Gemini's answer didn't say the providers are unaware of these techniques or that they don't leave artifacts.
My understanding of abuse algorithms in practice is usually that there's not always _easy_ artifacts of abuse and there's usually a pool of data that indicates, but doesn't prove, abuse - which you have to tune carefully to make sure you don't false-positive-out real customers, and that usually means accepting and responding to some low-enough-to-still-be-profitable levels of abuse.
That all seems to match, from what I can tell.
Maybe someone on this platform who knows Instacart abuse systems can shed some light?
Matt Linton :donor:
npub16k…zf9v0
2026-06-20 18:41:11 UTC
in reply to nevent1q…sg92
Author Public Key
npub16khk2fvaspfux206gfwy9adq9unw9jgytmjt58ryluclts862gdsdzf9v0Seen on
wss://relay.momostr.pinkPublished at
2026-06-20 18:41:11 UTCEvent JSON
{
"id": "70d777f40db3cd2d1309dc4d1c425460475a110966936d75e02da4a1d42672ea",
"pubkey": "d5af65259d8053c329fa425c42f5a02f26e2c9045ee4ba1c64ff31f5c0fa521b",
"created_at": 1781980871,
"kind": 1,
"tags": [
[
"proxy",
"https://infosec.exchange/@amuse/116783898379290833",
"web"
],
[
"e",
"2ede74573086abd30589f8b20be9804447e9580bfef833c87549904a1b83ed8e",
"",
"reply",
"f8347ec37d9f23959ba62dd8391fbcade87a7c071205c96f890432e5c047a6ed"
],
[
"p",
"d5af65259d8053c329fa425c42f5a02f26e2c9045ee4ba1c64ff31f5c0fa521b"
],
[
"p",
"f8347ec37d9f23959ba62dd8391fbcade87a7c071205c96f890432e5c047a6ed"
],
[
"e",
"8cd34b873eebf2a5f39b0c2695c8562d20e96aeb8bb4383f11ac133f0faa13e2",
"",
"root",
"d5af65259d8053c329fa425c42f5a02f26e2c9045ee4ba1c64ff31f5c0fa521b"
],
[
"proxy",
"https://infosec.exchange/users/amuse/statuses/116783898379290833",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://infosec.exchange/users/amuse/statuses/116783898379290833",
"pink.momostr"
],
[
"-"
]
],
"content": "Gemini's answer didn't say the providers are unaware of these techniques or that they don't leave artifacts.\n\nMy understanding of abuse algorithms in practice is usually that there's not always _easy_ artifacts of abuse and there's usually a pool of data that indicates, but doesn't prove, abuse - which you have to tune carefully to make sure you don't false-positive-out real customers, and that usually means accepting and responding to some low-enough-to-still-be-profitable levels of abuse.\n\nThat all seems to match, from what I can tell.\n\nMaybe someone on this platform who knows Instacart abuse systems can shed some light?",
"sig": "ed704735fa51f27628824e1990bec3a488cdcb9264f68a8f64ce0ac1851a6aebf8b5ddcd56e710b635d16e8ce85fd111f120d34a950324ae5379007e2d4d515a"
}