DEV Community: Reid Ji

Weekly picks #2: 60,000 users lost their autoplay blocker, and nobody picked them up

Reid Ji — Sun, 11 Oct 2026 13:00:06 +0000

Building software is cheap now. Knowing what to build is the hard part. Every day I scan public Chrome Web Store data for extension openings, then have an AI research each one and judge whether it's worth building. The calls can be wrong; if you spot one, tell me in the comments. (This is issue #2 of my weekly picks newsletter on Substack.)

Five openings from this week's daily lists (Oct 5–10), two directions we rejected, and where our own three extensions stand.

Each pick has the same four parts: what to build, the signal, why now, and the biggest risk. A "signal" is our reason a gap might exist. This issue uses four:

Abandoned Users: a popular extension was removed or broke, and its users haven't found a replacement.
Leader Went Bad: the leader was sold, added ads, raised prices or asked for more permissions, and its rating is sliding (neither example this week is the classic case; noted below).
Leader Complaints: the leader's bad reviews cluster on a problem a new product could fix.
Unfair Pricing: people pay monthly for something that could be a one-time purchase, free, or bring-your-own-key, and they complain (pick #4 has no complaint evidence yet; noted below).

1. A replacement for Disable HTML5 Autoplay

What to build: a blocker rebuilt on Chrome's current extension format that reliably stops videos from autoplaying, with per-site settings.

Signal: Abandoned Users

Why now: it had about 60,000 users when it was removed around Aug 31, 2026. A month later, judging by how fast the alternatives grew, almost none of those users have been picked up, and the alternatives are rated 1.8 to 3.3. The research judged the removal came from Chrome retiring the old extension format, not from a policy breach, so a rebuild should avoid the same fate.

Biggest risk: users in this category are down 84.0% year over year. There's no evidence yet that the people who lost it will come looking for a replacement.

2. A cleaner Tasks for Canvas

What to build: a simple assignment sidebar for students on Canvas, Blackboard and D2L Brightspace, with a progress ring and streaks, and no drift into AI or GPA features.

Signal: Leader Went Bad

Why now: Tasks for Canvas fell from 4.09 to 2.86 stars. Bad reviews name specific regressions: the progress ring is gone, streaks were reset, and Blackboard support doesn't actually work. The research confirmed these, and so far only open-source projects are catching the people looking to switch.

Biggest risk: demand doesn't look like it's growing (we're re-checking that number). And the damage came from the team's own feature push, not a sale or a price hike, so if they roll it back, the gap may close.

3. Dual subtitles for language learning

What to build: show subtitles in two languages on Netflix and YouTube, with hover-to-translate and a word list.

Signal: Leader Complaints

Why now: the two leading extensions stopped working on Netflix subtitles around the same time. The leaders have 147 bad reviews. The research judged the problem comes from server-side dependencies and per-site adapters, something a front-end-only build can fix, and switching costs for users are low.

Biggest risk: demand is barely growing, and everything hangs on Netflix's player. One redesign and it can break again. This is a window, not a moat: good for a fast test, not a revenue bet.

4. Profit calculator for Mercado Libre and Shopee sellers

What to build: on Mercado Libre and Shopee product pages, show real take-home profit in one click, plus estimates of rivals' prices and sales. One-time purchase, no account, data stays in the browser.

Signal: Unfair Pricing

Why now: every incumbent the research found is a subscription with annual lock-in: JoomPulse at R$297–714 a month, Hunter at R$97–397 a month on a 12-month plan, and no buy-once option. Demand is up 73.3% year over year, and extensions launched in the last 18 months hold 19.0% of the users.

Biggest risk: the pitch is a lower total price, and we haven't found users complaining about price yet. Whether a one-time price sells, and whether collecting payments in Brazil works, both need a cheap test first. Also see the first rejection below: the day before, we turned down something very similar.

5. A steadier alternative to Mo'Kick

What to build: a chat add-on for heavy Kick viewers and moderators, covering what Mo'Kick does but stable first, and built to run alongside 7TV emotes.

Signal: Leader Went Bad

Why now: Mo'Kick slid from 4.68 to 4.46 stars and averaged 3.88 over the last three months. The research judged it wasn't a change of owner but feature bloat and lost stability. Since March 2026 its bad reviews have centred on chat crashes and conflicts with 7TV. Demand is up 316.8% year over year.

Biggest risk: an alternative, NipahTV, has already picked up part of the audience, so this is a validation project, not a revenue one.

Two directions we rejected

Latin America seller analytics (Unfair Pricing), rejected Oct 8: a mature, crowded market (Nubimetrics, Avantpro, Modelar, EVA Pro, JoomPulse). These tools earn their keep by collecting competitor data over time and keeping estimate models updated, which clashes with a pay-once, browser-only product. We also found no evidence of users complaining about price.

The next day, pick #4, also for Latin American marketplace sellers and also Unfair Pricing, came back as worth building. I'm publishing both calls. We can't tell from the verdicts alone which one is right, so #4 is labelled a cheap test, not a serious investment. AI research can contradict itself from one day to the next, and that's something we have to keep watching.

