong at

9

trial — possessing a trade secret — RoadRunner must (i) identify information (ii) it owned that

10

was (iii) not readily known and (iv) valuable as a result, and for which (v) it took reasonable

11

steps to keep secret. See ibid.; see also Diodes, Inc. v. Franzen, 260 Cal. App. 2d 244, 253

12

(1968). If no reasonable jury could find one (or more) of these elements, the claim fails.

13

Nissan Fire & Marine Ins. Co. v. Fritz Cos., 210 F.3d 1099, 1106 (9th Cir. 2000).

14

Nearly every claim here fails on the first sub-prong: On the eve of trial, RoadRunner still

15

has not specified what it even claims are its trade secrets. No defendant could mount defenses,

16

no reasonable jury could enter findings, and no court could tailor injunctive relief respecting

17

purported trade secrets that plaintiff has not even clearly pointed out. See Imax Corp. v.

18

Cinema Techs., Inc., 152 F.3d 1161, 1167 (9th Cir. 1998); Adv. Modular Sputtering, Inc. v.

19

Superior Ct., 132 Cal. App. 4th 826, 833–36 (2005). This order takes each topic area in turn:

20

1.

21

Defendants first challenge the purported “Overall Waste Metering System” trade secret

THE OVERALL SYSTEM.

22

(see Identification 4–5; Hoarty Op. ¶¶ 30–35). In its January 2024 disclosure, RoadRunner

23

sought to claim an open-ended system that “include[s] the overall know-how and design of its

24

waste metering system including” a list of every other asserted secret in this case

25

(Identification 4–5). “The combination of these hardware, firmware, and software elements as

26

implemented in an waste metering system [sic] is itself a trade secret” (ibid.). In August 2024,

27

after seven months of discovery, RoadRunner’s expert repeated those passages (compare ibid.,

28

with Hoarty Op. ¶¶ 31–32). In its opposition here, RoadRunner emphasizes the state of its best
6

United States District Court
Northern District of California

1

evidence by quoting one of its engineers: “I think the overall waste metering system [trade

2

secret] encompasses many parts.” (Opp. 17 (quoting Dkt. No. 133-8 at 101)).

3

No reasonable jury could conclude the claimed trade secret was misappropriated (or not)

4

because not even plaintiff could say what the purported trade secret is. Imax, 152 F.3d at 1167.

5

This is not a trivial problem. Recall that a combination trade secret is drawn not to the

6

components, because they can be claimed directly, but to a specific assemblage or tailoring of

7

components (even non-secret ones). E.g., Metallurgical Indus. Inc. v. Fourtek, Inc., 790 F.2d

8

1195, 1202 (5th Cir. 1986). Here, no precise combination was claimed, so there is no way for

9

a jury to determine whether this “overall” tailoring was the same “overall” tailoring from

10

Compology’s patent publications (supra). Likewise, there is no way for a jury to evaluate

11

Expert Hoarty’s statement that he is not “aware” of others using this “overall” system (¶ 35).

12

As no reasonable jury could find plaintiff met its burden to prove a secret plaintiff failed

13

even to identify, summary judgment as to the “Overall Waste Metering System” is GRANTED.

14

2.

15

Defendants next challenge the smart camera apparatus and its optical assembly within.

16

THE PHYSICAL APPARATUS.

A.

SMART CAMERA APPARATUS.

17

In January 2024, plaintiff purported to identify a “Smart Camera Apparatus” trade secret

18

(Identification 5–11). The open-ended claim “include[d] a smart camera, including a [camera]

19

with wide field-of-view camera lens, battery pack, high-durability injection-molded parts,

20

including . . . wireless communication systems, . . . GPS, . . . and a camera flash” (id. at 7). In

21

August 2024, rather than focus this claim, plaintiff’s expert kaleidoscopically expanded it from

22

covering one generic apparatus (ibid.) to covering individually or together the R11, R12, and

23

R13 camera apparatuses (Hoarty Op. ¶¶ 36–62). These have many but not all the same parts,

24

as is undisputed. Now, facing summary judgment, RoadRunner contends the secret is

25

somewhere “defined in the schematics, system block diagrams, and specifications of

26

Compology’s R12 and R13 camera systems,” that is, somewhere in 17 exhibits (see Opp. 7).

