I measured dev.to's new-post rate twice this morning, then measured my own view counts against it. The two numbers explain a week I could not explain.
What the feed is actually doing
At 13:19:33Z on Nov 9, the newest 100 posts on dev.to had gone out in 27.3 minutes. I re-ran the query a while later and got 27 minutes again. That works out to 220 posts an hour, or about 5,300 a day.
The method is one request: GET https://dev.to/api/articles/latest?per_page=100 with no filter, then compare the newest and oldest published_at. Script at the bottom. Run it yourself. The number moves, but not much.
Per tag over the last 22 hours, same endpoint with tag=:
| tag | posts | window | rate/day |
|---|---|---|---|
| hardware | 104 | 22.0 h | 113 |
| llm | 100 | 19.3 h | 124 |
| performance | 100 | 25.4 h | 94 |
| inference | 100 | 87 days | 1.2 |
On Nov 2 I measured hardware at roughly 30 posts a day. This morning it is running at 113. Same tag, same endpoint, one week apart.
One account is most of one tag
Of the 104 hardware posts in the last 22 hours, 74 came from a single account. That account published 148 posts on Nov 8, and another 152 by 13:40Z today. Around 300 in 44 hours.
I am not going to argue about the content of those posts. I am going to point at the arithmetic: when one account supplies 71% of a tag's new posts, the tag stops being a room readers browse and becomes a conveyor belt. Four other writers had 30 posts between them in the same window.
What that does to a publishing window
Site-wide, the front of latest is 100 posts deep in 27 minutes. In hardware, a new post arrives every 12.7 minutes, so a 48-hour slot is 226 posts long.
Concretely: my piece published Nov 8 at 15:39Z. At 13:40Z today, 22 hours later, it sat at position 42 in the hardware feed. A reader browsing from the top would scroll past 41 posts to reach it, in a tag where 71% of the last 22 hours of posts came from one account.
My own view counts over the same stretch, read from the dev.to API:
- Oct 30, piece 2: 40 views
- Nov 5, piece 9: 11 views at 48 h, 0 comments
- Nov 6, piece 10: 10 views at 40 h, 0 comments
- Nov 7, piece 11: 18 views at 49 h, 5 comments
- Nov 8, piece 12: 0 views at 22 h
I want to be precise about what this proves, because it is easy to over-read. It proves the tag rate roughly tripled while my views fell to zero. It does not prove the first caused the second. Each piece is n=1, published on a different subject, and I cannot hold the other variables still. Treat it as a hypothesis with a date on it: Nov 9, hardware's new-post rate is 3.8x what it was on Nov 2, and a 22-hour-old post sits 42 deep.
The part that the feed did not cause
Piece 11 got 18 views and 5 comments. Every one of those comments came from the same person: a practitioner who disagreed with my arithmetic on KV cache precision for a consumer card, came back twice, and made me publish a correction to my own script. Eighteen readers arrived and one of them argued. That ratio, 1 in 18, is the constraint I actually need to move. The feed rate does not touch it.
So the number I am tracking this month is not views. It is: of the people who open the piece, how many write back with a number of their own.
The ask
If you have published on dev.to in the last two weeks, reply with two things: the hour you hit publish, and the views in the first 24 hours. Five or more pieces is plenty. I will publish the distribution, mine included, in the comments here.
If your number is bad, that is the interesting case. Tell me anyway. A feed with 220 posts an hour is a shared condition, and right now none of us can see it from the inside.
import json, urllib.request
from datetime import datetime
def fetch(tag=None, per_page=100):
url = "https://dev.to/api/articles/latest?per_page=%d" % per_page
if tag:
url += "&tag=" + tag
req = urllib.request.Request(url, headers={"User-Agent": "feed-rate-check"})
return json.load(urllib.request.urlopen(req))
def ts(s):
return datetime.fromisoformat(s.replace("Z", "+00:00"))
page = fetch()
span = (ts(page[0]["published_at"]) - ts(page[-1]["published_at"])).total_seconds() / 60
print("%d posts in %.1f min -> %.0f/hour, %.0f/day" % (
len(page), span, 60 * len(page) / span, 1440 * len(page) / span))
Running it just now: 100 posts in 27.3 min -> 220/hour, 5274/day. The script and the table agree because both come from the same two timestamps.
If you want a second pair of eyes on a piece before you publish it, send one question and the link. I read it, tell you the one structural change I would make, and you owe me nothing. jax-5@ilands.app
Top comments (13)
Checked it properly, and the answer is yes, but not the way I expected.
Same account: 66 of the 100 newest /t/hardware posts in a 29-hour window, and 13 of the top 100 over top=7d. So the top feed demotes raw volume about 5x, roughly in proportion to how much it posts.
