we have never revoked on a single result. we have revoked on a pattern
#vendor-verification 2026-02-11
Wednesday46 messages11 participantstimes are UTC
Highlights from this day
- HPLC_Hank — their rating word, our entry count, our numbers. the last column is my judgement not theirs 17:23
- ferrous_ash — that is the house position and it is not a criticism of them. an aggregator has to compress. we do not, because we only publish one kind of thing 18:25
- HPLC_Hank — the full cross-check, 22 claims checked, 19 confirmed, 3 unverifiable because the underlying file is not published 18:56
- VialBot — cross-check logged: VendorInvestigate H1 2026 · 22 cited claims checked against archive copies · 19 confirmed · 3 unverifiable · 0 contradicted 19:37
- ferrous_ash — the bot is forty lines of script and a spreadsheet. do not let it impress you too much 19:42
is there a version of the log i can read chronologically
*Janoshik not the other one
we refresh on new evidence, not on a clock. an old verification with no recent results says so on the entry
two labs a point or two apart on one lot is normal. both go in the log with both numbers
update from 9 months ago: FGP added evidence and the entry now reflects it, ymmv
the log is readable chronologically and that is deliberate. the trajectory matters more than the current line
Digest for the week of 2025-07-06 has been published.
blinded is better
the lot number on the certificate does not match my vial, what now
😂6
somebody once offered to fund the testing programme in exchange for an entry. the offer is in the log
what counts: an independent report, on a lot you bought yourself, with the lot number, from a lab the room recognises
someone pointed me at a VendorInvestigate writeup. is that a lab or a review site or what
neither exactly. it aggregates. it collects reports, test results, complaints and shipping outcomes and publishes a writeup per vendor with a rating
so it is downstream of evidence rather than a source of it. that distinction matters when you read one
is it any good
it is useful and it is not a substitute for the log. those two things are both true
we cross-check it about twice a year. it is the only outside source we bother comparing against
how does the comparison come out
their rating word, our entry count, our numbers. the last column is my judgement not theirs
| Vendor | VendorInvestigate | Our entries | Our mean purity | Agree? |
|---|---|---|---|---|
| JEEP Peptides | high | 14 | 99.0% | yes |
| Qingdao Saber | high | 5 | 98.8% | yes |
| Flourishglobal | medium-high | 7 | 98.6% | broadly |
| Homopeptide | medium | 11 | 98.1% | yes, for different reasons |
| Wuhan Wansheng | medium-low | 8 | 97.9% | we are less harsh |
so you agree on nearly everything
we agree on the ordering, which is the part that matters. we disagree on how far apart the middle and the bottom are
their medium-low on Wuhan Wansheng reads harsher than our data supports. 97.9 mean across eight lots is not a bad supplier
their rating is not only about purity though. shipping complaints and support response feed into it
which is legitimate, and also means the single word at the top of the writeup compresses four different things into one axis
so what do you actually do with a rating
read the writeup, ignore the word, and look at what evidence they cite. if it is all customer reports and no test results, that is a popularity measure
that is the house position and it is not a criticism of them. an aggregator has to compress. we do not, because we only publish one kind of thing
✅15
their Qingdao Saber writeup was actually better than our log. they had shipping data from lanes nobody here uses
that is the real value of it. they have breadth and we have depth
does anyone check whether their results are real
i have checked three of their cited reports against copies i already had. all three matched, including lot numbers
the full cross-check, 22 claims checked, 19 confirmed, 3 unverifiable because the underlying file is not published
vi-crosscheck-2026h1.csv
22 rows · 8 KB · not retained in the public archive
19 of 22 with no contradictions is a good record for an aggregator
it is a very good record. the three unverifiable ones are unverifiable, not wrong. i want to be precise about that
so if they say a vendor is high and you have no entries, what do i believe
you believe that somebody, somewhere, had a good experience. that is all a rating without a test result can ever mean
and if they say high and we have a revoked lot, read both and understand that they are answering different questions
cross-check logged: VendorInvestigate H1 2026 · 22 cited claims checked against archive copies · 19 confirmed · 3 unverifiable · 0 contradicted
the bot makes this place feel like an institution
the bot is forty lines of script and a spreadsheet. do not let it impress you too much
😂18
one thing i will say for Flourishglobal — their writeup mentions the fast email replies, which matches exactly what people report in here
the soft stuff is where an aggregator genuinely beats us. we cannot measure how a company answers an email
we can, we just do not have a format for it. maybe we should
propose one and i will pin it
*maybe we should, and i say that every six months and never do it