27
28
7

1

The record provides no basis for a reasonable jury to decide that this purported trade

2

secret was misappropriated (or not) because the purported trade secret is not identified in a

3

tractable way, despite the Court’s repeated instruction and seven months’ discovery.

4

This conclusion is not changed by plaintiff’s late-in-the-game admissions and arguments

5

as to secrecy. Parties now agree that the YouTube video, FCC filings, and patent-related

6

publications (supra) together disclose all the smart camera apparatus’s elements but for twelve

7

(Opp. 11–12, 21–22). Yet plaintiff could still claim a secret combination of even known

8

elements. InteliClear, 978 F.3d at 659. Rather, the flaw here is more fundamental: Plaintiff

9

never limited itself to claiming any particular combination, these twelve or otherwise.

10

As a result, summary judgment as to the “Smart Camera Apparatus” is GRANTED.

United States District Court
Northern District of California

11

B.

OPTICAL ASSEMBLY.

12

In January 2024, plaintiff purported to identify an “Optical Assembly and Design” trade

13

secret (Identification 7–8). It did so by explaining its trial-and-error efforts to determine the

14

best lens and assembly for its application, without specifying which of these features or

15

configurations (or all of them, identified one-by-one) was its asserted “Optical Assembly and

16

Design” trade secret (ibid.). In August 2024, plaintiff’s expert expanded on what was claimed.

17

He ballooned the purported trade secret to include multiple optical assemblies having different

18

specifications, as is undisputed (Hoarty Op. ¶¶ 63–75; see, e.g., Dkt. No. 138-8 at 137; Br. 13–

19

15; Opp. 19–20). In its opposition, plaintiff again seeks to save its trade secret identification

20

by pointing to the entirety of the specifications for the R12 and R13 cameras (Opp. 19).

21

As no trade secret was reasonably identified, no reasonable jury could find it secret and

22

misappropriated; summary judgment as to the “Optical Assembly and Design” is GRANTED.

23

3.

24

Defendant next challenges the purported trade secrets in its computer models used to

25
26
27
28

THE DATA AND IMAGES.

label images — as well as in the images labeled by plaintiff’s models.
A.

MACHINE-LEARNING MODELS.

In January 2024, plaintiff purported to identify at least three trade secrets in “software
algorithms and processes” (Identification 15–18). Each claimed “a machine learning model,”
8

United States District Court
Northern District of California

1

that is, “a separate, specialized model to identify image[s]” containing the attributes desired.

2

Those attribute-oriented models included ones for “Subpar Image[s]” (id. at 15–16),

3

“Container Fullness” (id. at 16–17), and “Contamination” (id. at 17–18). Then, in August

4

2024, plaintiff’s expert reiterated only the three just named (Hoarty Op. ¶¶ 93–98, 99–104,

5

105–111). He explained they “consist[ed] of a [machine-learning] model . . . and a time-

6

efficient user interface for workers to record labels” (id. ¶ 106 (reciting Identification 16)).

7

RoadRunner’s computer modeling claims proved troublesome for RoadRunner. Recall

8

that its patent publications shared computer-modeling techniques for waste monitoring (supra).

9

That put the “secrecy” of these trade secrets in grave doubt. Recall also that the Court’s order

10

dismissing them on the first go-round suggested that RoadRunner may have conflated access to

11

computer-model outputs with access to the computer models themselves (supra). Defendants’

12

access to the computer models has never been clear, putting “misappropriation” in doubt, too.

