Case 750
The backlog was counted in items and the items grew
the_backlog_was_counted_in_items_and_the_items_grew.eml - The backlog is tracked as a count of unprocessed items, charted daily, and it has been flat for a year. What a flat count means is computed below.
ok: true — round-trip fixpoint reached (python1 == python2)updated 2026-09-08
EML
eml# Self-authored for the EML case corpus (no external origin). The backlog is
# tracked as a count of unprocessed items, charted daily, and it has been flat
# for a year. What a flat count means is computed below.
#
# Tracking it is better than not, and the chart is honest. The count is taken
# from the store rather than from the worker's own opinion, it is charted daily
# so a trend is visible rather than a snapshot, the target was agreed rather
# than assumed, and the number has genuinely not moved: the backlog held between
# eleven and thirteen thousand items every week of the year.
#
# The count is of ITEMS. An item is a document to be indexed, and the mean
# document has grown from four kilobytes to thirty-one over the same year,
# because the product started accepting attachments.
#
# The work is in the bytes.
12000 => items_in_the_backlog
4 => mean_item_kilobytes_a_year_ago
31 => mean_item_kilobytes_now
365 => days_charted
0 => charts_of_the_backlog_in_bytes
1 => targets_agreed_with_the_business
0 => alerts_on_the_mean_item_size
items_in_the_backlog * mean_item_kilobytes_a_year_ago => backlog_kilobytes_a_year_ago
items_in_the_backlog * mean_item_kilobytes_now => backlog_kilobytes_now
backlog_kilobytes_now - backlog_kilobytes_a_year_ago => kilobytes_added_with_no_change_in_the_count
int(mean_item_kilobytes_now * 100 / mean_item_kilobytes_a_year_ago) => size_now_as_percent_of_then
"items in the backlog : " + str(items_in_the_backlog) ^0
"days charted : " + str(days_charted) ^0
"targets agreed with the business: " + str(targets_agreed_with_the_business) ^0
"charts of the backlog in bytes : " + str(charts_of_the_backlog_in_bytes) ^0
"" ^0
"mean item, KB a year ago : " + str(mean_item_kilobytes_a_year_ago) ^0
"mean item, KB now : " + str(mean_item_kilobytes_now) ^0
" as a percent of then : " + str(size_now_as_percent_of_then) ^0
"alerts on the mean item size : " + str(alerts_on_the_mean_item_size) ^0
"" ^0
"backlog KB a year ago : " + str(backlog_kilobytes_a_year_ago) ^0
"backlog KB now : " + str(backlog_kilobytes_now) ^0
" added while the count held flat : " + str(kilobytes_added_with_no_change_in_the_count) ^0
"" ^0
# ---- what the chart verified ----
"the backlog chart" ^0
" where the number comes from : the store, not the" ^0
" worker's own opinion" ^0
" charted : daily, so a trend is visible rather than a" ^0
" snapshot" ^0
" target : agreed, not assumed" ^0
" days charted : " + str(days_charted) ^0
" has the number moved : no; between eleven and thirteen" ^0
" thousand every week" ^0
" verdict : STABLE, AND HONESTLY MEASURED" ^0
"" ^0
" reading the count from the store rather than from the" ^0
" consumer is the difference between a metric and a" ^0
" self-report, and it was done" ^0
"" ^0
# ---- what an item is ----
"the counted unit" ^0
" what one item is : a document to be indexed" ^0
" what indexing one costs : proportional to its bytes" ^0
" mean bytes a year ago : " + str(mean_item_kilobytes_a_year_ago) + " KB" ^0
" mean bytes now : " + str(mean_item_kilobytes_now) + " KB" ^0
" why : the product started accepting attachments" ^0
" charts in bytes : " + str(charts_of_the_backlog_in_bytes) ^0
"" ^0
" the count is exact and the unit it counts changed size" ^0
" underneath it" ^0
"" ^0
# ---- the flat line is the finding ----
# Holding the count flat while each item grew by a factor means the workers
# absorbed the growth. The chart records the absence of a change in a number
# whose stability now costs eight times what it did.
"what a flat line meant then and now" ^0
" a year ago : the workers keep up with the arrival rate" ^0
" now : the workers keep up with an arrival rate" ^0
" carrying " + str(size_now_as_percent_of_then) + " percent of the bytes" ^0
" did anything about the chart change : no" ^0
" did anything about the system change : the work per" ^0
" item, by that factor" ^0
" bytes added while the line stayed flat : " ^0
" " + str(kilobytes_added_with_no_change_in_the_count) + " KB" ^0
"" ^0
# ---- what the target means now ----
"the agreed target" ^0
" what it says : keep the backlog under a count" ^0
" when it was agreed : when an item was " + str(mean_item_kilobytes_a_year_ago) + " KB" ^0
