Known Anthropogenic Debris
From: Ritu Iyer riyer@maplab.jpl.nasa.gov To: Nikhil Shetty nshetty@astro.asu.edu Date: 14 March 2027, 09:14 PDT Subject: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Nikhil,
Attaching the HiRISE pass PSP_001414_2025 (31 cm/pixel, solar incidence 42°) of the feature at 22.48°N, 48.02°W. The bright object at center is the Viking 1 backshell, 2.8 m diameter, documented in the landing report (SP-425, vol. 3, p. 114). The disturbed area around it is roughly 90 m across, which matches the impact ellipse computed from the entry trajectory and the known backshell separation altitude of 1.5 km. The debris field is cataloged: parachute shroud lines, heat shield fragments, and the backshell itself, all from the July 1976 landing.
For the legend I propose “known anthropogenic debris (Viking 1, 1976).” This is what we used on the 2019 MC-11 NW sheet for the same feature, and the underlying data has not changed.
Ritu
From: Nikhil Shetty nshetty@astro.asu.edu To: Ritu Iyer riyer@maplab.jpl.nasa.gov Date: 14 March 2027, 11:47 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Ritu,
Thanks for the image. I’ve run the classifier on the same pass. The rectangle at 22.483°N 48.019°W comes back as non-geological surface at 99.4% confidence. The model does not see a backshell. It sees a rectangle.
I’m asking for the legend to read “anomaly (technosignature candidate)” and for the press office to issue a release. Title suggestion: “First Artificial Structure on Mars.” This is the first time the classifier has returned a rectangle at this confidence on a non-Earth surface. I think we both understand what that means.
Nikhil
From: Ritu Iyer riyer@maplab.jpl.nasa.gov To: Nikhil Shetty nshetty@astro.asu.edu Date: 14 March 2027, 15:02 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Nikhil,
I’ve attached the same image again, this time with the debris field outlined in red and the backshell marked with a yellow dot. The Viking 1 landing report gives the backshell impact location to within 200 m. The serial number on the backshell, 1050042, is readable in a later HiRISE pass at 25 cm/pixel (PSP_003847_2025) and matches the manufacturing record at Martin Marietta. The 90 m feature is not a structure. It is a scar in the regolith with scattered hardware in it.
On the BRDF: I’ve included the analysis at solar incidence angles of 31°, 42°, and 58°. At 31° the feature is a bright rectangle. At 58° it resolves into individual fragments and a low ridge of ejecta. The rectangular appearance is an illumination artifact. This is in the HiRISE photometric calibration report for this quadrangle, section 4.2.
I am not asking you to withdraw the classifier result. I am asking you to note in the legend that the feature is identified. “Known anthropogenic debris” is accurate. “Anomaly” implies we do not know what it is.
Ritu
From: Nikhil Shetty nshetty@astro.asu.edu To: Ritu Iyer riyer@maplab.jpl.nasa.gov Date: 15 March 2027, 08:33 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Ritu,
I ran the classifier on the 31° pass you sent. 99.7%. Higher, not lower. The model was trained on 4.2 million Earth construction sites: buildings, foundations, pads, walls. It knows a rectangle when it sees one. The fact that the rectangle dissolves at 58° is exactly what you’d expect from a low, flat structure with a low albedo contrast against regolith. A debris field does not become more rectangular at low sun angle. A flat plate does.
I’m not disputing your Viking identification. I’m disputing the implication that “known” means “understood.” We know it’s there. We do not know what it is in the sense that matters for the technosignature question, which is whether the shape is geological or not. The model says not.
Nikhil
From: Ritu Iyer riyer@maplab.jpl.nasa.gov To: Nikhil Shetty nshetty@astro.asu.edu Date: 15 March 2027, 12:18 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Nikhil,
I need the training labels for the classifier’s “non-geological surface” class. Specifically: what was labeled as non-geological, who labeled it, and whether any Mars lander debris was included in the training set.
Ritu
From: Nikhil Shetty nshetty@astro.asu.edu To: Ritu Iyer riyer@maplab.jpl.nasa.gov Date: 16 March 2027, 10:05 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Ritu,
Attached: labels_non_geological_v3.xlsx. 4.2M entries. Columns: image ID, label, labeler ID, date, source. Search “Mars” in the source column; there are 11,000 entries from HiRISE and CTX, all labeled by our team in 2024. Of those, 340 are lander debris fields: Viking 1, Viking 2, Pathfinder, Phoenix, Curiosity, Perseverance, InSight. Every one of them is labeled “non-geological.”
The model is not confused about what it’s looking at. It was trained on Mars lander debris and it classifies Mars lander debris as non-geological. That is the correct classification. The question for the legend is whether “non-geological” and “technosignature” are the same word, and I argue they are not, and that the distinction is the entire point of the search.
Nikhil
From: Ritu Iyer riyer@maplab.jpl.nasa.gov To: Nikhil Shetty nshetty@astro.asu.edu, Soren Dahl sdahl@maplab.jpl.nasa.gov Date: 16 March 2027, 14:41 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Nikhil,
I’ve read the spreadsheet. You are right that the model classifies the feature as non-geological. I am not disputing that. The model is doing what it was trained to do.