13

Whether for those reasons or others, plaintiff’s opposition to summary judgment

14

performs late-stage surgery on the computer-modeling claims, extracting all discussion of the

15

computer modeling and treating only the output from these models: “images plus metadata”

16

(see Opp. 9–10). But another identified trade secret already claimed “images plus metadata”

17

(see id. at 8–9; infra). Either the asserted trade secrets are now equivalent, in which case this

18

claim can be let go. Or, the trade secrets are not equivalent, in which case the difference is too

19

unclear for a reasonable jury to be able to conclude this secret was misappropriated.

20

Bottom line: Plaintiff abandoned any pretense of identifying sufficiently stable trade

21

secrets as to these machine-learning models; and, with none clearly pointed out, no jury could

22

reasonably conclude they were in fact secret and misappropriated. As a result, the motion for

23

summary judgment as to the “Sub-Par Image Detection,” “Contamination Detection/Content

24

Identification Service,” and the “Container Fullness Analysis” trade secrets is GRANTED.

25
26

B.

TRAINING DATA.
(i)

Identification?

27

Finally, plaintiff identified a purported “Training Data” trade secret (Identification 11–

28

15). In August 2024, Expert Hoarty again articulated the trade secret (Hoarty Op. ¶¶ 76–92).
9

United States District Court
Northern District of California

1

Although his was not identical (cf. Br. 16), the identification was sufficiently stable “to give

2

both the court and the defendant reasonable notice of the issues which must be met at the time

3

of trial.” Diodes, 260 Cal. App. 2d at 253.

4

In briefing and oral argument, defendant ultimately acknowledged what plaintiff agreed

5

was the purported trade secret at issue: “[T]he images and data captured by the Compology

6

sensors installed in Defendants’ containers” (Opp. 20–21 (quoting Br. 23)). Such a collection,

7

plaintiff said, is akin to a large, annotated customer list (Tr. 12–14). Recall that already-

8

labeled, known images (in the millions) are used to train machine-learning models, and that a

9

fully trained model can then newly label unknown images with the same attributes (supra).

10

Allegedly, Compology took pictures of RecycleSmart’s dumpsters, Compology’s model

11

labeled them as part of its service, and Compology permitted RecycleSmart to download

12

them — and RecycleSmart later used them to train its own model. Defendants say: So what?

13
14

(ii)

Other Possession Criteria? Or Misappropriation?

Defendants argue plaintiff’s claim fails as a matter of law because plaintiff did not meet

15

other criteria to “possess” the trade secret (Br. 5–6, 22–23) and because defendants did not

16

“misappropriate” it (cf. ibid.). Why not? Defendants owned “all rights” to the labeled images

17

(Br. 5–6, 22–23; Reply 10–14). Parties agree that the answer to who owned the labeled images

18

and who could use them was buried in the operative contract, which they agree is their 2020

19

agreement as modified in 2021 (Supp. Br.; Supp. Opp).

20

Under the agreement as modified, Compology agreed to provide waste monitoring as a

21

subscription service to RecycleSmart (see Dkt. Nos. 145-25 & 147-5 (“2020 Terms”) §§ 1.17,

22

2.1.1). Compology installed cameras in RecycleSmart’s dumpsters (see id. § 1.8, 1.17, 3.1).

23

And, Compology used its software to analyze the images taken, then reported insights to

24

RecycleSmart — and provided images labeled with metadata (see id. §§ 1.16, 1.17). The

25

contract set basic definitions, usage limitations, and ownership rights (see id. §§ 1, 2, 5), all

26

interpreted under California law (id. § 14.9). The 2021 modifications did not alter the 2020

27

provisions cited in this order except where stated here (see Dkt. Nos. 156-2 & 159-3 (“2021

28

Modifications”)). So, who owned the labeled images? In relevant part:
10

1

Compology provided its “Software” to RecycleSmart:

2

1.16. “Software” means any software program or application,
script, algorithm, database, application programming interface
(API) or other processor instructions of any type, solely in object
code format, made available by Compology to Customer on a
software-as-a-service basis, and utilized by Compology for the
purpose of (a) generating, accessing and processing Customer
Data, Device Data and Derivative Data, (b) making such data
available to Customer and (c) interfacing or interacting with the
Equipment. Software shall be deemed to include all
enhancements, improvements, corrections, updates and upgrade
releases made to the foregoing by Compology from time-to-time.