" what it constrains today : a queue holding " ^0
" " + str(backlog_kilobytes_now) + " KB" ^0
" was the target renegotiated : it did not appear to need" ^0
" renegotiating; the number it names is inside it" ^0
" alerts on the quantity that moved : " ^0
" " + str(alerts_on_the_mean_item_size) ^0
"" ^0
# ---- null control ----
# The same chart, with the backlog also plotted in bytes, and the target
# expressed in the quantity the workers actually spend.
1 => nc_charts_of_the_backlog_in_bytes
1 => nc_alerts_on_the_mean_item_size
"null control - plot the backlog in bytes as well" ^0
" items in the backlog : " + str(items_in_the_backlog) + ", unchanged and still exact" ^0
" charts in bytes : " + str(nc_charts_of_the_backlog_in_bytes) ^0
" alerts on the mean item size : " + str(nc_alerts_on_the_mean_item_size) ^0
" the count did not become wrong; a second line appeared" ^0
" in the unit the work is denominated in" ^0
"" ^0
# ---- the rule ----
"what a flat backlog count guarantees" ^0
" the number of unprocessed items is not growing :" ^0
" exactly, measured from the store, charted daily for" ^0
" " + str(days_charted) + " days" ^0
" the backlog is not growing : not addressed; the count" ^0
" is in items and the work is in bytes, and nothing" ^0
" holds the conversion still" ^0
"" ^0
"a count is a measurement only while the thing counted keeps" ^0
"its size; where the unit can grow, a flat line records that" ^0
"the growth was absorbed rather than that it did not happen" ^0
"" ^0
"The chart is honestly built - read from the store rather than the worker, drawn" ^0
"daily for " + str(days_charted) + " days against an agreed target - and the count has genuinely not" ^0
"moved. It counts items, and the mean item went from " + str(mean_item_kilobytes_a_year_ago) + " KB to " + str(mean_item_kilobytes_now) + ", " ^0
str(size_now_as_percent_of_then) + " percent of what it was, so " + str(kilobytes_added_with_no_change_in_the_count) + " KB entered the backlog while the" ^0
"line stayed flat, under " + str(charts_of_the_backlog_in_bytes) + " charts in bytes and " + str(alerts_on_the_mean_item_size) + " alerts on the size." ^0Python (deterministic transpilation)
pythonitems_in_the_backlog = 12000
mean_item_kilobytes_a_year_ago = 4
mean_item_kilobytes_now = 31
days_charted = 365
charts_of_the_backlog_in_bytes = 0
targets_agreed_with_the_business = 1
alerts_on_the_mean_item_size = 0
backlog_kilobytes_a_year_ago = items_in_the_backlog * mean_item_kilobytes_a_year_ago
backlog_kilobytes_now = items_in_the_backlog * mean_item_kilobytes_now
kilobytes_added_with_no_change_in_the_count = backlog_kilobytes_now - backlog_kilobytes_a_year_ago
size_now_as_percent_of_then = int(mean_item_kilobytes_now * 100 / mean_item_kilobytes_a_year_ago)
print("items in the backlog : " + str(items_in_the_backlog))
print("days charted : " + str(days_charted))
print("targets agreed with the business: " + str(targets_agreed_with_the_business))
print("charts of the backlog in bytes : " + str(charts_of_the_backlog_in_bytes))
print("")
print("mean item, KB a year ago : " + str(mean_item_kilobytes_a_year_ago))
print("mean item, KB now : " + str(mean_item_kilobytes_now))
print(" as a percent of then : " + str(size_now_as_percent_of_then))
print("alerts on the mean item size : " + str(alerts_on_the_mean_item_size))
print("")
print("backlog KB a year ago : " + str(backlog_kilobytes_a_year_ago))
print("backlog KB now : " + str(backlog_kilobytes_now))
print(" added while the count held flat : " + str(kilobytes_added_with_no_change_in_the_count))
print("")
print("the backlog chart")
print(" where the number comes from : the store, not the")
print(" worker's own opinion")
print(" charted : daily, so a trend is visible rather than a")
print(" snapshot")
print(" target : agreed, not assumed")
print(" days charted : " + str(days_charted))
print(" has the number moved : no; between eleven and thirteen")
print(" thousand every week")
print(" verdict : STABLE, AND HONESTLY MEASURED")
print("")
print(" reading the count from the store rather than from the")
print(" consumer is the difference between a metric and a")
print(" self-report, and it was done")
print("")
print("the counted unit")
print(" what one item is : a document to be indexed")
print(" what indexing one costs : proportional to its bytes")
print(" mean bytes a year ago : " + str(mean_item_kilobytes_a_year_ago) + " KB")
print(" mean bytes now : " + str(mean_item_kilobytes_now) + " KB")
print(" why : the product started accepting attachments")
print(" charts in bytes : " + str(charts_of_the_backlog_in_bytes))
print("")
print(" the count is exact and the unit it counts changed size")
print(" underneath it")
print("")
print("what a flat line meant then and now")
print(" a year ago : the workers keep up with the arrival rate")
print(" now : the workers keep up with an arrival rate")
print(" carrying " + str(size_now_as_percent_of_then) + " percent of the bytes")
print(" did anything about the chart change : no")
print(" did anything about the system change : the work per")
print(" item, by that factor")
print(" bytes added while the line stayed flat : ")
print(" " + str(kilobytes_added_with_no_change_in_the_count) + " KB")