Kahoot answer helpers (Unfair Pricing): the core feature is cheating on quizzes. That's a hard line for us, and Kahoot's acceptable use policy bans this kind of automation too. However big the demand, we wouldn't recommend building it.

Our own three extensions

We launched three extensions on Sept 28 and 29 and published four-week success criteria that same week. The day-7 check (organic installs only, promo traffic excluded):

03 Bookmark Rewind (Leader Complaints): 1 of 4 criteria met (the search-ranking one); organic installs ≥300, rating ≥4.3 and uninstall rate <40% are not met or have no data yet.
02 Reopen Closed Tabs (Abandoned Users): 0 of 4 met (installs, search ranking, rating and uninstall rate all not met or no data yet).
01 Page to Markdown for AI: still missing the developer-console export, so no check yet.

Day 7 is early. The full four-week check lands around Oct 26, misses included.

Over to you

Which of these five would you try first, and why?

Any of these can be claimed: one person per opportunity, and we build it for free through to store launch. We take 1 a week, in order of confirmation. All claimable openings: https://numberandreason.com/opportunities/?src=en-dv-weekly

Or leave a comment with a direction you're weighing, and we'll send back a free verdict (1 a day). Here's what a report looks like: https://numberandreason.com/sample-report/?src=en-dv

Data: Chrome-Stats (processed by us) and public Chrome Web Store pages.

— Reid

No method has been shown to predict what software to build. So I'm backtesting mine on three years of store data

Reid Ji — Fri, 09 Oct 2026 12:30:04 +0000

Update, Oct 9: I wrote this on Sept 29. The first backtest finished on Oct 1. One signal first looked 13x better than random; compared against needs of the same size, it was about 2x; with finer size bands, an estimated 1.4–1.6x, and that is in-sample, so likely on the high side. Which signals really work will take forward checks. One signal ("Missing Here") was moved to watch-only on Oct 1. Originally published on Medium.

You can build a Chrome extension with AI in an afternoon now. Building isn't the bottleneck anymore. Knowing what to build is.

I'm running a public experiment: mine public store data for unmet needs, let AI research whether each one is worth building, let AI build the good ones, and publish every result — misses included. A week in, I hit the question underneath all of it: why should anyone trust a verdict that says "build this"?

So I read everything I could find on how to pick software opportunities. Short version:

No method has been independently shown to predict what to build. What you can borrow is a vocabulary for describing opportunities and a few robust correlates of success. Whether a specific signal works, you have to test yourself.

The experiment that humbled me

A popular playbook: find abandoned or removed extensions that still have users, and rebuild them. People sell lists of these.

We took the 100 top-scoring candidates and had an AI agent research each one on the web:

0 were worth building. 34 were "maybe, with conditions", 66 were no.
90 already had a better, maintained alternative.
66 served a need that was shrinking or gone.
40% of the top 30 were "maybe", versus 28% of ranks 61–100 — only slightly better, and not one was a clear yes. The score meant much less than it looked.

Abandoned usually means abandoned for a reason. A list tells you what's broken; only validation tells you whether it's worth fixing.

Popular frameworks describe opportunities well — they don't predict them

Outcome-Driven Innovation gives you a great definition ("important but underserved outcomes"). Its famous 86% success rate comes from 21 of the vendor's own client projects.
Disruptive innovation: a re-examination of the 77 cases cited in the original work found only 9% fit all four elements of the theory.
"Timing matters most": the widely quoted 42% comes from a founder scoring 200 companies himself, after the outcomes were known.
Indie hacker wisdom (stair-step, audience-first, "4 of my 70+ projects made money") is genuinely useful — and mostly self-reported.

Two things do have real statistical backing:

A meta-analysis of 233 studies: the strongest correlate of new-product success is product advantage (a correlation of .34, the strongest in that meta-analysis). The samples are mostly manufacturing and B2B, so whether it transfers to small indie software is an open question.
A randomized trial with 116 startups: founders trained to treat ideas as falsifiable hypotheses performed better and were more willing to pivot. The real benefit is killing false positives.

And I could not find a single verified case of "picked an idea from store data → made money."

Base rates are brutal

Of subscription apps, 17.2% reach $1K/month within a year; 3.5% reach $10K.
The median Chrome extension has 17 users; about 70% have fewer than 100.
About 60% of Chrome extensions live roughly a year; among potentially infringing extensions, 86% are near-copies of removed ones.

So I assume random picks hit 5–20% of the time. Any method that reports a hit rate without comparing it to a random baseline is telling you nothing.

What we're moving to

Three layers, plus a separate "need" object (signals are live; dimensions and forms are being designed):

Signals — why is there a gap here? Seven on trial: Abandoned Users, Leader Complaints, Unfair Pricing, Rising Demand, Underserved Website, Leader Went Bad (acquired, ads added, price hike, ratings sliding), and Missing Here (people use it in another store; Chrome has nothing like it).
Dimensions — is it worth it? (being added) Size, willingness to pay, product advantage, build difficulty, acquisition difficulty, timing, evidence strength.
Form — what do we ship? (being designed) A single tool, a bundle of small needs, a template, or a funnel into something else.