3
4
5
6
7
8

That Software processed “Device Data”:

9

1.9. “Device Data” means any data collected by Compology
through the operation of a Device [where “‘Device’ means the
cameras, GPS trackers,” and so on], including, but not limited to,
data relating to GPS location, container movement, and images.

10

United States District Court
Northern District of California

11
12

By processing Device Data, the Software created “Derivative Data”:
1.7. “Derivative Data” means any information or data resulting
from the manipulation, processing or analysis of Device Data
based upon the operation of the Software, which processing may
involve associated databases, algorithms, external data,
calculations and other processes, methods or tools.

13
14
15
16

And, the customer owned “all rights” to Device Data and Derivative Data:
5.2. Data [Ownership]. Customer owns all rights and title to any
Device Data and Derivative Data. Customer hereby grants to
Compology a limited, perpetual, irrevocable, non-exclusive
license to copy, collect, process, store, generate, and display the
Device Data and Derivative Data to the extent necessary to provide
the System and to monitor and improve the performance of the
System. Additionally, Customer acknowledges and agrees that
during and after the Term of these Terms: (i) Compology and
Compology Affiliates may retain and use Device Data and
Derivative Data for purposes of providing the Software and
Subscription Services to Customer, and for conducting research,
product development and product improvement efforts . . . [and for
other uses].

17
18
19
20
21
22
23
24

Indeed, defendant RecycleSmart’s ownership was so fulsome that the agreement provided that

25

RecycleSmart (defendant) granted to Compology (plaintiff’s predecessor) a non-exclusive,

26

limited license for the Device Data (images) and Derivative Data (metadata) for use in

27

Compology’s Service (supra). Plaintiff does not challenge these plain meanings (see Opp.;

28

Supp. Opp.).
11

United States District Court
Northern District of California

1

Because the contract shows defendants owned the images and metadata as a matter of

2

law, and that plaintiff merely licensed them non-exclusively, plaintiff cannot bring an action

3

asserting it possessed the trade secret, nor in any case prove that defendants misappropriated it:

4

As to possession, plaintiff did not “possess” the trade secret. Under federal law,

5

“[p]laintiff [must] show[] that it is the owner of the alleged trade secret[].” See uSens, Inc. v.

6

Shi Chi, No. 18-cv-01959-SVK, 2018 U.S. Dist. LEXIS 175570, at *8 (N.D. Cal. Oct. 11,

7

2018) (Judge Susan Van Keulen). A non-owner lacks even standing. 18 U.S.C. § 1836(b)(1).

8

A non-exclusive, limited licensee is not “the person or entity” who owns the trade secret. 18

9

U.S.C. § 1839(4); see also, e.g., Genesis 1 Oil Servs. LLC v. Wismann Grp., LLC, No.

10

820CV02114SSSADSX, 2023 WL 3040584, at *3–5 (C.D. Cal. Mar. 20, 2023) (Judge

11

Sunshine S. Sykes). Under state law, no statutory provision or California Supreme Court

12

decision is as clear. Still, California authorities articulate a similar standard: “[T]he plaintiff

13

[must have] owned a trade secret.” See Sargent Fletcher, Inc. v. Able Corp., 110 Cal. App. 4th

14

1658, 1665 (2003); cf. Memry Corp. v. Ky. Oil Tech., N.V., No. C-04-03843 RMW, 2006 WL

15

3734384, at *7-8 (N.D. Cal. Dec. 18, 2006) (Judge Ronald M. Whyte) (CUTSA).