print("")
print("the agreed target")
print(" what it says : keep the backlog under a count")
print(" when it was agreed : when an item was " + str(mean_item_kilobytes_a_year_ago) + " KB")
print(" what it constrains today : a queue holding ")
print(" " + str(backlog_kilobytes_now) + " KB")
print(" was the target renegotiated : it did not appear to need")
print(" renegotiating; the number it names is inside it")
print(" alerts on the quantity that moved : ")
print(" " + str(alerts_on_the_mean_item_size))
print("")
nc_charts_of_the_backlog_in_bytes = 1
nc_alerts_on_the_mean_item_size = 1
print("null control - plot the backlog in bytes as well")
print(" items in the backlog : " + str(items_in_the_backlog) + ", unchanged and still exact")
print(" charts in bytes : " + str(nc_charts_of_the_backlog_in_bytes))
print(" alerts on the mean item size : " + str(nc_alerts_on_the_mean_item_size))
print(" the count did not become wrong; a second line appeared")
print(" in the unit the work is denominated in")
print("")
print("what a flat backlog count guarantees")
print(" the number of unprocessed items is not growing :")
print(" exactly, measured from the store, charted daily for")
print(" " + str(days_charted) + " days")
print(" the backlog is not growing : not addressed; the count")
print(" is in items and the work is in bytes, and nothing")
print(" holds the conversion still")
print("")
print("a count is a measurement only while the thing counted keeps")
print("its size; where the unit can grow, a flat line records that")
print("the growth was absorbed rather than that it did not happen")
print("")
print("The chart is honestly built - read from the store rather than the worker, drawn")
print("daily for " + str(days_charted) + " days against an agreed target - and the count has genuinely not")
print("moved. It counts items, and the mean item went from " + str(mean_item_kilobytes_a_year_ago) + " KB to " + str(mean_item_kilobytes_now) + ", ")
print(str(size_now_as_percent_of_then) + " percent of what it was, so " + str(kilobytes_added_with_no_change_in_the_count) + " KB entered the backlog while the")
print("line stayed flat, under " + str(charts_of_the_backlog_in_bytes) + " charts in bytes and " + str(alerts_on_the_mean_item_size) + " alerts on the size.")stdout (executed)
textitems in the backlog : 12000
days charted : 365
targets agreed with the business: 1
charts of the backlog in bytes : 0
mean item, KB a year ago : 4
mean item, KB now : 31
as a percent of then : 775
alerts on the mean item size : 0
backlog KB a year ago : 48000
backlog KB now : 372000
added while the count held flat : 324000
the backlog chart
where the number comes from : the store, not the
worker's own opinion
charted : daily, so a trend is visible rather than a
snapshot
target : agreed, not assumed
days charted : 365
has the number moved : no; between eleven and thirteen
thousand every week
verdict : STABLE, AND HONESTLY MEASURED
reading the count from the store rather than from the
consumer is the difference between a metric and a
self-report, and it was done
the counted unit
what one item is : a document to be indexed
what indexing one costs : proportional to its bytes
mean bytes a year ago : 4 KB
mean bytes now : 31 KB
why : the product started accepting attachments
charts in bytes : 0
the count is exact and the unit it counts changed size
underneath it
what a flat line meant then and now
a year ago : the workers keep up with the arrival rate
now : the workers keep up with an arrival rate
carrying 775 percent of the bytes
did anything about the chart change : no
did anything about the system change : the work per
item, by that factor
bytes added while the line stayed flat :
324000 KB
the agreed target
what it says : keep the backlog under a count
when it was agreed : when an item was 4 KB
what it constrains today : a queue holding
372000 KB
was the target renegotiated : it did not appear to need
renegotiating; the number it names is inside it
alerts on the quantity that moved :
0
null control - plot the backlog in bytes as well
items in the backlog : 12000, unchanged and still exact
charts in bytes : 1
alerts on the mean item size : 1
the count did not become wrong; a second line appeared
in the unit the work is denominated in
what a flat backlog count guarantees
the number of unprocessed items is not growing :
exactly, measured from the store, charted daily for
365 days
the backlog is not growing : not addressed; the count
is in items and the work is in bytes, and nothing
holds the conversion still
a count is a measurement only while the thing counted keeps
its size; where the unit can grow, a flat line records that
the growth was absorbed rather than that it did not happen
The chart is honestly built - read from the store rather than the worker, drawn
daily for 365 days against an agreed target - and the count has genuinely not
moved. It counts items, and the mean item went from 4 KB to 31,
775 percent of what it was, so 324000 KB entered the backlog while the
line stayed flat, under 0 charts in bytes and 0 alerts on the size.Trace event types
eml:run:starteml:assigneml:outputeml:run:done