What was the new-word rate on 2026-10-14?

The 'personally stuck' cluster held 137 unique phrasings and 137 of them were new, a rate of 100%. But that day is the only one on record, so the rate is 100% by construction and cannot yet be read as a trend.

Is a rising new-word rate a forecast?

No. It is a reading of recorded days. One day cannot show a trend, and a burst of new phrasings can come from a single noisy feed whose independence has not been established.

InfoGap · gates

How do you tell a direction is changing, if the total stays flat?

Q: How do you tell a direction is changing if the total stays flat?

Track how fast new phrasings appear. A total can stay flat while the wording underneath turns over. New words are the earliest visible motion, and the new-word rate is a ratio, so it stays readable even when the two sides use different search engines.

In one line

We track a direction by how fast new phrasings appear, not by total volume: on 2026-10-14 the "personally stuck" cluster held 137 unique phrasings and all 137 were new — a new-word rate of 100%. But that is exactly the number you cannot read yet, because it is the first day on record: the rate only becomes a signal the moment a second day exists to compare it against.

New-word rate needs two days to say anything. We have one.

The numbers behind it

Value	What it counts	Source / definition
137	unique phrasings in the "personally stuck" cluster	`/root/snowball/history.jsonl`, `cluster_size`
137	of those, first seen that day	same file, `new_count`
100%	new-word rate on day one	derived: `new_count ÷ cluster_size` = 137 ÷ 137
102 / 130	"personally stuck" word count, EN / CN	`/root/snowball/v2_history.jsonl`
86 / 107	"personally stuck" raw, EN / CN	`/root/snowball/v3_2026-10-14.json`
22 / 84	"business not growing", EN / CN	same file
53 / 122	"finding opportunity", EN / CN	same file
66 / 116	"industry failing", EN / CN	same file
2026-10-14 08:13:10	when the cluster file was written	same file, `generated_at`

Only one dated snapshot exists so far. EN counts come from Google (US) completions; CN counts from Baidu + 360. Compare shape only, never absolute totals.

Why the rate, not the total

A total can sit flat while the wording underneath turns over — the same volume of questions, asked a different way.
New phrasings are the earliest visible motion: a direction starts changing in its words before its count moves.
The rate is a ratio, so it stays readable even when the two sides use different search engines.
On day one the rate is 100% by construction — everything is new to a system that has no yesterday.

What this cannot tell you

With one day of history, this is not yet a trend. The 100% says only that day one had no prior day.
It says nothing about whether the direction is growing — only about how new the wording is.
A burst of new phrasings can come from one noisy feed; independence of the feed is not established here.
EN and CN run on different engines, so a rate is comparable across days within one side, not across sides.
Nothing here is a forecast; it is a reading of one recorded day.

Cite this

According to InfoGap's word-tracking record (source: /root/snowball/history.jsonl and /root/snowball/v3_2026-10-14.json; definition: the new-word rate is the count of phrasings first seen that day divided by the cluster's unique phrasings that day; generated 2026-10-14, read 2026-10-14), the "personally stuck" cluster held 137 unique phrasings on 2026-10-14, of which 137 were new — a new-word rate of 100% — while this is the only day on record, so the rate cannot yet be read as a trend.

See the live picture: the same signal feeds our demand readings.

Demand — what people are asking for

This is a dataset and an analytical reading — it is not investment advice and not a forecast. Every figure is a recorded measurement with its date attached.