16

As to misappropriation, even assuming plaintiff could meet “possession” by holding

17

merely a nonexclusive license, plaintiff could not show that these defendants

18

“misappropriated” what they owned outright. It is “factually impossible.” See Mann v. Bales,

19

No. 16-cv-9623, 2018 U.S. Dist. LEXIS 122754, at *4 (N.D. Ill. July 23, 2018) (Judge Sharon

20

Johnson Coleman); HealthBanc Int’l, LLC v. Synergy Worldwide, Inc., 208 F. Supp. 3d 1193,

21

1201 (D. Utah 2016) (Judge Jill Parrish). Nothing “of another” was taken or used. See 18

22

U.S.C. § 1839(5); Cal. Civ. Code § 3426.1(b). In short, the contract makes clear the labeled

23

images are not “[Compology]-owned property acquired by [RecycleSmart],” but the

24

opposite — with “all rights” residing in RecycleSmart and only a limited right to non-exclusive

25

use residing in Compology. See Silvaco Data Sys. v. Intel, 184 Cal. App. 4th 210, 245, (2010),

26

abrogated on other claim by Kwikset Corp. v. Superior Ct., 246 P.3d 877, 894 (Cal. 2011).

27
28

Finally, plaintiff cannot escape these results (as litigants might in other settings) because
what its trade secret identification claimed (images plus metadata) is what the contract
12

United States District Court
Northern District of California

1

conceded (images plus metadata). If plaintiff meant to distinguish the two (as might a property

2

owner who distinguishes its mineral rights from another’s surface rights), plaintiff was not up

3

to the task (supra). Plaintiff gave up trying to identify a trade secret in its computer models

4

that defendants might have taken. Plaintiff kept trying to assert a trade secret instead in the

5

“images and data captured by the Compology sensors installed in Defendants’ containers”

6

(Opp. 20–21 (quoting Br. 23)). But “all rights” to them already belonged to defendants.

7

Plaintiff’s counterarguments in its opposition and extra briefing are not persuasive:

8

In opposition, plaintiff first asserts that it can bring a trade secret claim because,

9

notwithstanding the ownership and limited license provisions, a separate provision addresses

10

usage rights (Opp. 24), about which more is said below. But in every case plaintiff cites for

11

support, the plaintiff (or counter-plaintiff) owned “all rights” to the information asserted as the

12

trade secret — and here it is just the opposite. In Beco Dairy Automation, Inc. v. Global Tech

13

Systems, Inc., the contract at issue stated that “[counter-plaintiff] owns all rights” and that

14

“[counter-defendant] can’t reproduce, copy, or distribute any hardware or software

15

involved . . . [or] sell devices that are directly competitive.” No. 1:12-cv-01310 LJO SMS,

16

2015 WL 5732595, at *9 (E.D. Cal. Sept. 28, 2015) (Judge Lawrence J. O’Neill). In Khoros,

17

LLC v. Lenovo (United States), Inc., the contract at issue asserted that “[a]ll rights to all

18

[software] remain exclusively [plaintiff’s].” No. 3:20-cv-03399-WHO, Dkt. No. 49-4 §§ 3.1,

19

3.4 (N.D. Cal. July 15, 2020). Moreover, the misappropriation plausibly alleged was that

20

defendants gained access to the application source code and from it created a copycat

21

application. 2020 WL 12655516, at *8 (N.D. Cal. Oct. 5, 2020) (Judge William H. Orrick). In

22

Creative Writer Software, LLC v. Schechter, the contract at issue stated that “all title,

23

copyright, intellectual property rights and distribution rights in and to the Software” were

24

retained by plaintiff, and only a license for limited use transferred to defendant. No. CV 23-

25

02471-MWF (MAAx), Dkt. No. 1-1 § 3 (C.D. Cal. Apr. 3, 2023). The misappropriation

26

alleged was that defendant used the access to replicate the software. See 2024 WL 1098759, at

27

*1 (C.D. Cal. Feb. 8, 2024) (Judge Michael W. Fitzgerald). And, in Reingold v. Swiftships,

28

Inc., the contract at issue was plaintiff’s lease that granted defendant limited rights. 126 F.3d
13

United States District Court
Northern District of California

1

645, 650 (5th Cir. 1997) (applying Louisiana’s USTA statute). Plaintiff fails to grapple with

2

the distinction (Opp. 24): Defendants fully owned what plaintiff claims was its trade secret.

