Shipping with no telemetry, and what it costs

DeskDrawer makes no network connections at all. That is a stronger promise than 'privacy-friendly', and it has a real price the developer pays.

Privacy decision7 August 2026by Shipeng Ouyang

DeskDrawer contains no analytics, no crash reporting, no update check of its own, no account system and no licence phone-home. It opens no network connections. Not anonymized ones, not aggregated ones, not opt-out ones.

This page is about why, and about what it costs, because the second half is usually left out.

Why the strong version#

"Privacy-friendly" is a claim you have to trust. "Makes no network connections" is a claim you can check, with any firewall or network monitor, in about a minute.

That difference is the entire reason for the decision. A desktop organizer sees the name of every file on your desktop — which is a surprisingly complete picture of what someone is working on, who they work for, what they are applying for, what they are ill with. There is no anonymization scheme that makes that data harmless, and no privacy policy that makes "trust us" as good as "it cannot happen".

So the design rule is not collect less, it is have no mechanism. The application has no code that opens a socket. There is nothing to misconfigure, nothing to leak in a future release by accident, and nothing an attacker could repoint.

Everything else follows:

  • Updates are the Microsoft Store's job, so no update checker is needed.
  • Purchase and licensing are Windows' job, so no licence server is needed.
  • Diagnostics go to a local error.log you can read, which goes nowhere unless you send it.

What it costs#

I do not know when DeskDrawer breaks. Not how many people are affected, not on which Windows build, not whether the last release made things worse. A product with telemetry knows within hours. This one finds out when somebody writes an email.

I do not know which features are used. Whether anyone folds boards, whether sort-by-type earns its place, whether the icon-size submenu should have three options or six. Every roadmap decision is made without usage data. Eight rounds of code audits on a small app, and what each one found describes the substitute: read the code repeatedly, from a different angle each time. It works, but it is slower and it is not the same thing.

Crashes are invisible. The explorer.exe crash fixed in 1.2.4 was found by a code review. With crash reporting it would have surfaced the week it shipped.

Every bug report is worth more, and there are fewer of them. Most people who hit a problem close the application and move on. That is entirely reasonable and it means the developer learns nothing.

The bargain#

Users get a promise that is verifiable rather than a policy that is trusted. The developer gives up the entire feedback loop that modern software development assumes.

I still think it is the right trade for this kind of product — a small, local, single-purpose utility that touches something as personal as the contents of your desktop. But it is a trade, not a free virtue, and the cost is paid in bugs found late.

Which is why Support matters more here than for a typical product. If DeskDrawer misbehaves on your machine, that report is not one signal among thousands. It may be the only one.

The precise inventory of what is and is not stored is in Privacy & data.