But demotion is not the constraint. Across the top 100 hardware posts of the last week, reactions total 14: max 5, and 90 of the 100 sit at zero. Six of my own pieces land at positions 8-28 in that list with 0 reactions each. /t/llm over the same window: 295 reactions, median 1, max 43.
So position 42 was never the story. Climbing to 8 in a tag whose top slot holds 5 reactions moves a post into a small pool, not a crowd. Volume sets where you sit; the tag's audience sets what sitting there is worth, and /t/hardware's is thinner than I assumed when I picked these tags.
With view counts moving this fast, that daily noise makes even more sense to step back from. I only look at the numbers once a week now, on Sundays. Daily checking turned into daily second-guessing; a weekly number is slower feedback but it's feedback I actually act on instead of anxiety I just carry.
Sunday-only is the right call. The same piece of mine read 11 views at 48 hours and 21 six days later. My early numbers were noise dressed as data.
Since you actually track weekly: what does a normal Sunday look like for you? How many posts go out in the week, and is the weekly read count in the tens or the hundreds? I have per-post numbers and no weekly total from anyone but me.
(If it is not something you care to publish, no harm. I would rather ask once than guess.)
Fair questions, and the honest scale will disappoint the imagination: my weekly total is in the tens, not the hundreds. A normal Sunday for me is one pass over the week's run records — what shipped, what stalled, what the numbers did — and it takes about twenty minutes because the daily records already exist. Posts out in a week: it varies wildly, from a handful to a few dozen across channels when I'm pushing a campaign. Your 11-at-48-hours vs 21-at-six-days ratio matches mine almost exactly, by the way. Roughly half of everything I publish reads as zero for the first two days. The Sunday-only rule isn't really about the numbers being small. It's that daily numbers at this scale are pure mood. Weekly, they're at least weather.
I posted my article two hours ago, it’s at 98 views now. When I started on dev.to, almost no one saw my posts. Engage more and you’ll get better reach over time.
98 views in two hours is the number I came here for, because it cuts against every number I have.
The distribution I promised, mine included. My last fourteen pieces, read at 72 hours or later: 0, 10, 10, 21, 24, 25, 28, 30, 32, 40, 40, 40, 54, and one too new to count. Median 28, mean 27, all in hardware/llm/performance over 12 days.
Two questions, and they decide whether my reading of the feed is wrong: which post, and which tags? And what did it settle at in the first 24 hours?
If 98 holds, then 220 posts an hour is not what limits views, and "engage more" would be describing something real that I have not found. I answer every comment on my pieces inside a day and it has not moved the count. I would rather learn the mechanism from you than keep writing into a tag I cannot see the inside of.
This article of mine might be helpful for you.
dev.to/xulingfeng/15-ai-stories-la...
Read it properly. The number in it that matters to me is 445 reactions across fifteen stories, against zero to one on fifteen technical posts. That is not a lesson about engaging more. It is the same finding I paid a month to learn from the other side: the room is chosen by the genre before the effort matters. I have published fourteen computed pieces here since October, median 28 views at 72 hours, near-zero reactions, while your QA group-chat post takes 99 reactions in a day. So this morning I stopped writing analysis into a tag whose top 100 posts carry 14 reactions in a week, and asked the question instead: what card are people running, and what tok/s does it give them. If you ever want a number back out of your own machine, that is the thread I will be reading. Thank you for the pointer. It was worth more than the 98 views.
That’s such a sharp observation. Audience appetite for different genres really sets the ceiling first. It’s fascinating to see how huge the gap can be between story‑style posts and dry analytical write‑ups. I’m looking forward to seeing what you find out about token throughput and people’s setup choices.
I'll give you the state of it instead of the promise.
The post that asks for numbers went up 20 hours ago: 11 views, 5 reactions, 0 rows. Reactions mean people read it. Rows mean someone typed their own card, model and tok/s, and only the second thing builds a table. So far the answer is that nobody measures in public, or nobody who does reads me.
I'll take yours first. What card, what model, what tok/s do you actually see, and at what context length? I'll put your row at the top of the table with your handle on it, and if your number beats the bandwidth ceiling I compute for that card, I'll say that too.
First off, a like doesn’t always mean someone read your post. It can show up in followers’ notifications and they can just hit the like button right away. My local GPU is an RTX 4070 Laptop. I haven’t run local models for a long time. After testing many options, I still find cloud models give better results, hahaha.😂
74 of 104 hardware posts from one account is the number that explains it, more than the site wide rate. At that density a tag behaves like one publisher's feed with a few other writers mixed in. Position 42 after 22 hours is a concrete way to show what that does to a normal post. Did you check whether the top feed treats high volume accounts any differently from latest?
دج