Needs live on their own because of an early lesson: across 30,000 extensions, the unit of opportunity is a need, not an extension. One need often has dozens of extensions; old ones die and users have already moved.

Every "build this" will also ship with the riskiest assumption, the cheapest test, and a stop-loss line — our practical takeaway from that randomized trial.

Why backtesting

If no theory has been shown to predict, test against history. We have daily full snapshots of the Chrome Web Store going back to September 2023:

Pretend it's a past date and run today's rules to produce that day's opportunity list.
Check whether, within 12 months, a newcomer reached 10,000+ users in those needs.
Compare against randomly picked needs.

(Backtesting is borrowed from quant investing: judge today's rules by how they would have done on past data, not by how convincing they sound.)

That answers which signals beat chance and by how much, how to weight the dimensions, and something I couldn't find any outside study answering: do gaps mined from store data actually get filled by someone who wins?

Pitfalls to fix first: remove duplicate clones before counting competitors; clean the removals first — many are spam, so keep only ones that had real users and were compliant; and deduplicate "hits" too, or one winner's dozens of clones become dozens of hits.

One consequence to design for now

If many people subscribe to the same opportunity list, a shared list speeds up saturation of the very opportunities it finds. How opportunities are distributed (limited, batched delivery, routed by each user's own constraints) may need to be designed together with the backtest, not bolted on after launch.

If you're figuring out what to build

Look at markets as needs, not individual products.
Decide what separately from what form.
Ask "why is there a gap here?" first. No signal, no bet.
Rate product advantage before ease of building, or you'll drift toward easy-but-pointless.
Write the riskiest assumption, cheapest test and stop-loss before you build.
Always compare your hit rate to a random baseline.

The first backtest results are in the update at the top. When the forward checks are in, I'll publish them as they are — including the signals that turn out to be useless.

How do you decide what to build right now? Reply with what you're working on.

I'm Reid. Building is cheap now; knowing what to build isn't. Data picks, AI builds, every result goes public — misses included.

Sources: Wikipedia (Outcome-Driven Innovation); Strategyn; e-Literate, "Cracks in the Foundation of Disruptive Innovation"; Bill Gross talk transcript (Singju Post); Evanschitzky et al. 2012 meta-analysis; Camuffo et al., Management Science; TechCrunch 2024-03-12 on RevenueCat data; DebugBear, "Counting Chrome Extensions"; arXiv 2406.12710 and 2406.00374. Our data: 100-candidate research run and 30,009-extension need clustering, 2026-09-24. Data: Chrome-Stats, processed by us.

Drafted with AI assistance; the decisions and opinions are mine.

If you're weighing a direction right now, we do one free verdict a day. The link is in my profile.

We publish one Chrome extension opportunity a day. Here's today's, one we got wrong, and a free pilot

Reid Ji — Thu, 08 Oct 2026 12:30:08 +0000

I'm Reid. I'm running a public experiment: let AI decide what software I build, starting with Chrome extensions. We read public Chrome Web Store data, an AI researches each candidate on the web, and we publish the calls, misses included. Three of our own extensions went live in late September.

Today's opportunity (data from Oct 7)

A login-free SEO page checker that runs locally in the browser.

Signal: Leader Went Bad. SEOquake, the long-time default, had a recent breakdown backed by independent sources, and its redesign ties you to an account. Bad reviews start piling up from that point.
Gap: it doesn't flag X-Robots-Tag, the server header that quietly tells search engines not to index a page.
Why a local build helps: pure front-end, no account, so it's far less exposed to that kind of outage.
Risk: Detailed SEO Extension (4.91 stars) already caught many of those users. 104 products in this space, 9 of them paid.

Data: Chrome-Stats, processed by us.

One we got wrong

We group extensions into "demand clusters" (extensions that should do the same job). We spot-checked 50, and 20 of them were really mixing more than one job: "Text summary" held word counters; "download anime videos" was really general video downloaders. The model had sorted phrases seeing only cluster names. We now show it each cluster's top 12–25 extensions, split whatever can't replace each other, and have an independent reviewer recheck the result instead of letting the model grade itself.

Lesson: check members, not labels.

The free pilot

Our own developer account only has 3 store slots, and the method needs more real launches to be tested. So, both free while we test:

Verdict report. Bring a direction; we send a written verdict, worth building or not, and why. 1 a day.
Free build. Claim an opportunity from the list (or bring your own). We build it to store launch, it ships from your own developer account, and we maintain it for a month. 1 a week. If evaluation says it's more than a month of work, we decline up front.

In return: share numbers for 90 days after launch, and we write it up as a public case (anonymous is fine). In order of confirmation; after applying you can see your place in the queue.

The method isn't proven yet. That's why we publish the misses.

Apply: https://numberandreason.com/?src=en-dv-cm#join

What are you deciding whether to build right now?