Different kinds of coding and recoding

🌻 Different kinds of coding and recoding#


	Hard coding	Hard recoding	Links recoding	Factors recoding	Soft recoding
Accuracy	Highest	...	...	...	Lowest
Speed	Slowest	...	...	...	Fastest

Manual	Just code manually	Make a copy of your file, delete links and start again	Edit manually in Links table or Map, - or use search/replace in Links table	Edit manually in Factors table or Map, - or use search/replace in Factors table - or Bulk Edit	-
AI	Just code with AI, with/without a codebook	As above, or just put the switch "skip coded sources" to off	AI Answers / Links. Recode labels permanently or into temporary cause/effect columns.	AI Answers / Filters. Recode labels permanently or into temporary cause/effect columns	Apply magnetic labels in Soft Recode filter

What's the point of Links and Factors recoding? What's the difference?

Soft recoding is only as good as the underlying embedding space, and it is never perfect.
Hard recoding can take a long time, is expensive, and does not encourage experimentation
With Links/Factors recoding, you can:
- Recode just the currently filtered sources/links (or all links)
- Recode the permanent cause/effect labels or recode into one or more sets of temporary columns, e.g. experiment1_cause and experiment1_effect, and then use these temporary labels in your maps and tables.
- There is also another option Answers which is not about recoding; it is simply a way to send your links and/or factors data to an AI and getting a text answer.

But the main point is that rather than just hoping the magnetisation will work the way you want it to, you can do smart recoding as if you had an assistant to work through each label. For example you can say "Relabel everything which expresses a decrease or lack of something with a ~" or "Look at all these labels and tag each with [Food] or `[Health]"

You can even bring other columns into play, for example citation count, source count etc.
Links recoding is significantly more powerful because you can also include the actual Quote as well as both Cause and Effect. This means the AI can make its decision with a lot more context. So this is almost like recoding from scratch, but the original coding has already identified causal claims and all we have to do is relabel the labels with the same complete information about the claim.
Be careful: it's tempting to say things like "Find 3-8 top-level factor labels which cover the meaning of all these labels and recode them with the new top-level labels", but remember the "Rows per call" slider: with a large set of links and lots of quotes you will probably have to break up your work into multiple chunks, and each call may come up with different labels. In this case you could use the Answers mode (or the Cluster part of Soft Recode filter) first to develop some labels.

Recoding labels temporarily

Sometimes you want to improve your factor labels (cause/effect text) without changing the original data. You might want to:

Experiment safely — try different prompts or AI settings without overwriting what you coded
Iterate — run factor relabelling several times, refining the prompt each time, until you’re happy
Compare — switch between original and improved labels to see the difference
Review before committing — only merge into the main cause/effect fields when you’re satisfied

The app supports this with two features that work together: Temporary Cause/Effect Fields (a filter) and Target suffix (in AI Answers → Factors).

How it works

Create temporary columns. When you run factor relabelling, you can choose a “Target suffix” (e.g. _temp or _version1). Instead of overwriting cause and effect, the AI writes to cause_temp/effect_temp (or cause_version1/effect_version1). Your original labels stay untouched.
Show them on the map. Add the Temporary Cause/Effect Fields filter in the Filter Links tab. Point it at those same columns (e.g. cause_temp, effect_temp). The map will display the recoded labels instead of the originals.
Iterate. With the filter active, you can run factor relabelling again. The AI will work on the current temp labels (what you see on the map), not the originals. So you can refine prompts, fix odd results, and run again — all without touching the underlying data.
When you’re happy, you can
1. leave one or more sets of temporary columns as a separate view for analysis. You can switch between different sets and the permanent labels with the Temporary filter.
2. or rewrite the permanent cause/effect labels with these temporary labels if you want to make the changes permanent. The easiest way to do that is to apply just the Temporary filter and then Save As Currently Filtered.

Summary

Why: Experiment, iterate, and compare label improvements without changing your original coding.
How: Use a Target suffix when running factor relabelling, then add the Temporary Cause/Effect Fields filter to display those labels on the map. You can then run factor relabelling again to refine the temp labels further.

Open page →