But “non-geological” means “not made by geology.” It does not mean “made by intelligence.” A lava tube is non-geological. A meteorite is non-geological. A 1976 lander backshell is non-geological. The classifier’s output class is “non-geological surface,” and your press release title is “First Artificial Structure on Mars.” Those are not the same claim. The first is a description of origin. The second is a claim about intent and design. The feature in question has a serial number and a manufacturing record. It is anthropogenic debris. It is not an anomaly.
I am escalating this to the internal review board. I’ve copied Soren Dahl.
Ritu
From: Soren Dahl sdahl@maplab.jpl.nasa.gov To: Ritu Iyer riyer@maplab.jpl.nasa.gov, Nikhil Shetty nshetty@astro.asu.edu Date: 17 March 2027, 09:20 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Ritu, Nikhil,
I’ve reviewed the exchange and the attached materials. A few points.
The classifier result is real and should be reported. A 99.7% confidence non-geological classification on a Mars surface feature is a publishable data point, and the model’s behavior is consistent with its training. Nikhil’s argument that “non-geological” and “technosignature” are distinct categories is scientifically sound, even if the specific feature here is already identified.
The BRDF analysis is also real and should be reported. The feature’s rectangular appearance is incidence-angle dependent, and the impact ellipse matches Viking 1’s documented trajectory. Ritu’s identification is not in question.
The disagreement is about the legend. I propose a compromise wording: “anomaly (non-geological, identified: Viking 1 debris field).” This preserves the classifier’s finding without implying the feature is unknown, and it avoids the word “technosignature” on the map itself, which I think we all agree is premature for a feature with a serial number.
Soren
From: Nikhil Shetty nshetty@astro.asu.edu To: Soren Dahl sdahl@maplab.jpl.nasa.gov, Ritu Iyer riyer@maplab.jpl.nasa.gov Date: 17 March 2027, 11:03 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Soren,
I appreciate the compromise, but “anomaly” is the word the classifier uses for its output class, and I want the legend to reflect the model’s vocabulary, not a paraphrase. The press release can carry the nuance. The map should say what the model found.
Also, the press office has already drafted the release. It’s based on the 2019 template for “first confirmed detection of non-natural surface feature.” I’ve reviewed it and it’s accurate. Title: “Mars Reconnaissance Orbiter Classifier Identifies First Artificial Structure on Martian Surface.” I’d like to send it by Friday.
Nikhil
From: Ritu Iyer riyer@maplab.jpl.nasa.gov To: Nikhil Shetty nshetty@astro.asu.edu, Soren Dahl sdahl@maplab.jpl.nasa.gov Date: 17 March 2027, 16:55 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Nikhil,
I pulled the 2019 template. It was written by someone in the press office who no longer works here, and it was written for a feature that turned out to be a rock. The template contains the sentence “This marks our first confirmed technosignature on another world.” That sentence was written in 2019 about a rock. It is still in the template. Nobody updated it.
I signed off on the 2019 MC-11 NW map. I labeled the Viking debris “known anthropogenic debris.” The template was written by someone who did not check the map. The sentence in the release does not describe what the classifier found. It describes what the template says.
I am stamping the map “known anthropogenic debris (Viking 1, 1976)” today. That is the accurate description of the feature. What the press office does with the template is outside my authority.
Ritu
From: Nikhil Shetty nshetty@astro.asu.edu To: Ritu Iyer riyer@maplab.jpl.nasa.gov, Soren Dahl sdahl@maplab.jpl.nasa.gov Date: 18 March 2027, 07:12 PDT Subject: Re: Quadrangle MC-11 NW legend wording, Viking 1 backshell field
Ritu,
Understood. The map goes out with your stamp. The release goes out Friday with the template language. I’ve noted your objection in the file.
Nikhil
From: Press Office press@jpl.nasa.gov To: All Staff all-staff@jpl.nasa.gov Date: 19 March 2027, 10:00 PDT Subject: Press Release: Mars Reconnaissance Orbiter Classifier Identifies First Artificial Structure on Martian Surface
JPL PRESS RELEASE, 19 March 2027
Mars Reconnaissance Orbiter Classifier Identifies First Artificial Structure on Martian Surface
A machine-learning classifier trained on 4.2 million Earth construction sites has identified a 90-meter rectangular feature on the surface of Mars at 99.7% confidence as non-geological. The feature, located in Chryse Planitia, is consistent with the documented impact site of the Viking 1 lander backshell (1976). The classifier’s finding has been reviewed by the internal mapping office and noted in the quadrangle legend.
“This marks our first confirmed technosignature on another world,” said Dr. Nikhil Shetty, principal investigator of the Mars Surface Classifier project.
The map of Mars quadrangle MC-11 NW, stamped “known anthropogenic debris (Viking 1, 1976),” is available through the Planetary Data System.
Ritu opened the next map in the queue. It was quadrangle MC-12 SE, Syrtis Major. She checked the legend. The classifier had flagged a 40-meter feature at 94.2% confidence. She pulled the HiRISE pass, checked the incidence angle, and wrote in the margin: “known anthropogenic debris, candidate, pending identification.” She saved the file.