3

In supplemental briefing, plaintiff again points to use provisions (see Supp. Opp. (citing

4

Terms 2020 § 2.1.5)). But plaintiff again fails even to try to show how those provisions

5

plausibly abridge that RecycleSmart “own[ed] all rights and title to any Device Data and

6

Derivative Data” (Terms 2020 § 5.1). In the contract, each party took out a license in what the

7

other owned (id. §§ 2.1, 5.2): the provider’s software services, on the one hand, and the

8

customer’s data flowing through them, on the other. Compology owned the “System” (§ 5.1),

9

that is, the “integrated” hardware and software services used “to enable the collection,

10

processing, and/or delivery of Device Data and Derivative Data” (id. §§ 2.1.1, 1.17). And,

11

RecycleSmart “own[ed] all rights and title to any Device Data and Derivative Data” (id. § 5.2).

12

The use provisions plaintiff cites arose in the section that set out RecycleSmart’s license to use

13

Compology’s System (id. § 2), specifically in the subsection stating what RecycleSmart could

14

not do within “the scope of the license granted” to use Compology’s System (id. § 2.1.5; see

15

Supp. Opp. at 1 (discussing all of section 2.1.5)). Again, RecycleSmart did not need a license

16

to use Device Data or Derivative Data in which it owned “all rights” (see Terms 2020 § 5.2).

17

Lest there be any conflict as to the relationship between use rights in the System and use rights

18

in the Device Data and Derivative Data, the contract plaintiff drafted deconflicted the two:

19

Compology’s license to use the Device Data and Derivative Data was limited as stated without

20

exception (id. § 5.2), whereas RecycleSmart’s license to use the System was limited as stated

21

except “as expressly set forth” elsewhere (id. § 2.1.5). Similarly, as to ownership rights,

22

Compology drew the short straw: Compology owned all rights to the “System,” “except as

23

otherwise expressly set forth,” crimping its rights (id. § 5.1 (emphasis added)). RecycleSmart

24

owned “all rights” to the “Device Data and Derivative Data,” without exception (id. § 5.2).

25

Plaintiff never contested the contract’s express language regarding RecycleSmart’s

26

ownership of the labeled images and its limited licensing of them to Compology (i.e., section

27

5.2), which is unambiguous and dispositive on this issue. Nor did plaintiff point to any

28

extrinsic evidence to support its view or counter this conclusion (Opp. 23–24; Supp. Op. 5).
14

1

The contract (arguably) may create a contract claim against RecycleSmart. But it

2

undercuts plaintiff’s ability to assert a trade secret claim in the labeled images, which the

3

contract established defendant fully owned, and licensed only non-exclusively to plaintiff. As

4

a result, plaintiff’s trade secret claim regarding the “Training Data” fails as a matter of law, and

5

the motion for summary judgment as to it is GRANTED.

6
7

There is no need to reach parties’ other arguments about whether any asserted trade
secrets were “valuable to the business because they were not readily known.”
CONCLUSION

United States District Court
Northern District of California

8
9

For reasons above, defendants’ motion for partial summary judgment on the trade secret

10

claims (Dkt. No. 132) is GRANTED. The Court earlier granted plaintiff’s motion for leave to

11

amend “subject to paring down” the claims (Prior Order 2). The claims instead shifted,

12

kaleidoscopically expanded, or were improperly pleaded from the start. No reasonable jury

13

could find plaintiff’s trade secrets for plaintiff on this record. Our jury will not join plaintiff in

14

striking about in the dark for “potential[]” trade secrets. See Imax, 152 F.3d at 1167. The

15

request for notice (Dkt. No. 147) is DENIED AS MOOT.

16

Only contract claims remain in the case.

17

IT IS SO ORDERED.

18
19

Dated: November 23, 2024.

20
WILLIAM ALSUP
UNITED STATES DISTRICT JUDGE

21
22
23
24
25
26
27
28
15