The sandbox, for invited testers

The sandbox is a private copy of Cabdell for testing. It runs the same contract, the same indexer and the same web app as TestNet, on a small Algorand chain that runs on Cabdell’s own server in development mode. In the sandbox you can try in an afternoon what takes weeks and several people on TestNet:

publishing and new versions;
votes, reviews and comments;
reputation and the Peer-reviewed seal;
flags, disputes and juries.

It is by invitation only (DECISIONS D-153). Without an invitation the sandbox does not exist: it is not in the network menu, and its pages and addresses answer “not found”.

Getting in

The owner sends each tester a link of their own. Opening it lets that browser into the sandbox for 180 days and opens the Test tools page. The owner can revoke a link at any time, and the sandbox then closes to that person.

A purple SANDBOX band stays at the top of every page, so that nobody mistakes the sandbox for TestNet or MainNet. It also shows how far ahead the sandbox’s clock runs.

Test accounts

The sandbox needs no wallet app. In Test tools (or the wallet button), New test account makes an account in your browser and gives it 100 test ALGO from the sandbox’s dispenser. Make several to try co-authors, votes between people, or reviewing someone else’s article. Switch to… changes the account in use with one click.

The keys of these accounts are kept in your browser and exist only on the sandbox’s chain; they hold nothing of value. Rename changes an account’s name, which its test identity also shows (on profiles, in citations, on juries). Forget removes an account from the browser for good. After a visit to TestNet, the account you used last is connected again by itself.

The tools on each page

The Sandbox tools button in the bottom-right corner opens the tools that fit the page you are on:

Page	Tools
any page	the test account in use, a new account, the clock (+1 hour, +1 day, +7 days, +14 days)
an article	+5 votes, give the seal, dispute it with a jury (or with you on the jury); a helper comments, replies or mentions you; once disputed: how the jurors answer, a seat for you, and governance’s decision
a profile	+100 ALGO; for your own test accounts, a test ORCID iD and reputation in a field
a list or journal	helper addresses to invite as editors; helpers accept; a helper editor selects an article
other pages	helper accounts publish three articles; a helper founds a journal and invites you

The Test tools page has all of them in four groups (you and time, an article, other people, the sandbox), with an index at the top: the problems reported and the log of what testers did are under The sandbox. For admins it adds invitations and starting again. Escape closes the Sandbox tools panel and returns to its button.

Reporting a problem

When something does not work or looks wrong, open the Sandbox tools button and choose Report a problem. Say what you did, what you expected and what happened instead. The report carries what is needed to find the problem again, and Show what is sent lists it before you send:

the page and the test account in use;
the last transactions this page sent, and whether each was confirmed or why it failed;
the last errors the page met;
your browser and screen size;
the sandbox’s round and clock, and the tools you used in the last 15 minutes.

Nothing secret is sent: no keys, only transaction ids and messages. You see your reports, and whether they were resolved, on the Test tools page. The owner sees every report there, marks them resolved with a note, and can copy the open ones as text.

Helper accounts

Some tests need other people. For those, the sandbox has helper accounts that act like other people, always through the contract’s own methods:

they publish articles;
they vote for articles;
they vote reviews useful;
they write reviews;
they comment, reply and mention people;
they flag;
they answer as jurors;
they found journals and edit them.

Helpers have made-up names (“Helper Ana Ruiz”), so that profiles, citations and juries read like real ones.

Each tool uses fresh helpers where the rules require it. For example, seal reviewers are never an article’s authors, and jurors are never its flaggers.

Time

The sandbox’s clock only moves forward. Moving it ahead makes a block at the new time. Everything that depends on time follows it:

the daily limit of 50 reputation per field;
the jury’s 7 and 14 days;
the windows for co-authorship and returned reviews;
relative dates on the pages.

Times shown on pages are the sandbox’s time.

What each tool does

Reputation. Reputation is earned when others find your reviews useful. The tool does exactly that:

Your account reviews the field’s host article. The review is signed in your browser.
Helpers vote the review useful, until your account reaches the number you asked for. A useful vote from a helper gives 3 points.
When the daily limit of 50 holds the points back, the tool moves the clock a day ahead.

Reputation 10 lets you flag, and makes your reviews count toward the seal.

The Peer-reviewed seal. Two helpers write favourable reviews of at least 200 words. Each has a test ORCID iD and reputation 10 in the article’s field. Every author also needs a verified ORCID iD, so the tool first declares test iDs for the authors whose accounts are in your browser.

Disputes and juries. The tool works in this order:

Helpers qualify as possible jurors.
Other helpers flag the article until the flags weigh 30, and the article becomes disputed.
The indexer draws a jury of seven from the qualified helpers, as it would on TestNet.

Then you choose how the jurors answer:

all retract;
all clear;
split 4 to 3;
only three answer.

Governance resolves the dispute (in the sandbox the tools hold its key). To see the other cases of the jury’s rules, move the clock: +7 days replaces silent jurors, and +14 days reaches the deadline. After a jury without enough answers, More people for a second jury qualifies new helpers before the deadline; the second jury is drawn within a minute of it.

You on the jury. Dispute it, with me on the jury gives your account a test ORCID iD and reputation 10 in the article’s field, opens the dispute, and once the jury is drawn, helpers ahead of you in the draw step aside until you hold a seat. Then you answer from the dispute page, with the juror’s form. On a jury already drawn, Put me on the jury does the last step, if your account was qualified when it was drawn.

Comments from other people. A helper comments on an article, replies to the latest comment that is not a helper’s (yours), or mentions your account. Your inbox, replies, mentions and resolving comments can then be tried alone.

Journals. A helper founds a journal and invites your account as an editor; accept it from your inbox, which leads to your journals. For a journal of yours, invite helper addresses from its settings; helpers accept, and a helper editor selects articles.

DOIs. Any well-formed DOI counts as verified at once, and nobody outside is asked: Make up a test DOI in an article’s DOI form. Test DOIs start with 10.5072/ and are shown as “test DOI”, without a link: they name no real record (D-158).

How the sandbox differs from TestNet

	TestNet	Sandbox
Who can use it	everyone	invited testers
Chain	Algorand TestNet	a private chain on Cabdell’s server
Wallet	Pera, Defly, Lute	test accounts in the browser
ORCID iDs	checked against your ORCID record	any well-formed iD counts as verified at once; ORCID is never asked
DOIs	checked against Zenodo or DataCite	any well-formed DOI counts as verified at once
Files	IPFS, with a copy on Cabdell’s node	kept on Cabdell’s server only, never on IPFS
Clock	real time	can be moved ahead
Lasts	until MainNet	until an admin starts it again

Starting again. An admin can start the sandbox again. This deploys a new application on the same chain:

every article, review, vote, list and reputation is left behind;
accounts keep their ALGO;
the indexer starts again on its own;
a page left open from before says so, and signs nothing until it is reloaded.

Nothing in the sandbox carries over to TestNet or MainNet.